gnu/packages/machine-learning.scm

   1 ;;; GNU Guix --- Functional package management for GNU
   2 ;;; Copyright © 2015, 2016, 2017, 2018, 2019, 2020, 2021 Ricardo Wurmus <rekado@elephly.net>
   3 ;;; Copyright © 2016, 2020, 2021 Efraim Flashner <efraim@flashner.co.il>
   4 ;;; Copyright © 2016, 2017, 2020 Marius Bakke <mbakke@fastmail.com>
   5 ;;; Copyright © 2016 Hartmut Goebel <h.goebel@crazy-compilers.com>
   6 ;;; Copyright © 2018, 2019, 2020 Tobias Geerinckx-Rice <me@tobias.gr>
   7 ;;; Copyright © 2018 Kei Kebreau <kkebreau@posteo.net>
   8 ;;; Copyright © 2018 Mark Meyer <mark@ofosos.org>
   9 ;;; Copyright © 2018 Ben Woodcroft <donttrustben@gmail.com>
  10 ;;; Copyright © 2018 Fis Trivial <ybbs.daans@hotmail.com>
  11 ;;; Copyright © 2018 Julien Lepiller <julien@lepiller.eu>
  12 ;;; Copyright © 2018 Björn Höfling <bjoern.hoefling@bjoernhoefling.de>
  13 ;;; Copyright © 2019 Nicolas Goaziou <mail@nicolasgoaziou.fr>
  14 ;;; Copyright © 2019, 2020 Guillaume Le Vaillant <glv@posteo.net>
  15 ;;; Copyright © 2019 Brett Gilio <brettg@gnu.org>
  16 ;;; Copyright © 2020 Konrad Hinsen <konrad.hinsen@fastmail.net>
  17 ;;; Copyright © 2020 Edouard Klein <edk@beaver-labs.com>
  18 ;;; Copyright © 2020, 2021 Vinicius Monego <monego@posteo.net>
  19 ;;; Copyright © 2020 Maxim Cournoyer <maxim.cournoyer@gmail.com>
  20 ;;;
  21 ;;; This file is part of GNU Guix.
  22 ;;;
  23 ;;; GNU Guix is free software; you can redistribute it and/or modify it
  24 ;;; under the terms of the GNU General Public License as published by
  25 ;;; the Free Software Foundation; either version 3 of the License, or (at
  26 ;;; your option) any later version.
  27 ;;;
  28 ;;; GNU Guix is distributed in the hope that it will be useful, but
  29 ;;; WITHOUT ANY WARRANTY; without even the implied warranty of
  30 ;;; MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
  31 ;;; GNU General Public License for more details.
  32 ;;;
  33 ;;; You should have received a copy of the GNU General Public License
  34 ;;; along with GNU Guix.  If not, see <http://www.gnu.org/licenses/>.
  35
  36 (define-module (gnu packages machine-learning)
  37   #:use-module ((guix licenses) #:prefix license:)
  38   #:use-module (guix packages)
  39   #:use-module (guix utils)
  40   #:use-module (guix download)
  41   #:use-module (guix svn-download)
  42   #:use-module (guix build-system cmake)
  43   #:use-module (guix build-system gnu)
  44   #:use-module (guix build-system ocaml)
  45   #:use-module (guix build-system python)
  46   #:use-module (guix build-system r)
  47   #:use-module (guix git-download)
  48   #:use-module (gnu packages)
  49   #:use-module (gnu packages adns)
  50   #:use-module (gnu packages algebra)
  51   #:use-module (gnu packages audio)
  52   #:use-module (gnu packages autotools)
  53   #:use-module (gnu packages base)
  54   #:use-module (gnu packages bash)
  55   #:use-module (gnu packages boost)
  56   #:use-module (gnu packages check)
  57   #:use-module (gnu packages compression)
  58   #:use-module (gnu packages cmake)
  59   #:use-module (gnu packages cran)
  60   #:use-module (gnu packages databases)
  61   #:use-module (gnu packages dejagnu)
  62   #:use-module (gnu packages gcc)
  63   #:use-module (gnu packages glib)
  64   #:use-module (gnu packages graphviz)
  65   #:use-module (gnu packages gstreamer)
  66   #:use-module (gnu packages image)
  67   #:use-module (gnu packages linux)
  68   #:use-module (gnu packages llvm)
  69   #:use-module (gnu packages maths)
  70   #:use-module (gnu packages mpi)
  71   #:use-module (gnu packages ocaml)
  72   #:use-module (gnu packages onc-rpc)
  73   #:use-module (gnu packages perl)
  74   #:use-module (gnu packages pkg-config)
  75   #:use-module (gnu packages protobuf)
  76   #:use-module (gnu packages python)
  77   #:use-module (gnu packages python-build)
  78   #:use-module (gnu packages python-check)
  79   #:use-module (gnu packages python-science)
  80   #:use-module (gnu packages python-web)
  81   #:use-module (gnu packages python-xyz)
  82   #:use-module (gnu packages rpc)
  83   #:use-module (gnu packages serialization)
  84   #:use-module (gnu packages sphinx)
  85   #:use-module (gnu packages statistics)
  86   #:use-module (gnu packages sqlite)
  87   #:use-module (gnu packages swig)
  88   #:use-module (gnu packages web)
  89   #:use-module (gnu packages xml)
  90   #:use-module (gnu packages xorg)
  91   #:use-module (ice-9 match))
  92
  93 (define-public fann
  94   ;; The last release is >100 commits behind, so we package from git.
  95   (let ((commit "d71d54788bee56ba4cf7522801270152da5209d7"))
  96     (package
  97       (name "fann")
  98       (version (string-append "2.2.0-1." (string-take commit 8)))
  99       (source (origin
 100                 (method git-fetch)
 101                 (uri (git-reference
 102                       (url "https://github.com/libfann/fann")
 103                       (commit commit)))
 104                 (file-name (string-append name "-" version "-checkout"))
 105                 (sha256
 106                  (base32
 107                   "0ibwpfrjs6q2lijs8slxjgzb2llcl6rk3v2ski4r6215g5jjhg3x"))))
 108       (build-system cmake-build-system)
 109       (arguments
 110        `(#:phases
 111          (modify-phases %standard-phases
 112            (replace 'check
 113              (lambda* (#:key outputs #:allow-other-keys)
 114                (let* ((out (assoc-ref outputs "out")))
 115                  (with-directory-excursion (string-append (getcwd) "/tests")
 116                    (invoke "./fann_tests"))))))))
 117       (home-page "http://leenissen.dk/fann/wp/")
 118       (synopsis "Fast Artificial Neural Network")
 119       (description
 120        "FANN is a neural network library, which implements multilayer
 121 artificial neural networks in C with support for both fully connected and
 122 sparsely connected networks.")
 123       (license license:lgpl2.1))))
 124
 125 (define-public libsvm
 126   (package
 127     (name "libsvm")
 128     (version "3.23")
 129     (source
 130      (origin
 131        (method url-fetch)
 132        (uri (string-append "https://www.csie.ntu.edu.tw/~cjlin/libsvm/"
 133                            name "-" version ".tar.gz"))
 134        (sha256
 135         (base32 "0jpaq0rr92x38p4nk3gjan79ip67m6p80anb28z1d8601miysyi5"))))
 136     (build-system gnu-build-system)
 137     (arguments
 138      `(#:tests? #f                      ; no "check" target
 139        #:phases (modify-phases %standard-phases
 140                   (delete 'configure)
 141                   (replace
 142                    'install             ; no ‘install’ target
 143                    (lambda* (#:key outputs #:allow-other-keys)
 144                      (let* ((out (assoc-ref outputs "out"))
 145                             (bin (string-append out "/bin/")))
 146                        (mkdir-p bin)
 147                        (for-each (lambda (file)
 148                                    (copy-file file (string-append bin file)))
 149                                  '("svm-train"
 150                                    "svm-predict"
 151                                    "svm-scale")))
 152                      #t)))))
 153     (home-page "https://www.csie.ntu.edu.tw/~cjlin/libsvm/")
 154     (synopsis "Library for Support Vector Machines")
 155     (description
 156      "LIBSVM is a machine learning library for support vector
 157 classification, (C-SVC, nu-SVC), regression (epsilon-SVR, nu-SVR) and
 158 distribution estimation (one-class SVM).  It supports multi-class
 159 classification.")
 160     (license license:bsd-3)))
 161
 162 (define-public python-libsvm
 163   (package (inherit libsvm)
 164     (name "python-libsvm")
 165     (build-system gnu-build-system)
 166     (arguments
 167      `(#:tests? #f                      ; no "check" target
 168        #:make-flags '("-C" "python")
 169        #:phases
 170        (modify-phases %standard-phases
 171          (delete 'configure)
 172          (replace
 173           'install                      ; no ‘install’ target
 174           (lambda* (#:key inputs outputs #:allow-other-keys)
 175             (let ((site (string-append (assoc-ref outputs "out")
 176                                        "/lib/python"
 177                                        (string-take
 178                                         (string-take-right
 179                                          (assoc-ref inputs "python") 5) 3)
 180                                        "/site-packages/")))
 181               (substitute* "python/svm.py"
 182                 (("../libsvm.so.2") "libsvm.so.2"))
 183               (mkdir-p site)
 184               (for-each (lambda (file)
 185                           (copy-file file (string-append site (basename file))))
 186                         (find-files "python" "\\.py"))
 187               (copy-file "libsvm.so.2"
 188                          (string-append site "libsvm.so.2")))
 189             #t)))))
 190     (inputs
 191      `(("python" ,python)))
 192     (synopsis "Python bindings of libSVM")))
 193
 194 (define-public ghmm
 195   ;; The latest release candidate is several years and a couple of fixes have
 196   ;; been published since.  This is why we download the sources from the SVN
 197   ;; repository.
 198   (let ((svn-revision 2341))
 199     (package
 200       (name "ghmm")
 201       (version (string-append "0.9-rc3-0." (number->string svn-revision)))
 202       (source (origin
 203                 (method svn-fetch)
 204                 (uri (svn-reference
 205                       (url "http://svn.code.sf.net/p/ghmm/code/trunk")
 206                       (revision svn-revision)))
 207                 (file-name (string-append name "-" version "-checkout"))
 208                 (sha256
 209                  (base32
 210                   "0qbq1rqp94l530f043qzp8aw5lj7dng9wq0miffd7spd1ff638wq"))))
 211       (build-system gnu-build-system)
 212       (arguments
 213        `(#:imported-modules (,@%gnu-build-system-modules
 214                              (guix build python-build-system))
 215          #:modules          ((guix build python-build-system)
 216                              ,@%gnu-build-system-modules)
 217          #:phases
 218          (modify-phases %standard-phases
 219            (add-after 'unpack 'enter-dir
 220              (lambda _ (chdir "ghmm") #t))
 221            (delete 'check)
 222            (add-after 'install 'check
 223              (assoc-ref %standard-phases 'check))
 224            (add-before 'check 'fix-PYTHONPATH
 225              (lambda* (#:key inputs outputs #:allow-other-keys)
 226                (let ((python-version (python-version
 227                                       (assoc-ref inputs "python"))))
 228                  (setenv "PYTHONPATH"
 229                          (string-append (getenv "PYTHONPATH")
 230                                         ":" (assoc-ref outputs "out")
 231                                         "/lib/python" python-version
 232                                         "/site-packages")))
 233                #t))
 234            (add-after 'enter-dir 'fix-runpath
 235              (lambda* (#:key outputs #:allow-other-keys)
 236                (substitute* "ghmmwrapper/setup.py"
 237                  (("^(.*)extra_compile_args = \\[" line indent)
 238                   (string-append indent
 239                                  "extra_link_args = [\"-Wl,-rpath="
 240                                  (assoc-ref outputs "out") "/lib\"],\n"
 241                                  line
 242                                  "\"-Wl,-rpath="
 243                                  (assoc-ref outputs "out")
 244                                  "/lib\", ")))
 245                #t))
 246            (add-after 'enter-dir 'disable-broken-tests
 247              (lambda _
 248                (substitute* "tests/Makefile.am"
 249                  ;; GHMM_SILENT_TESTS is assumed to be a command.
 250                  (("TESTS_ENVIRONMENT.*") "")
 251                  ;; Do not build broken tests.
 252                  (("chmm .*") "")
 253                  (("read_fa .*") "")
 254                  (("mcmc .*") "")
 255                  (("label_higher_order_test.*$")
 256                   "label_higher_order_test\n"))
 257
 258                ;; These Python unittests are broken as there is no gato.
 259                ;; See https://sourceforge.net/p/ghmm/support-requests/3/
 260                (substitute* "ghmmwrapper/ghmmunittests.py"
 261                  (("^(.*)def (testNewXML|testMultipleTransitionClasses|testNewXML)"
 262                    line indent)
 263                   (string-append indent
 264                                  "@unittest.skip(\"Disabled by Guix\")\n"
 265                                  line)))
 266                #t)))))
 267       (inputs
 268        `(("python" ,python-2) ; only Python 2 is supported
 269          ("libxml2" ,libxml2)))
 270       (native-inputs
 271        `(("pkg-config" ,pkg-config)
 272          ("dejagnu" ,dejagnu)
 273          ("swig" ,swig)
 274          ("autoconf" ,autoconf)
 275          ("automake" ,automake)
 276          ("libtool" ,libtool)))
 277       (home-page "http://ghmm.org")
 278       (synopsis "Hidden Markov Model library")
 279       (description
 280        "The General Hidden Markov Model library (GHMM) is a C library with
 281 additional Python bindings implementing a wide range of types of @dfn{Hidden
 282 Markov Models} (HMM) and algorithms: discrete, continuous emissions, basic
 283 training, HMM clustering, HMM mixtures.")
 284       (license license:lgpl2.0+))))
 285
 286 (define-public mcl
 287   (package
 288     (name "mcl")
 289     (version "14.137")
 290     (source (origin
 291               (method url-fetch)
 292               (uri (string-append
 293                     "http://micans.org/mcl/src/mcl-"
 294                     (string-replace-substring version "." "-")
 295                     ".tar.gz"))
 296               (sha256
 297                (base32
 298                 "15xlax3z31lsn62vlg94hkm75nm40q4679amnfg13jm8m2bnhy5m"))))
 299     (build-system gnu-build-system)
 300     (arguments
 301      `(#:configure-flags (list "--enable-blast")))
 302     (inputs
 303      `(("perl" ,perl)))
 304     (home-page "http://micans.org/mcl/")
 305     (synopsis "Clustering algorithm for graphs")
 306     (description
 307      "The MCL algorithm is short for the @dfn{Markov Cluster Algorithm}, a
 308 fast and scalable unsupervised cluster algorithm for graphs (also known as
 309 networks) based on simulation of (stochastic) flow in graphs.")
 310     ;; In the LICENCE file and web page it says "The software is licensed
 311     ;; under the GNU General Public License, version 3.", but in several of
 312     ;; the source code files it suggests GPL3 or later.
 313     ;; http://listserver.ebi.ac.uk/pipermail/mcl-users/2016/000376.html
 314     (license license:gpl3)))
 315
 316 (define-public ocaml-mcl
 317   (package
 318     (name "ocaml-mcl")
 319     (version "12-068oasis4")
 320     (source
 321      (origin
 322        (method git-fetch)
 323        (uri (git-reference
 324              (url "https://github.com/fhcrc/mcl")
 325              (commit version)))
 326        (file-name (git-file-name name version))
 327        (sha256
 328         (base32
 329          "0009dc3h2jp3qg5val452wngpqnbfyhbcxylghq0mrjqxx0jdq5p"))))
 330     (build-system ocaml-build-system)
 331     (arguments
 332      `(#:phases
 333        (modify-phases %standard-phases
 334          (add-before 'configure 'patch-paths
 335            (lambda _
 336              (substitute* "setup.ml"
 337                (("LDFLAGS=-fPIC")
 338                 (string-append "LDFLAGS=-fPIC\"; \"SHELL=" (which "sh")))
 339                (("-std=c89") "-std=gnu99")
 340
 341                ;; This is a mutable string, which is no longer supported.  Use
 342                ;; a byte buffer instead.
 343                (("String.make \\(String.length s\\)")
 344                 "Bytes.make (String.length s)")
 345
 346                ;; These two belong together.
 347                (("OASISString.replace_chars")
 348                 "Bytes.to_string (OASISString.replace_chars")
 349                ((" s;")
 350                 " s);"))
 351              (substitute* "myocamlbuild.ml"
 352                (("std=c89") "std=gnu99"))
 353              ;; Since we build with a more recent OCaml, we have to use C99 or
 354              ;; later.  This causes problems with the old C code.
 355              (substitute* "src/impala/matrix.c"
 356                (("restrict") "restrict_"))
 357              #t)))))
 358     (native-inputs
 359      `(("ocamlbuild" ,ocamlbuild)))
 360     (home-page "https://github.com/fhcrc/mcl")
 361     (synopsis "OCaml wrappers around MCL")
 362     (description
 363      "This package provides OCaml bindings for the MCL graph clustering
 364 algorithm.")
 365     (license license:gpl3)))
 366
 367 (define-public randomjungle
 368   (package
 369     (name "randomjungle")
 370     (version "2.1.0")
 371     (source
 372      (origin
 373        (method url-fetch)
 374        (uri (string-append
 375              "https://www.imbs.uni-luebeck.de/fileadmin/files/Software"
 376              "/randomjungle/randomjungle-" version ".tar_.gz"))
 377        (patches (search-patches "randomjungle-disable-static-build.patch"))
 378        (sha256
 379         (base32
 380          "12c8rf30cla71swx2mf4ww9mfd8jbdw5lnxd7dxhyw1ygrvg6y4w"))))
 381     (build-system gnu-build-system)
 382     (arguments
 383      `(#:configure-flags
 384        (list "--disable-static"
 385              (string-append "--with-boost="
 386                             (assoc-ref %build-inputs "boost")))
 387        #:phases
 388        (modify-phases %standard-phases
 389          (add-before
 390           'configure 'set-CXXFLAGS
 391           (lambda _
 392             (setenv "CXXFLAGS" "-fpermissive ")
 393             #t)))))
 394     (inputs
 395      `(("boost" ,boost)
 396        ("gsl" ,gsl)
 397        ("libxml2" ,libxml2)
 398        ("zlib" ,zlib)))
 399     (native-inputs
 400      `(("gfortran" ,gfortran)
 401        ("gfortran:lib" ,gfortran "lib")))
 402     ;; Non-portable assembly instructions are used so building fails on
 403     ;; platforms other than x86_64 or i686.
 404     (supported-systems '("x86_64-linux" "i686-linux"))
 405     (home-page "https://www.imbs.uni-luebeck.de/forschung/software/details.html#c224")
 406     (synopsis "Implementation of the Random Forests machine learning method")
 407     (description
 408      "Random Jungle is an implementation of Random Forests.  It is supposed to
 409 analyse high dimensional data.  In genetics, it can be used for analysing big
 410 Genome Wide Association (GWA) data.  Random Forests is a powerful machine
 411 learning method.  Most interesting features are variable selection, missing
 412 value imputation, classifier creation, generalization error estimation and
 413 sample proximities between pairs of cases.")
 414     (license license:gpl3+)))
 415
 416 (define-public openfst
 417   (package
 418     (name "openfst")
 419     (version "1.7.9")
 420     (source (origin
 421               (method url-fetch)
 422               (uri (string-append "http://www.openfst.org/twiki/pub/FST/"
 423                                   "FstDownload/openfst-" version ".tar.gz"))
 424               (sha256
 425                (base32
 426                 "1pmx1yhn2gknj0an0zwqmzgwjaycapi896244np50a8y3nrsw6ck"))))
 427     (build-system gnu-build-system)
 428     (home-page "http://www.openfst.org")
 429     (synopsis "Library for weighted finite-state transducers")
 430     (description "OpenFst is a library for constructing, combining,
 431 optimizing, and searching weighted finite-state transducers (FSTs).")
 432     (license license:asl2.0)))
 433
 434 (define-public shogun
 435   (package
 436     (name "shogun")
 437     (version "6.1.3")
 438     (source
 439      (origin
 440        (method url-fetch)
 441        (uri (string-append
 442              "ftp://shogun-toolbox.org/shogun/releases/"
 443              (version-major+minor version)
 444              "/sources/shogun-" version ".tar.bz2"))
 445        (sha256
 446         (base32
 447          "1rn9skm3nw6hr7mr3lgp2gfqhi7ii0lyxck7qmqnf8avq349s5jp"))
 448        (modules '((guix build utils)
 449                   (ice-9 rdelim)))
 450        (snippet
 451         '(begin
 452            ;; Remove non-free sources and files referencing them
 453            (for-each delete-file
 454                      (find-files "src/shogun/classifier/svm/"
 455                                  "SVMLight\\.(cpp|h)"))
 456            (for-each delete-file
 457                      (find-files "examples/undocumented/libshogun/"
 458                                  (string-append
 459                                   "(classifier_.*svmlight.*|"
 460                                   "evaluation_cross_validation_locked_comparison).cpp")))
 461            ;; Remove non-free functions.
 462            (define (delete-ifdefs file)
 463              (with-atomic-file-replacement file
 464                (lambda (in out)
 465                  (let loop ((line (read-line in 'concat))
 466                             (skipping? #f))
 467                    (if (eof-object? line)
 468                        #t
 469                        (let ((skip-next?
 470                               (or (and skipping?
 471                                        (not (string-prefix?
 472                                              "#endif //USE_SVMLIGHT" line)))
 473                                   (string-prefix?
 474                                    "#ifdef USE_SVMLIGHT" line))))
 475                          (when (or (not skipping?)
 476                                    (and skipping? (not skip-next?)))
 477                            (display line out))
 478                          (loop (read-line in 'concat) skip-next?)))))))
 479            (for-each delete-ifdefs
 480                      (append
 481                       (find-files "src/shogun/classifier/mkl"
 482                                   "^MKLClassification\\.cpp")
 483                       (find-files "src/shogun/classifier/svm"
 484                                   "^SVMLightOneClass\\.(cpp|h)")
 485                       (find-files "src/shogun/multiclass"
 486                                   "^ScatterSVM\\.(cpp|h)")
 487                       (find-files "src/shogun/kernel/"
 488                                   "^(Kernel|CombinedKernel|ProductKernel)\\.(cpp|h)")
 489                       (find-files "src/shogun/regression/svr"
 490                                   "^(MKLRegression|SVRLight)\\.(cpp|h)")
 491                       (find-files "src/shogun/transfer/domain_adaptation"
 492                                   "^DomainAdaptationSVM\\.(cpp|h)")))
 493            #t))))
 494     (build-system cmake-build-system)
 495     (arguments
 496      '(#:tests? #f ;no check target
 497        #:phases
 498        (modify-phases %standard-phases
 499          (add-after 'unpack 'delete-broken-symlinks
 500            (lambda _
 501              (for-each delete-file '("applications/arts/data"
 502                                      "applications/asp/data"
 503                                      "applications/easysvm/data"
 504                                      "applications/msplicer/data"
 505                                      "applications/ocr/data"
 506                                      "examples/meta/data"
 507                                      "examples/undocumented/data"))
 508              #t))
 509          (add-after 'unpack 'change-R-target-path
 510            (lambda* (#:key outputs #:allow-other-keys)
 511              (substitute* '("src/interfaces/r/CMakeLists.txt"
 512                             "examples/meta/r/CMakeLists.txt")
 513                (("\\$\\{R_COMPONENT_LIB_PATH\\}")
 514                 (string-append (assoc-ref outputs "out")
 515                                "/lib/R/library/")))
 516              #t))
 517          (add-after 'unpack 'fix-octave-modules
 518            (lambda* (#:key outputs #:allow-other-keys)
 519              (substitute* "src/interfaces/octave/CMakeLists.txt"
 520                (("^include_directories\\(\\$\\{OCTAVE_INCLUDE_DIRS\\}")
 521                 "include_directories(${OCTAVE_INCLUDE_DIRS} ${OCTAVE_INCLUDE_DIRS}/octave")
 522                ;; change target directory
 523                (("\\$\\{OCTAVE_OCT_LOCAL_API_FILE_DIR\\}")
 524                 (string-append (assoc-ref outputs "out")
 525                                "/share/octave/packages")))
 526              (substitute* '("src/interfaces/octave/swig_typemaps.i"
 527                             "src/interfaces/octave/sg_print_functions.cpp")
 528                ;; "octave/config.h" and "octave/oct-obj.h" deprecated in Octave.
 529                (("octave/config\\.h") "octave/octave-config.h")
 530                (("octave/oct-obj.h") "octave/ovl.h"))
 531              #t))
 532          (add-after 'unpack 'move-rxcpp
 533            (lambda* (#:key inputs #:allow-other-keys)
 534              (let ((rxcpp-dir "shogun/third-party/rxcpp"))
 535                (mkdir-p rxcpp-dir)
 536                (install-file (assoc-ref inputs "rxcpp") rxcpp-dir)
 537                #t)))
 538          (add-before 'build 'set-HOME
 539            ;; $HOME needs to be set at some point during the build phase
 540            (lambda _ (setenv "HOME" "/tmp") #t)))
 541        #:configure-flags
 542        (list "-DCMAKE_BUILD_WITH_INSTALL_RPATH=TRUE"
 543              "-DUSE_SVMLIGHT=OFF" ;disable proprietary SVMLIGHT
 544              "-DBUILD_META_EXAMPLES=OFF" ;requires unpackaged ctags
 545              ;;"-DINTERFACE_JAVA=ON" ;requires unpackaged jblas
 546              ;;"-DINTERFACE_RUBY=ON" ;requires unpackaged ruby-narray
 547              ;;"-DINTERFACE_PERL=ON" ;"FindPerlLibs" does not exist
 548              ;;"-DINTERFACE_LUA=ON"  ;fails because lua doesn't build pkgconfig file
 549              "-DINTERFACE_OCTAVE=ON"
 550              "-DINTERFACE_PYTHON=ON"
 551              "-DINTERFACE_R=ON")))
 552     (inputs
 553      `(("python" ,python)
 554        ("numpy" ,python-numpy)
 555        ("r-minimal" ,r-minimal)
 556        ("octave" ,octave-cli)
 557        ("swig" ,swig)
 558        ("eigen" ,eigen)
 559        ("hdf5" ,hdf5)
 560        ("atlas" ,atlas)
 561        ("arpack" ,arpack-ng)
 562        ("lapack" ,lapack)
 563        ("glpk" ,glpk)
 564        ("libxml2" ,libxml2)
 565        ("lzo" ,lzo)
 566        ("zlib" ,zlib)))
 567     (native-inputs
 568      `(("pkg-config" ,pkg-config)
 569        ("rxcpp" ,rxcpp)))
 570     ;; Non-portable SSE instructions are used so building fails on platforms
 571     ;; other than x86_64.
 572     (supported-systems '("x86_64-linux"))
 573     (home-page "https://shogun-toolbox.org/")
 574     (synopsis "Machine learning toolbox")
 575     (description
 576      "The Shogun Machine learning toolbox provides a wide range of unified and
 577 efficient Machine Learning (ML) methods.  The toolbox seamlessly
 578 combines multiple data representations, algorithm classes, and general purpose
 579 tools.  This enables both rapid prototyping of data pipelines and extensibility
 580 in terms of new algorithms.")
 581     (license license:gpl3+)))
 582
 583 (define-public python-onnx
 584   (package
 585     (name "python-onnx")
 586     (version "1.8.1")
 587     (source
 588      (origin
 589        (method url-fetch)
 590        (uri (pypi-uri "onnx" version))
 591        ;; ONNX will build googletest from a git checkout.  Patch CMake
 592        ;; to use googletest from Guix and enable tests by default.
 593        (patches (search-patches "python-onnx-use-system-googletest.patch"))
 594        (sha256
 595         (base32 "1ys5f4kqkabm4mgivsw80zz8xj1svanfbpszqbw9j15914hcarcx"))))
 596     (build-system python-build-system)
 597     (native-inputs
 598      `(("cmake" ,cmake)
 599        ("googletest" ,googletest)
 600        ("pybind11" ,pybind11)
 601        ("python-coverage" ,python-coverage)
 602        ("python-nbval" ,python-nbval)
 603        ("python-pytest" ,python-pytest)
 604        ("python-pytest-runner" ,python-pytest-runner)))
 605     (inputs
 606      `(("protobuf" ,protobuf)))
 607     (propagated-inputs
 608      `(("python-numpy" ,python-numpy)
 609        ("python-protobuf" ,python-protobuf)
 610        ("python-six" ,python-six)
 611        ("python-tabulate" ,python-tabulate)
 612        ("python-typing-extensions"
 613         ,python-typing-extensions)))
 614     (home-page "https://onnx.ai/")
 615     (synopsis "Open Neural Network Exchange")
 616     (description
 617      "Open Neural Network Exchange (ONNX) provides an open source format for
 618 AI models, both deep learning and traditional ML.  It defines an extensible
 619 computation graph model, as well as definitions of built-in operators and
 620 standard data types.")
 621     (license license:expat)))
 622
 623 (define-public rxcpp
 624   (package
 625     (name "rxcpp")
 626     (version "4.1.0")
 627     (source
 628      (origin
 629        (method git-fetch)
 630        (uri (git-reference
 631              (url "https://github.com/ReactiveX/RxCpp")
 632              (commit (string-append "v" version))))
 633        (sha256
 634         (base32 "1rdpa3jlc181jd08nk437aar085h28i45s6nzrv65apb3xyyz0ij"))
 635        (file-name (git-file-name name version))))
 636     (build-system cmake-build-system)
 637     (arguments
 638      `(#:phases
 639        (modify-phases %standard-phases
 640          (add-after 'unpack 'remove-werror
 641            (lambda _
 642              (substitute* (find-files ".")
 643                (("-Werror") ""))
 644              #t))
 645          (replace 'check
 646            (lambda _
 647              (invoke "ctest"))))))
 648     (native-inputs
 649      `(("catch" ,catch-framework)))
 650     (home-page "http://reactivex.io/")
 651     (synopsis "Reactive Extensions for C++")
 652     (description
 653      "The Reactive Extensions for C++ (RxCpp) is a library of algorithms for
 654 values-distributed-in-time.  ReactiveX is a library for composing asynchronous
 655 and event-based programs by using observable sequences.
 656
 657 It extends the observer pattern to support sequences of data and/or events and
 658 adds operators that allow you to compose sequences together declaratively while
 659 abstracting away concerns about things like low-level threading,
 660 synchronization, thread-safety, concurrent data structures, and non-blocking
 661 I/O.")
 662     (license license:asl2.0)))
 663
 664 (define-public gemmlowp-for-tensorflow
 665   ;; The commit hash is taken from "tensorflow/workspace.bzl".
 666   (let ((commit "38ebac7b059e84692f53e5938f97a9943c120d98")
 667         (revision "2"))
 668     (package
 669       (name "gemmlowp")
 670       (version (git-version "0" revision commit))
 671       (source (origin
 672                 (method url-fetch)
 673                 (uri (string-append "https://mirror.bazel.build/"
 674                                     "github.com/google/gemmlowp/archive/"
 675                                     commit ".zip"))
 676                 (file-name (string-append "gemmlowp-" version ".zip"))
 677                 (sha256
 678                  (base32
 679                   "0n56s2g8hrssm4w8qj1v58gfm56a04n9v992ixkmvk6zjiralzxq"))))
 680       (build-system cmake-build-system)
 681       (arguments
 682        `(#:configure-flags
 683          (list ,@(match (%current-system)
 684                    ((or "x86_64-linux" "i686-linux")
 685                     '("-DCMAKE_CXX_FLAGS=-msse2"))
 686                    (_ '())))
 687          #:phases
 688          (modify-phases %standard-phases
 689            ;; This directory contains the CMakeLists.txt.
 690            (add-after 'unpack 'chdir
 691              (lambda _ (chdir "contrib") #t))
 692            ;; There is no install target
 693            (replace 'install
 694              (lambda* (#:key outputs #:allow-other-keys)
 695                (let* ((out (assoc-ref outputs "out"))
 696                       (lib (string-append out "/lib/"))
 697                       (inc (string-append out "/include/")))
 698                  (install-file "../build/libeight_bit_int_gemm.so" lib)
 699                  (for-each (lambda (dir)
 700                              (let ((target (string-append inc "/" dir)))
 701                                (mkdir-p target)
 702                                (for-each (lambda (h)
 703                                            (install-file h target))
 704                                          (find-files (string-append "../" dir)
 705                                                      "\\.h$"))))
 706                            '("meta" "profiling" "public" "fixedpoint"
 707                              "eight_bit_int_gemm" "internal"))
 708                  #t))))))
 709       (native-inputs
 710        `(("unzip" ,unzip)))
 711       (home-page "https://github.com/google/gemmlowp")
 712       (synopsis "Small self-contained low-precision GEMM library")
 713       (description
 714        "This is a small self-contained low-precision @dfn{general matrix
 715 multiplication} (GEMM) library.  It is not a full linear algebra library.
 716 Low-precision means that the input and output matrix entries are integers on
 717 at most 8 bits.  To avoid overflow, results are internally accumulated on more
 718 than 8 bits, and at the end only some significant 8 bits are kept.")
 719       (license license:asl2.0))))
 720
 721 (define-public dlib
 722   (package
 723     (name "dlib")
 724     (version "19.20")
 725     (source (origin
 726               (method url-fetch)
 727               (uri (string-append
 728                     "http://dlib.net/files/dlib-" version ".tar.bz2"))
 729               (sha256
 730                (base32
 731                 "139jyi19qz37wwmmy48gil9d1kkh2r3w3bwdzabha6ayxmba96nz"))
 732               (modules '((guix build utils)))
 733               (snippet
 734                '(begin
 735                   ;; Delete ~13MB of bundled dependencies.
 736                   (delete-file-recursively "dlib/external")
 737                   (delete-file-recursively "docs/dlib/external")
 738                   #t))))
 739     (build-system cmake-build-system)
 740     (arguments
 741      `(#:configure-flags '("-DBUILD_SHARED_LIBS=ON")
 742        #:phases
 743        (modify-phases %standard-phases
 744          (add-after 'unpack 'disable-asserts
 745            (lambda _
 746              ;; config.h recommends explicitly enabling or disabling asserts
 747              ;; when building as a shared library. By default neither is set.
 748              (substitute* "dlib/config.h"
 749                (("^//#define DLIB_DISABLE_ASSERTS") "#define DLIB_DISABLE_ASSERTS"))
 750              #t))
 751          (add-after 'disable-asserts 'disable-failing-tests
 752            (lambda _
 753              ;; One test times out on MIPS, so we need to disable it.
 754              ;; Others are flaky on some platforms.
 755              (let* ((system ,(or (%current-target-system)
 756                                  (%current-system)))
 757                     (disabled-tests (cond
 758                                      ((string-prefix? "mips64" system)
 759                                       '("object_detector" ; timeout
 760                                         "data_io"))
 761                                      ((string-prefix? "armhf" system)
 762                                       '("learning_to_track"))
 763                                      ((string-prefix? "i686" system)
 764                                       '("optimization"))
 765                                      (else '()))))
 766                (for-each
 767                 (lambda (test)
 768                   (substitute* "dlib/test/makefile"
 769                     (((string-append "SRC \\+= " test "\\.cpp")) "")))
 770                 disabled-tests)
 771                #t)))
 772          (replace 'check
 773            (lambda _
 774              ;; No test target, so we build and run the unit tests here.
 775              (let ((test-dir (string-append "../dlib-" ,version "/dlib/test")))
 776                (with-directory-excursion test-dir
 777                  (invoke "make" "-j" (number->string (parallel-job-count)))
 778                  (invoke "./dtest" "--runall"))
 779                #t))))))
 780     (native-inputs
 781      `(("pkg-config" ,pkg-config)
 782        ;; For tests.
 783        ("libnsl" ,libnsl)))
 784     (inputs
 785      `(("giflib" ,giflib)
 786        ("lapack" ,lapack)
 787        ("libjpeg" ,libjpeg-turbo)
 788        ("libpng" ,libpng)
 789        ("libx11" ,libx11)
 790        ("openblas" ,openblas)
 791        ("zlib" ,zlib)))
 792     (synopsis
 793      "Toolkit for making machine learning and data analysis applications in C++")
 794     (description
 795      "Dlib is a modern C++ toolkit containing machine learning algorithms and
 796 tools.  It is used in both industry and academia in a wide range of domains
 797 including robotics, embedded devices, mobile phones, and large high performance
 798 computing environments.")
 799     (home-page "http://dlib.net")
 800     (license license:boost1.0)))
 801
 802 (define-public python-scikit-learn
 803   (package
 804     (name "python-scikit-learn")
 805     (version "0.24.1")
 806     (source
 807      (origin
 808        (method git-fetch)
 809        (uri (git-reference
 810              (url "https://github.com/scikit-learn/scikit-learn")
 811              (commit version)))
 812        (file-name (git-file-name name version))
 813        (sha256
 814         (base32
 815          "0dd854hi9h81pa3y6gwa6r4qjwrwq5fndi312h6dkqzfh7jbvgvd"))))
 816     (build-system python-build-system)
 817     (arguments
 818      `(#:phases
 819        (modify-phases %standard-phases
 820          (add-after 'build 'build-ext
 821            (lambda _ (invoke "python" "setup.py" "build_ext" "--inplace") #t))
 822          (replace 'check
 823            (lambda _
 824              ;; Restrict OpenBLAS threads to prevent segfaults while testing!
 825              (setenv "OPENBLAS_NUM_THREADS" "1")
 826
 827              ;; Some tests require write access to $HOME.
 828              (setenv "HOME" "/tmp")
 829
 830              (invoke "pytest" "sklearn" "-m" "not network")))
 831          (add-before 'reset-gzip-timestamps 'make-files-writable
 832            (lambda* (#:key outputs #:allow-other-keys)
 833              ;; Make sure .gz files are writable so that the
 834              ;; 'reset-gzip-timestamps' phase can do its work.
 835              (let ((out (assoc-ref outputs "out")))
 836                (for-each make-file-writable
 837                          (find-files out "\\.gz$"))
 838                #t))))))
 839     (inputs
 840      `(("openblas" ,openblas)))
 841     (native-inputs
 842      `(("python-pytest" ,python-pytest)
 843        ("python-pandas" ,python-pandas) ;for tests
 844        ("python-cython" ,python-cython)))
 845     (propagated-inputs
 846      `(("python-numpy" ,python-numpy)
 847        ("python-threadpoolctl" ,python-threadpoolctl)
 848        ("python-scipy" ,python-scipy)
 849        ("python-joblib" ,python-joblib)))
 850     (home-page "https://scikit-learn.org/")
 851     (synopsis "Machine Learning in Python")
 852     (description
 853      "Scikit-learn provides simple and efficient tools for data mining and
 854 data analysis.")
 855     (properties `((python2-variant . ,(delay python2-scikit-learn))))
 856     (license license:bsd-3)))
 857
 858 ;; scikit-learn 0.22 and later only supports Python 3, so we stick with
 859 ;; an older version here.
 860 (define-public python2-scikit-learn
 861   (let ((base (package-with-python2 (strip-python2-variant python-scikit-learn))))
 862     (package
 863       (inherit base)
 864       (version "0.20.4")
 865       (source (origin
 866                 (method git-fetch)
 867                 (uri (git-reference
 868                       (url "https://github.com/scikit-learn/scikit-learn")
 869                       (commit version)))
 870                 (file-name (git-file-name "python-scikit-learn" version))
 871                 (sha256
 872                  (base32
 873                   "08zbzi8yx5wdlxfx9jap61vg1malc9ajf576w7a0liv6jvvrxlpj"))))
 874       (inputs
 875        `(("openblas" ,openblas)))
 876       (native-inputs
 877        `(("python2-pytest" ,python2-pytest)
 878          ("python2-pandas" ,python2-pandas) ;for tests
 879          ("python2-cython" ,python2-cython)))
 880       (propagated-inputs
 881        `(("python2-numpy" ,python2-numpy)
 882          ("python2-scipy" ,python2-scipy)
 883          ("python2-joblib" ,python2-joblib))))))
 884
 885 (define-public python-threadpoolctl
 886   (package
 887     (name "python-threadpoolctl")
 888     (version "2.1.0")
 889     (source
 890       (origin
 891         (method url-fetch)
 892         (uri (pypi-uri "threadpoolctl" version))
 893         (sha256
 894          (base32
 895           "0szsxcm2fbxrn83iynn42bnvrdh7mfsmkhfn8pdn7swblfb7rifx"))))
 896     (build-system python-build-system)
 897     (arguments
 898      `(#:phases
 899        (modify-phases %standard-phases
 900          (replace 'check
 901            (lambda* (#:key tests? inputs outputs #:allow-other-keys)
 902              (when tests?
 903                (add-installed-pythonpath inputs outputs)
 904                (invoke "pytest"))
 905              #t)))))
 906     (native-inputs
 907      `(("python-pytest" ,python-pytest)))
 908     (home-page "https://github.com/joblib/threadpoolctl")
 909     (synopsis "Python helpers for common threading libraries")
 910     (description "Thread-pool Controls provides Python helpers to limit the
 911 number of threads used in the threadpool-backed of common native libraries used
 912 for scientific computing and data science (e.g. BLAS and OpenMP).")
 913     (license license:bsd-3)))
 914
 915 (define-public python-pynndescent
 916   (package
 917     (name "python-pynndescent")
 918     (version "0.4.8")
 919     (source
 920      (origin
 921        (method url-fetch)
 922        (uri (pypi-uri "pynndescent" version))
 923        (sha256
 924         (base32 "0li1fclif50v6xrq7wh3lif9vv5jpj7xhrb0z6g89wwjnp9b9833"))))
 925     (build-system python-build-system)
 926     (native-inputs
 927      `(("python-nose" ,python-nose)))
 928     (propagated-inputs
 929      `(("python-joblib" ,python-joblib)
 930        ("python-llvmlite" ,python-llvmlite)
 931        ("python-numba" ,python-numba)
 932        ("python-scikit-learn" ,python-scikit-learn)
 933        ("python-scipy" ,python-scipy)))
 934     (home-page "https://github.com/lmcinnes/pynndescent")
 935     (synopsis "Nearest neighbor descent for approximate nearest neighbors")
 936     (description
 937      "PyNNDescent provides a Python implementation of Nearest Neighbor Descent
 938 for k-neighbor-graph construction and approximate nearest neighbor search.")
 939     (license license:bsd-2)))
 940
 941 (define-public python-opentsne
 942   (package
 943     (name "python-opentsne")
 944     (version "0.4.4")
 945     (source
 946      (origin
 947        ;; No tests in the PyPI tarball.
 948        (method git-fetch)
 949        (uri (git-reference
 950              (url "https://github.com/pavlin-policar/openTSNE")
 951              (commit (string-append "v" version))))
 952        (file-name (string-append name "-" version "-checkout"))
 953        (sha256
 954         (base32 "08wamsssmyf6511cbmglm67dp48i6xazs89m1cskdk219v90bc76"))))
 955     (build-system python-build-system)
 956     (arguments
 957      `(#:phases
 958        (modify-phases %standard-phases
 959          ;; Benchmarks require the 'macosko2015' data files.
 960          (add-after 'unpack 'delete-benchmark
 961            (lambda _
 962              (delete-file-recursively "benchmarks")
 963              #t))
 964          ;; Numba needs a writable dir to cache functions.
 965          (add-before 'check 'set-numba-cache-dir
 966            (lambda _
 967              (setenv "NUMBA_CACHE_DIR" "/tmp")
 968              #t)))))
 969     (native-inputs
 970      `(("python-cython" ,python-cython)))
 971     (inputs
 972      `(("fftw" ,fftw)))
 973     (propagated-inputs
 974      `(("python-numpy" ,python-numpy)
 975        ("python-pynndescent" ,python-pynndescent)
 976        ("python-scikit-learn" ,python-scikit-learn)
 977        ("python-scipy" ,python-scipy)))
 978     (home-page "https://github.com/pavlin-policar/openTSNE")
 979     (synopsis "Extensible, parallel implementations of t-SNE")
 980     (description
 981      "This is a modular Python implementation of t-Distributed Stochastic
 982 Neighbor Embedding (t-SNE), a popular dimensionality-reduction algorithm for
 983 visualizing high-dimensional data sets.")
 984     (license license:bsd-3)))
 985
 986 (define-public python-scikit-rebate
 987   (package
 988     (name "python-scikit-rebate")
 989     (version "0.6")
 990     (source (origin
 991               (method url-fetch)
 992               (uri (pypi-uri "skrebate" version))
 993               (sha256
 994                (base32
 995                 "1h7qs9gjxpzqabzhb8rmpv3jpmi5iq41kqdibg48299h94iikiw7"))))
 996     (build-system python-build-system)
 997     ;; Pandas is only needed to run the tests.
 998     (native-inputs
 999      `(("python-pandas" ,python-pandas)))
1000     (propagated-inputs
1001      `(("python-numpy" ,python-numpy)
1002        ("python-scipy" ,python-scipy)
1003        ("python-scikit-learn" ,python-scikit-learn)
1004        ("python-joblib" ,python-joblib)))
1005     (home-page "https://epistasislab.github.io/scikit-rebate/")
1006     (synopsis "Relief-based feature selection algorithms for Python")
1007     (description "Scikit-rebate is a scikit-learn-compatible Python
1008 implementation of ReBATE, a suite of Relief-based feature selection algorithms
1009 for Machine Learning.  These algorithms excel at identifying features that are
1010 predictive of the outcome in supervised learning problems, and are especially
1011 good at identifying feature interactions that are normally overlooked by
1012 standard feature selection algorithms.")
1013     (license license:expat)))
1014
1015 (define-public python-autograd
1016   (let* ((commit "442205dfefe407beffb33550846434baa90c4de7")
1017          (revision "0")
1018          (version (git-version "0.0.0" revision commit)))
1019     (package
1020       (name "python-autograd")
1021       (home-page "https://github.com/HIPS/autograd")
1022       (source (origin
1023                 (method git-fetch)
1024                 (uri (git-reference
1025                       (url home-page)
1026                       (commit commit)))
1027                 (sha256
1028                  (base32
1029                   "189sv2xb0mwnjawa9z7mrgdglc1miaq93pnck26r28fi1jdwg0z4"))
1030                 (file-name (git-file-name name version))))
1031       (version version)
1032       (build-system python-build-system)
1033       (native-inputs
1034        `(("python-nose" ,python-nose)
1035          ("python-pytest" ,python-pytest)))
1036       (propagated-inputs
1037        `(("python-future" ,python-future)
1038          ("python-numpy" ,python-numpy)))
1039       (arguments
1040        `(#:phases (modify-phases %standard-phases
1041                     (replace 'check
1042                       (lambda _
1043                         (invoke "py.test" "-v"))))))
1044       (synopsis "Efficiently computes derivatives of NumPy code")
1045       (description "Autograd can automatically differentiate native Python and
1046 NumPy code.  It can handle a large subset of Python's features, including loops,
1047 ifs, recursion and closures, and it can even take derivatives of derivatives
1048 of derivatives.  It supports reverse-mode differentiation
1049 (a.k.a. backpropagation), which means it can efficiently take gradients of
1050 scalar-valued functions with respect to array-valued arguments, as well as
1051 forward-mode differentiation, and the two can be composed arbitrarily.  The
1052 main intended application of Autograd is gradient-based optimization.")
1053       (license license:expat))))
1054
1055 (define-public python2-autograd
1056   (package-with-python2 python-autograd))
1057
1058 (define-public lightgbm
1059   (package
1060     (name "lightgbm")
1061     (version "2.0.12")
1062     (source (origin
1063               (method git-fetch)
1064               (uri (git-reference
1065                      (url "https://github.com/Microsoft/LightGBM")
1066                      (commit (string-append "v" version))))
1067               (sha256
1068                (base32
1069                 "0jlvyn7k81dzrh9ij3zw576wbgiwmmr26rzpdxjn1dbpc3njpvzi"))
1070               (file-name (git-file-name name version))))
1071     (native-inputs
1072      `(("python-pytest" ,python-pytest)
1073        ("python-nose" ,python-nose)))
1074     (inputs
1075      `(("openmpi" ,openmpi)))
1076     (propagated-inputs
1077      `(("python-numpy" ,python-numpy)
1078        ("python-scipy" ,python-scipy)))
1079     (arguments
1080      `(#:configure-flags
1081        '("-DUSE_MPI=ON")
1082        #:phases
1083        (modify-phases %standard-phases
1084          (replace 'check
1085            (lambda _
1086              (with-directory-excursion "../source"
1087                (invoke "pytest" "tests/c_api_test/test_.py")))))))
1088     (build-system cmake-build-system)
1089     (home-page "https://github.com/Microsoft/LightGBM")
1090     (synopsis "Gradient boosting framework based on decision tree algorithms")
1091     (description "LightGBM is a gradient boosting framework that uses tree
1092 based learning algorithms.  It is designed to be distributed and efficient with
1093 the following advantages:
1094
1095 @itemize
1096 @item Faster training speed and higher efficiency
1097 @item Lower memory usage
1098 @item Better accuracy
1099 @item Parallel and GPU learning supported (not enabled in this package)
1100 @item Capable of handling large-scale data
1101 @end itemize\n")
1102     (license license:expat)))
1103
1104 (define-public vowpal-wabbit
1105   ;; Language bindings not included.
1106   (package
1107     (name "vowpal-wabbit")
1108     (version "8.5.0")
1109     (source (origin
1110               (method git-fetch)
1111               (uri (git-reference
1112                      (url "https://github.com/JohnLangford/vowpal_wabbit")
1113                      (commit version)))
1114               (sha256
1115                (base32
1116                 "04bwzk6ifgnz3fmzid8b7avxf9n5pnx9xcjm61nkjng1vv0bpj8x"))
1117               (file-name (git-file-name name version))))
1118     (inputs
1119      `(("boost" ,boost)
1120        ("zlib" ,zlib)))
1121     (arguments
1122      `(#:configure-flags
1123        (list (string-append "--with-boost="
1124                             (assoc-ref %build-inputs "boost")))
1125        #:phases
1126        (modify-phases %standard-phases
1127          (add-after 'unpack 'make-files-writable
1128            (lambda _
1129              (for-each make-file-writable (find-files "." ".*")) #t))
1130          (add-after 'install 'install-more-headers
1131            (lambda* (#:key outputs #:allow-other-keys)
1132              (for-each
1133                (lambda (file)
1134                  (install-file file (string-append
1135                                       (assoc-ref outputs "out")
1136                                       "/include/vowpalwabbit")))
1137                (find-files "vowpalwabbit" "\\.h$"))
1138              #t)))))
1139     (build-system gnu-build-system)
1140     (home-page "https://github.com/JohnLangford/vowpal_wabbit")
1141     (synopsis "Fast machine learning library for online learning")
1142     (description "Vowpal Wabbit is a machine learning system with techniques
1143 such as online, hashing, allreduce, reductions, learning2search, active, and
1144 interactive learning.")
1145     (license license:bsd-3)))
1146
1147 (define-public python2-fastlmm
1148   (package
1149     (name "python2-fastlmm")
1150     (version "0.2.21")
1151     (source
1152      (origin
1153        (method url-fetch)
1154        (uri (pypi-uri "fastlmm" version ".zip"))
1155        (sha256
1156         (base32
1157          "1q8c34rpmwkfy3r4d5172pzdkpfryj561897z9r3x22gq7813x1m"))))
1158     (build-system python-build-system)
1159     (arguments
1160      `(#:tests? #f ; some test files are missing
1161        #:python ,python-2)) ; only Python 2.7 is supported
1162     (propagated-inputs
1163      `(("python2-numpy" ,python2-numpy)
1164        ("python2-scipy" ,python2-scipy)
1165        ("python2-matplotlib" ,python2-matplotlib)
1166        ("python2-pandas" ,python2-pandas)
1167        ("python2-scikit-learn" ,python2-scikit-learn)
1168        ("python2-pysnptools" ,python2-pysnptools)))
1169     (native-inputs
1170      `(("unzip" ,unzip)
1171        ("python2-cython" ,python2-cython)
1172        ("python2-mock" ,python2-mock)
1173        ("python2-nose" ,python2-nose)))
1174     (home-page "http://research.microsoft.com/en-us/um/redmond/projects/mscompbio/fastlmm/")
1175     (synopsis "Perform genome-wide association studies on large data sets")
1176     (description
1177      "FaST-LMM, which stands for Factored Spectrally Transformed Linear Mixed
1178 Models, is a program for performing both single-SNP and SNP-set genome-wide
1179 association studies (GWAS) on extremely large data sets.")
1180     (license license:asl2.0)))
1181
1182 ;; There have been no proper releases yet.
1183 (define-public kaldi
1184   (let ((commit "d4791c0f3fc1a09c042dac365e120899ee2ad21e")
1185         (revision "2"))
1186     (package
1187       (name "kaldi")
1188       (version (git-version "0" revision commit))
1189       (source (origin
1190                 (method git-fetch)
1191                 (uri (git-reference
1192                       (url "https://github.com/kaldi-asr/kaldi")
1193                       (commit commit)))
1194                 (file-name (git-file-name name version))
1195                 (sha256
1196                  (base32
1197                   "07k80my6f19mhrkwbzhjsnpf9871wmrwkl0ym468i830w67qyjrz"))))
1198       (build-system gnu-build-system)
1199       (arguments
1200        `(#:test-target "test"
1201          #:phases
1202          (modify-phases %standard-phases
1203            (add-after 'unpack 'chdir
1204              (lambda _ (chdir "src") #t))
1205            (replace 'configure
1206              (lambda* (#:key build system inputs outputs #:allow-other-keys)
1207                (when (not (or (string-prefix? "x86_64" system)
1208                               (string-prefix? "i686" system)))
1209                  (substitute* "makefiles/linux_openblas.mk"
1210                    (("-msse -msse2") "")))
1211                (substitute* "makefiles/default_rules.mk"
1212                  (("/bin/bash") (which "bash")))
1213                (substitute* "Makefile"
1214                  (("ext_depend: check_portaudio")
1215                   "ext_depend:"))
1216                (substitute* '("online/Makefile"
1217                               "onlinebin/Makefile"
1218                               "gst-plugin/Makefile")
1219                  (("../../tools/portaudio/install")
1220                   (assoc-ref inputs "portaudio")))
1221
1222                ;; This `configure' script doesn't support variables passed as
1223                ;; arguments, nor does it support "prefix".
1224                (let ((out (assoc-ref outputs "out"))
1225                      (openblas (assoc-ref inputs "openblas"))
1226                      (openfst (assoc-ref inputs "openfst")))
1227                  (substitute* "configure"
1228                    (("check_for_slow_expf;") "")
1229                    ;; This affects the RPATH and also serves as the installation
1230                    ;; directory.
1231                    (("KALDILIBDIR=`pwd`/lib")
1232                     (string-append "KALDILIBDIR=" out "/lib")))
1233                  (mkdir-p out) ; must exist
1234                  (setenv "CONFIG_SHELL" (which "bash"))
1235                  (setenv "OPENFST_VER" ,(package-version openfst))
1236                  (invoke "./configure"
1237                          "--use-cuda=no"
1238                          "--shared"
1239                          (string-append "--openblas-root=" openblas)
1240                          (string-append "--fst-root=" openfst)))))
1241            (add-after 'build 'build-ext-and-gstreamer-plugin
1242              (lambda _
1243                (invoke "make" "-C" "online" "depend")
1244                (invoke "make" "-C" "online")
1245                (invoke "make" "-C" "onlinebin" "depend")
1246                (invoke "make" "-C" "onlinebin")
1247                (invoke "make" "-C" "gst-plugin" "depend")
1248                (invoke "make" "-C" "gst-plugin")
1249                #t))
1250            ;; TODO: also install the executables.
1251            (replace 'install
1252              (lambda* (#:key outputs #:allow-other-keys)
1253                (let* ((out (assoc-ref outputs "out"))
1254                       (inc (string-append out "/include"))
1255                       (lib (string-append out "/lib")))
1256                  (mkdir-p lib)
1257                  ;; The build phase installed symlinks to the actual
1258                  ;; libraries.  Install the actual targets.
1259                  (for-each (lambda (file)
1260                              (let ((target (readlink file)))
1261                                (delete-file file)
1262                                (install-file target lib)))
1263                            (find-files lib "\\.so"))
1264                  ;; Install headers
1265                  (for-each (lambda (file)
1266                              (let ((target-dir (string-append inc "/" (dirname file))))
1267                                (install-file file target-dir)))
1268                            (find-files "." "\\.h"))
1269                  (install-file "gst-plugin/libgstonlinegmmdecodefaster.so"
1270                                (string-append lib "/gstreamer-1.0"))
1271                  #t))))))
1272       (inputs
1273        `(("alsa-lib" ,alsa-lib)
1274          ("gfortran" ,gfortran "lib")
1275          ("glib" ,glib)
1276          ("gstreamer" ,gstreamer)
1277          ("jack" ,jack-1)
1278          ("openblas" ,openblas)
1279          ("openfst" ,openfst)
1280          ("portaudio" ,portaudio)
1281          ("python" ,python)))
1282       (native-inputs
1283        `(("glib" ,glib "bin")             ; glib-genmarshal
1284          ("grep" ,grep)
1285          ("sed" ,sed)
1286          ("pkg-config" ,pkg-config)
1287          ("which" ,which)))
1288       (home-page "https://kaldi-asr.org/")
1289       (synopsis "Speech recognition toolkit")
1290       (description "Kaldi is an extensible toolkit for speech recognition
1291 written in C++.")
1292       (license license:asl2.0))))
1293
1294 (define-public gst-kaldi-nnet2-online
1295   (let ((commit "cb227ef43b66a9835c14eb0ad39e08ee03c210ad")
1296         (revision "2"))
1297     (package
1298       (name "gst-kaldi-nnet2-online")
1299       (version (git-version "0" revision commit))
1300       (source (origin
1301                 (method git-fetch)
1302                 (uri (git-reference
1303                       (url "https://github.com/alumae/gst-kaldi-nnet2-online")
1304                       (commit commit)))
1305                 (file-name (git-file-name name version))
1306                 (sha256
1307                  (base32
1308                   "1i6ffwiavxx07ri0lxix6s8q0r31x7i4xxvhys5jxkixf5q34w8g"))))
1309       (build-system gnu-build-system)
1310       (arguments
1311        `(#:tests? #f                    ; there are none
1312          #:make-flags
1313          (list (string-append "SHELL="
1314                               (assoc-ref %build-inputs "bash") "/bin/bash")
1315                (string-append "KALDI_ROOT="
1316                               (assoc-ref %build-inputs "kaldi-src"))
1317                (string-append "KALDILIBDIR="
1318                               (assoc-ref %build-inputs "kaldi") "/lib")
1319                "KALDI_FLAVOR=dynamic")
1320          #:phases
1321          (modify-phases %standard-phases
1322            (add-after 'unpack 'chdir
1323              (lambda _ (chdir "src") #t))
1324            (replace 'configure
1325              (lambda* (#:key inputs #:allow-other-keys)
1326                (let ((glib (assoc-ref inputs "glib")))
1327                  (setenv "CXXFLAGS" "-fPIC")
1328                  (setenv "CPLUS_INCLUDE_PATH"
1329                          (string-append glib "/include/glib-2.0:"
1330                                         glib "/lib/glib-2.0/include:"
1331                                         (assoc-ref inputs "gstreamer")
1332                                         "/include/gstreamer-1.0")))
1333                (substitute* "Makefile"
1334                  (("include \\$\\(KALDI_ROOT\\)/src/kaldi.mk") "")
1335                  (("\\$\\(error Cannot find") "#"))
1336                #t))
1337            (add-before 'build 'build-depend
1338              (lambda* (#:key make-flags #:allow-other-keys)
1339                (apply invoke "make" "depend" make-flags)))
1340            (replace 'install
1341              (lambda* (#:key outputs #:allow-other-keys)
1342                (let* ((out (assoc-ref outputs "out"))
1343                       (lib (string-append out "/lib/gstreamer-1.0")))
1344                  (install-file "libgstkaldinnet2onlinedecoder.so" lib)
1345                  #t))))))
1346       (inputs
1347        `(("glib" ,glib)
1348          ("gstreamer" ,gstreamer)
1349          ("jansson" ,jansson)
1350          ("openfst" ,openfst)
1351          ("kaldi" ,kaldi)))
1352       (native-inputs
1353        `(("bash" ,bash)
1354          ("glib:bin" ,glib "bin")       ; glib-genmarshal
1355          ("kaldi-src" ,(package-source kaldi))
1356          ("pkg-config" ,pkg-config)))
1357       (home-page "https://kaldi-asr.org/")
1358       (synopsis "Gstreamer plugin for decoding speech")
1359       (description "This package provides a GStreamer plugin that wraps
1360 Kaldi's @code{SingleUtteranceNnet2Decoder}.  It requires iVector-adapted DNN
1361 acoustic models.  The iVectors are adapted to the current audio stream
1362 automatically.")
1363       (license license:asl2.0))))
1364
1365 (define-public kaldi-gstreamer-server
1366   ;; This is the tip of the py3 branch
1367   (let ((commit "f68cab490be7eb0da2af1475fbc16655f50a60cb")
1368         (revision "2"))
1369     (package
1370       (name "kaldi-gstreamer-server")
1371       (version (git-version "0" revision commit))
1372       (source (origin
1373                 (method git-fetch)
1374                 (uri (git-reference
1375                       (url "https://github.com/alumae/kaldi-gstreamer-server")
1376                       (commit commit)))
1377                 (file-name (git-file-name name version))
1378                 (sha256
1379                  (base32
1380                   "17lh1368vkg8ngrcbn2phvigzlmalrqg6djx2gg61qq1a0nj87dm"))))
1381       (build-system gnu-build-system)
1382       (arguments
1383        `(#:tests? #f ; there are no tests that can be run automatically
1384          #:modules ((guix build utils)
1385                     (guix build gnu-build-system)
1386                     (srfi srfi-26))
1387          #:phases
1388          (modify-phases %standard-phases
1389            (delete 'configure)
1390            (replace 'build
1391              (lambda* (#:key outputs #:allow-other-keys)
1392                ;; Disable hash randomization to ensure the generated .pycs
1393                ;; are reproducible.
1394                (setenv "PYTHONHASHSEED" "0")
1395                (with-directory-excursion "kaldigstserver"
1396                  ;; See https://github.com/alumae/kaldi-gstreamer-server/issues/232
1397                  (substitute* "master_server.py"
1398                    (("\\.replace\\('\\\\.*") ")"))
1399
1400                  ;; This is a Python 2 file
1401                  (delete-file "decoder_test.py")
1402                  (delete-file "test-buffer.py")
1403
1404                  (for-each (lambda (file)
1405                              (apply invoke
1406                                     `("python"
1407                                       "-m" "compileall"
1408                                       "-f" ; force rebuild
1409                                       ,file)))
1410                            (find-files "." "\\.py$")))
1411                #t))
1412            (replace 'install
1413              (lambda* (#:key inputs outputs #:allow-other-keys)
1414                (let* ((out (assoc-ref outputs "out"))
1415                       (bin (string-append out "/bin"))
1416                       (share (string-append out "/share/kaldi-gstreamer-server/")))
1417                  ;; Install Python files
1418                  (with-directory-excursion "kaldigstserver"
1419                    (for-each (cut install-file <> share)
1420                              (find-files "." ".*")))
1421
1422                  ;; Install sample configuration files
1423                  (for-each (cut install-file <> share)
1424                            (find-files "." "\\.yaml"))
1425
1426                  ;; Install executables
1427                  (mkdir-p bin)
1428                  (let* ((server (string-append bin "/kaldi-gst-server"))
1429                         (client (string-append bin "/kaldi-gst-client"))
1430                         (worker (string-append bin "/kaldi-gst-worker"))
1431                         (PYTHONPATH (getenv "PYTHONPATH"))
1432                         (GST_PLUGIN_PATH (string-append
1433                                           (assoc-ref inputs "gst-kaldi-nnet2-online")
1434                                           "/lib/gstreamer-1.0:${GST_PLUGIN_PATH}"))
1435                         (wrap (lambda (wrapper what)
1436                                 (with-output-to-file wrapper
1437                                   (lambda _
1438                                     (format #t
1439                                             "#!~a
1440 export PYTHONPATH=~a
1441 export GST_PLUGIN_PATH=~a
1442 exec ~a ~a/~a \"$@\"~%"
1443                                             (which "bash") PYTHONPATH GST_PLUGIN_PATH
1444                                             (which "python") share what)))
1445                                 (chmod wrapper #o555))))
1446                    (for-each wrap
1447                              (list server client worker)
1448                              (list "master_server.py"
1449                                    "client.py"
1450                                    "worker.py")))
1451                  #t))))))
1452       (inputs
1453        `(("gst-kaldi-nnet2-online" ,gst-kaldi-nnet2-online)
1454          ("python" ,python-wrapper)
1455          ("python-pygobject" ,python-pygobject)
1456          ("python-pyyaml" ,python-pyyaml)
1457          ("python-tornado" ,python-tornado-6)))
1458       (home-page "https://github.com/alumae/kaldi-gstreamer-server")
1459       (synopsis "Real-time full-duplex speech recognition server")
1460       (description "This is a real-time full-duplex speech recognition server,
1461 based on the Kaldi toolkit and the GStreamer framework and implemented in
1462 Python.")
1463       (license license:bsd-2))))
1464
1465 ;; Note that Tensorflow includes a "third_party" directory, which seems to not
1466 ;; only contain modified subsets of upstream library source code, but also
1467 ;; adapter headers provided by Google (such as the fft.h header, which is not
1468 ;; part of the upstream project code).  The Tensorflow code includes headers
1469 ;; from the "third_party" directory.  It does not look like we can replace
1470 ;; these headers with unmodified upstream files, so we keep them.
1471 (define-public tensorflow
1472   (package
1473     (name "tensorflow")
1474     (version "1.9.0")
1475     (source
1476      (origin
1477        (method git-fetch)
1478        (uri (git-reference
1479              (url "https://github.com/tensorflow/tensorflow")
1480              (commit (string-append "v" version))))
1481        (file-name (string-append "tensorflow-" version "-checkout"))
1482        (sha256
1483         (base32
1484          "0a9kwha395g3wgxfwln5j8vn9nkspmd75xldrlqdq540w996g8xa"))))
1485     (build-system cmake-build-system)
1486     (arguments
1487      `(#:tests? #f                      ; no "check" target
1488        #:build-type "Release"
1489        #:configure-flags
1490        (let ((protobuf (assoc-ref %build-inputs "protobuf"))
1491              (protobuf:native (assoc-ref %build-inputs "protobuf:native"))
1492              (jsoncpp (assoc-ref %build-inputs "jsoncpp"))
1493              (snappy (assoc-ref %build-inputs "snappy"))
1494              (sqlite (assoc-ref %build-inputs "sqlite")))
1495          (list
1496           ;; Use protobuf from Guix
1497           (string-append "-Dprotobuf_STATIC_LIBRARIES="
1498                          protobuf "/lib/libprotobuf.so")
1499           (string-append "-DPROTOBUF_PROTOC_EXECUTABLE="
1500                          protobuf:native "/bin/protoc")
1501
1502           ;; Use snappy from Guix
1503           (string-append "-Dsnappy_STATIC_LIBRARIES="
1504                          snappy "/lib/libsnappy.so")
1505           ;; Yes, this is not actually the include directory but a prefix...
1506           (string-append "-Dsnappy_INCLUDE_DIR=" snappy)
1507
1508           ;; Use jsoncpp from Guix
1509           (string-append "-Djsoncpp_STATIC_LIBRARIES="
1510                          jsoncpp "/lib/libjsoncpp.so")
1511           ;; Yes, this is not actually the include directory but a prefix...
1512           (string-append "-Djsoncpp_INCLUDE_DIR=" jsoncpp)
1513
1514           ;; Use sqlite from Guix
1515           (string-append "-Dsqlite_STATIC_LIBRARIES="
1516                          sqlite "/lib/libsqlite.a")
1517
1518           ;; Use system libraries wherever possible.  Currently, this
1519           ;; only affects zlib.
1520           "-Dsystemlib_ALL=ON"
1521           "-Dtensorflow_ENABLE_POSITION_INDEPENDENT_CODE=ON"
1522           "-Dtensorflow_BUILD_SHARED_LIB=ON"
1523           "-Dtensorflow_OPTIMIZE_FOR_NATIVE_ARCH=OFF"
1524           "-Dtensorflow_ENABLE_SSL_SUPPORT=OFF"
1525           "-Dtensorflow_BUILD_CONTRIB_KERNELS=OFF"))
1526        #:make-flags
1527        (list "CC=gcc")
1528        #:modules ((ice-9 ftw)
1529                   (guix build utils)
1530                   (guix build cmake-build-system)
1531                   ((guix build python-build-system)
1532                    #:select (python-version)))
1533        #:imported-modules (,@%cmake-build-system-modules
1534                             (guix build python-build-system))
1535        #:phases
1536        (modify-phases %standard-phases
1537          (add-after 'unpack 'set-source-file-times-to-1980
1538            ;; At the end of the tf_python_build_pip_package target, a ZIP
1539            ;; archive should be generated via bdist_wheel, but it fails with
1540            ;; "ZIP does not support timestamps before 1980".  Luckily,
1541            ;; SOURCE_DATE_EPOCH is respected, which we set to some time in
1542            ;; 1980.
1543            (lambda _ (setenv "SOURCE_DATE_EPOCH" "315532800") #t))
1544          ;; See https://github.com/tensorflow/tensorflow/issues/20517#issuecomment-406373913
1545          (add-after 'unpack 'python3.7-compatibility
1546            (lambda _
1547              (substitute* '("tensorflow/python/eager/pywrap_tfe_src.cc"
1548                             "tensorflow/python/lib/core/ndarray_tensor.cc"
1549                             "tensorflow/python/lib/core/py_func.cc")
1550                (("PyUnicode_AsUTF8") "(char *)PyUnicode_AsUTF8"))
1551              (substitute* "tensorflow/c/eager/c_api.h"
1552                (("unsigned char async")
1553                 "unsigned char is_async"))
1554
1555              ;; Remove dependency on tensorboard, a complicated but probably
1556              ;; optional package.
1557              (substitute* "tensorflow/tools/pip_package/setup.py"
1558                ((".*'tensorboard >.*") ""))
1559
1560              ;; Fix the build with python-3.8, taken from rejected upstream patch:
1561              ;; https://github.com/tensorflow/tensorflow/issues/34197
1562              (substitute* (find-files "tensorflow/python" ".*\\.cc$")
1563                (("(nullptr,)(\\ +/. tp_print)" _ _ tp_print)
1564                 (string-append "NULL,   " tp_print)))
1565              #t))
1566          (add-after 'python3.7-compatibility 'chdir
1567            (lambda _ (chdir "tensorflow/contrib/cmake") #t))
1568          (add-after 'chdir 'disable-downloads
1569            (lambda* (#:key inputs #:allow-other-keys)
1570              (substitute* (find-files "external" "\\.cmake$")
1571                (("GIT_REPOSITORY.*") "")
1572                (("GIT_TAG.*") "")
1573                (("PREFIX ")
1574                 "DOWNLOAD_COMMAND \"\"\nPREFIX "))
1575
1576              ;; Use packages from Guix
1577              (let ((grpc (assoc-ref inputs "grpc")))
1578                (substitute* "CMakeLists.txt"
1579                  ;; Sqlite
1580                  (("include\\(sqlite\\)") "")
1581                  (("\\$\\{sqlite_STATIC_LIBRARIES\\}")
1582                   (string-append (assoc-ref inputs "sqlite")
1583                                  "/lib/libsqlite3.so"))
1584                  (("sqlite_copy_headers_to_destination") "")
1585
1586                  ;; PNG
1587                  (("include\\(png\\)") "")
1588                  (("\\$\\{png_STATIC_LIBRARIES\\}")
1589                   (string-append (assoc-ref inputs "libpng")
1590                                  "/lib/libpng16.so"))
1591                  (("png_copy_headers_to_destination") "")
1592
1593                  ;; JPEG
1594                  (("include\\(jpeg\\)") "")
1595                  (("\\$\\{jpeg_STATIC_LIBRARIES\\}")
1596                   (string-append (assoc-ref inputs "libjpeg")
1597                                  "/lib/libjpeg.so"))
1598                  (("jpeg_copy_headers_to_destination") "")
1599
1600                  ;; GIF
1601                  (("include\\(gif\\)") "")
1602                  (("\\$\\{gif_STATIC_LIBRARIES\\}")
1603                   (string-append (assoc-ref inputs "giflib")
1604                                  "/lib/libgif.so"))
1605                  (("gif_copy_headers_to_destination") "")
1606
1607                  ;; lmdb
1608                  (("include\\(lmdb\\)") "")
1609                  (("\\$\\{lmdb_STATIC_LIBRARIES\\}")
1610                   (string-append (assoc-ref inputs "lmdb")
1611                                  "/lib/liblmdb.so"))
1612                  (("lmdb_copy_headers_to_destination") "")
1613
1614                  ;; Protobuf
1615                  (("include\\(protobuf\\)") "")
1616                  (("protobuf_copy_headers_to_destination") "")
1617                  (("^ +protobuf") "")
1618
1619                  ;; gRPC
1620                  (("include\\(grpc\\)")
1621                   "find_package(grpc REQUIRED NAMES gRPC)")
1622                  (("list\\(APPEND tensorflow_EXTERNAL_DEPENDENCIES grpc\\)") "")
1623
1624                  ;; Eigen
1625                  (("include\\(eigen\\)")
1626                   (string-append "find_package(eigen REQUIRED NAMES Eigen3)
1627 set(eigen_INCLUDE_DIRS ${CMAKE_CURRENT_BINARY_DIR}/external/eigen_archive "
1628                                  (assoc-ref inputs "eigen") "/include/eigen3)"))
1629                  (("^ +eigen") "")
1630
1631                  ;; snappy
1632                  (("include\\(snappy\\)")
1633                   "add_definitions(-DTF_USE_SNAPPY)")
1634                  (("list\\(APPEND tensorflow_EXTERNAL_DEPENDENCIES snappy\\)") "")
1635
1636                  ;; jsoncpp
1637                  (("include\\(jsoncpp\\)") "")
1638                  (("^ +jsoncpp") ""))
1639
1640                (substitute* "tf_core_framework.cmake"
1641                  ((" grpc") "")
1642                  (("\\$\\{GRPC_BUILD\\}/grpc_cpp_plugin")
1643                   (which "grpc_cpp_plugin"))
1644                  ;; Link with gRPC libraries
1645                  (("add_library\\(tf_protos_cc.*" m)
1646                   (string-append m
1647                                  (format #f "\ntarget_link_libraries(tf_protos_cc PRIVATE \
1648 ~a/lib/libgrpc++_unsecure.a \
1649 ~a/lib/libgrpc_unsecure.a \
1650 ~a/lib/libaddress_sorting.a \
1651 ~a/lib/libgpr.a \
1652 ~a//lib/libcares.so
1653 )\n"
1654                                          grpc grpc grpc grpc
1655                                          (assoc-ref inputs "c-ares"))))))
1656              (substitute* "tf_tools.cmake"
1657                (("add_dependencies\\(\\$\\{proto_text.*") ""))
1658              ;; Remove dependency on bundled grpc
1659              (substitute* "tf_core_distributed_runtime.cmake"
1660                (("tf_core_cpu grpc") "tf_core_cpu"))
1661
1662              ;; This directory is a dependency of many targets.
1663              (mkdir-p "protobuf")
1664              #t))
1665          (add-after 'configure 'unpack-third-party-sources
1666            (lambda* (#:key inputs #:allow-other-keys)
1667              ;; This is needed to configure bundled packages properly.
1668              (setenv "CONFIG_SHELL" (which "bash"))
1669              (for-each
1670               (lambda (name)
1671                 (let* ((what  (assoc-ref inputs (string-append name "-src")))
1672                        (name* (string-map (lambda (c)
1673                                             (if (char=? c #\-)
1674                                                 #\_ c)) name))
1675                        (where (string-append "../build/" name* "/src/" name*)))
1676                   (cond
1677                    ((string-suffix? ".zip" what)
1678                     (mkdir-p where)
1679                     (with-directory-excursion where
1680                       (invoke "unzip" what)))
1681                    ((string-suffix? ".tar.gz" what)
1682                     (mkdir-p where)
1683                     (invoke "tar" "xf" what
1684                             "-C" where "--strip-components=1"))
1685                    (else
1686                     (let ((parent (dirname where)))
1687                       (mkdir-p parent)
1688                       (with-directory-excursion parent
1689                         (when (file-exists? name*)
1690                           (delete-file-recursively name*))
1691                         (copy-recursively what name*)
1692                         (map make-file-writable
1693                              (find-files name* ".*"))))))))
1694               (list "boringssl"
1695                     "cub"
1696                     "double-conversion"
1697                     "farmhash"
1698                     "fft2d"
1699                     "highwayhash"
1700                     "nsync"
1701                     "re2"))
1702
1703              (rename-file "../build/cub/src/cub/cub-1.8.0/"
1704                           "../build/cub/src/cub/cub/")
1705              #t))
1706          (add-after 'unpack 'fix-python-build
1707            (lambda* (#:key inputs outputs #:allow-other-keys)
1708              (mkdir-p "protobuf-src")
1709              (invoke "tar" "xf" (assoc-ref inputs "protobuf:src")
1710                      "-C" "protobuf-src" "--strip-components=1")
1711              (mkdir-p "eigen-src")
1712              (invoke "tar" "xf" (assoc-ref inputs "eigen:src")
1713                      "-C" "eigen-src" "--strip-components=1")
1714
1715              (substitute* "tensorflow/contrib/cmake/tf_python.cmake"
1716                ;; Ensure that all Python dependencies can be found at build time.
1717                (("PYTHONPATH=\\$\\{CMAKE_CURRENT_BINARY_DIR\\}/tf_python" m)
1718                 (string-append m ":" (getenv "PYTHONPATH")))
1719                ;; Take protobuf source files from our source package.
1720                (("\\$\\{CMAKE_CURRENT_BINARY_DIR\\}/protobuf/src/protobuf/src/google")
1721                 (string-append (getcwd) "/protobuf-src/src/google")))
1722
1723              (substitute* '("tensorflow/contrib/cmake/tf_shared_lib.cmake"
1724                             "tensorflow/contrib/cmake/tf_python.cmake")
1725                ;; Take Eigen source files from our source package.
1726                (("\\$\\{CMAKE_CURRENT_BINARY_DIR\\}/eigen/src/eigen/")
1727                 (string-append (getcwd) "/eigen-src/"))
1728                ;; Take Eigen headers from our own package.
1729                (("\\$\\{CMAKE_CURRENT_BINARY_DIR\\}/external/eigen_archive")
1730                 (string-append (assoc-ref inputs "eigen") "/include/eigen3")))
1731
1732              ;; Correct the RUNPATH of ops libraries generated for Python.
1733              ;; TODO: this doesn't work :(
1734              ;; /gnu/store/...-tensorflow-1.9.0/lib/python3.7/site-packages/tensorflow/contrib/seq2seq/python/ops/lib_beam_search_ops.so:
1735              ;; warning: RUNPATH contains bogus entries: ("/tmp/guix-build-tensorflow-1.9.0.drv-0/source/tensorflow/contrib/build")
1736              ;; /gnu/store/...-tensorflow-1.9.0/lib/python3.7/site-packages/tensorflow/contrib/seq2seq/python/ops/lib_beam_search_ops.so:
1737              ;; error: depends on 'libpywrap_tensorflow_internal.so', which
1738              ;; cannot be found in RUNPATH ...
1739              (substitute* "tensorflow/contrib/cmake/tf_cc_ops.cmake"
1740                (("set_target_properties.*")
1741                 (string-append "set_target_properties(${_AT_TARGET} PROPERTIES \
1742 COMPILE_FLAGS ${target_compile_flags} \
1743 INSTALL_RPATH_USE_LINK_PATH TRUE \
1744 INSTALL_RPATH " (assoc-ref outputs "out") "/lib)\n")))
1745              #t))
1746          (add-after 'build 'build-pip-package
1747            (lambda* (#:key outputs #:allow-other-keys)
1748              (setenv "LDFLAGS"
1749                      (string-append "-Wl,-rpath="
1750                                     (assoc-ref outputs "out") "/lib"))
1751              (invoke "make" "tf_python_build_pip_package")
1752              #t))
1753          (add-after 'build-pip-package 'install-python
1754            (lambda* (#:key inputs outputs #:allow-other-keys)
1755              (let ((out (assoc-ref outputs "out"))
1756                    (wheel (car (find-files "../build/tf_python/dist/" "\\.whl$")))
1757                    (python-version (python-version
1758                                      (assoc-ref inputs "python"))))
1759                (invoke "python" "-m" "pip" "install" wheel
1760                        (string-append "--prefix=" out))
1761
1762                ;; XXX: broken RUNPATH, see fix-python-build phase.
1763                (delete-file
1764                 (string-append
1765                  out "/lib/python" python-version
1766                  "/site-packages/tensorflow/contrib/"
1767                  "seq2seq/python/ops/lib_beam_search_ops.so"))
1768                #t))))))
1769     (native-inputs
1770      `(("pkg-config" ,pkg-config)
1771        ("protobuf:native" ,protobuf-3.6) ; protoc
1772        ("protobuf:src" ,(package-source protobuf-3.6))
1773        ("eigen:src" ,(package-source eigen-for-tensorflow))
1774        ;; install_pip_packages.sh wants setuptools 39.1.0 specifically.
1775        ("python-setuptools" ,python-setuptools-for-tensorflow)
1776
1777        ;; The commit hashes and URLs for third-party source code are taken
1778        ;; from "tensorflow/workspace.bzl".
1779        ("boringssl-src"
1780         ,(let ((commit "ee7aa02")
1781                (revision "1"))
1782            (origin
1783              (method git-fetch)
1784              (uri (git-reference
1785                    (url "https://boringssl.googlesource.com/boringssl")
1786                    (commit commit)))
1787              (file-name (string-append "boringssl-0-" revision
1788                                        (string-take commit 7)
1789                                        "-checkout"))
1790              (sha256
1791               (base32
1792                "1jf693q0nw0adsic6cgmbdx6g7wr4rj4vxa8j1hpn792fqhd8wgw")))))
1793        ("cub-src"
1794         ,(let ((version "1.8.0"))
1795            (origin
1796              (method url-fetch)
1797              (uri (string-append "https://mirror.bazel.build/github.com/NVlabs/"
1798                                  "cub/archive/" version ".zip"))
1799              (file-name (string-append "cub-" version ".zip"))
1800              (sha256
1801               (base32
1802                "1hsqikqridb90dkxkjr2918dcry6pfh46ccnwrzawl56aamhdykb")))))
1803        ("double-conversion-src"
1804         ,(let ((commit "5664746")
1805                (revision "1"))
1806            (origin
1807              (method git-fetch)
1808              (uri (git-reference
1809                    (url "https://github.com/google/double-conversion")
1810                    (commit commit)))
1811              (file-name
1812               (git-file-name "double-conversion"
1813                              (string-append "0-" revision "."
1814                                             (string-take commit 7))))
1815              (sha256
1816               (base32
1817                "1h5lppqqxcvdg5jq42i5msgwx20ryij3apvmndflngrgdpc04gn1")))))
1818        ("farmhash-src"
1819         ,(let ((commit "816a4ae622e964763ca0862d9dbd19324a1eaf45"))
1820            (origin
1821              (method url-fetch)
1822              (uri (string-append
1823                    "https://mirror.bazel.build/github.com/google/farmhash/archive/"
1824                    commit ".tar.gz"))
1825              (file-name (string-append "farmhash-0-" (string-take commit 7)
1826                                        ".tar.gz"))
1827              (sha256
1828               (base32
1829                "185b2xdxl4d4cnsnv6abg8s22gxvx8673jq2yaq85bz4cdy58q35")))))
1830        ;; The license notice on the home page at
1831        ;; http://www.kurims.kyoto-u.ac.jp/~ooura/fft.html says:
1832        ;;   Copyright Takuya OOURA, 1996-2001
1833        ;;
1834        ;;   You may use, copy, modify and distribute this code for any purpose
1835        ;;   (include commercial use) and without fee. Please refer to this
1836        ;;   package when you modify this code.
1837        ;;
1838        ;; We take the identical tarball from the Bazel mirror, because the URL
1839        ;; at the home page is not versioned and might change.
1840        ("fft2d-src"
1841         ,(origin
1842            (method url-fetch)
1843            (uri "https://mirror.bazel.build/www.kurims.kyoto-u.ac.jp/~ooura/fft.tgz")
1844            (file-name "fft2d.tar.gz")
1845            (sha256
1846             (base32
1847              "15jjkfvhqvl2c0753d2di8hz0pyzn598g74wqy79awdrf1y67fsj"))))
1848        ("highwayhash-src"
1849         ,(let ((commit "be5edafc2e1a455768e260ccd68ae7317b6690ee")
1850                (revision "1"))
1851            (origin
1852              (method git-fetch)
1853              (uri (git-reference
1854                    (url "https://github.com/google/highwayhash")
1855                    (commit commit)))
1856              (file-name (string-append "highwayhash-0-" revision
1857                                        (string-take commit 7)
1858                                        "-checkout"))
1859              (sha256
1860               (base32
1861                "154jwf98cyy54hldr94pgjn85zynly3abpnc1avmb8a18lzwjyb6")))))
1862        ("nsync-src"
1863         ,(let ((version "0559ce013feac8db639ee1bf776aca0325d28777")
1864                (revision "1"))
1865            (origin
1866              (method url-fetch)
1867              (uri (string-append "https://mirror.bazel.build/"
1868                                  "github.com/google/nsync/archive/"
1869                                  version ".tar.gz"))
1870              (file-name (string-append "nsync-0." revision
1871                                        "-" (string-take version 7)
1872                                        ".tar.gz"))
1873              (sha256
1874               (base32
1875                "0qdkyqym34x739mmzv97ah5r7ph462v5xkxqxvidmcfqbi64b132")))))
1876        ("re2-src"
1877         ,(let ((commit "e7efc48")
1878                (revision "1"))
1879            (origin
1880              (method git-fetch)
1881              (uri (git-reference
1882                    (url "https://github.com/google/re2")
1883                    (commit commit)))
1884              (file-name (string-append "re2-0-" revision
1885                                        (string-take commit 7)
1886                                        "-checkout"))
1887              (sha256
1888               (base32
1889                "161g9841rjfsy5pn52fcis0s9hdr7rxvb06pad38j5rppfihvign")))))
1890        ("googletest" ,googletest)
1891        ("swig" ,swig)
1892        ("unzip" ,unzip)))
1893     (propagated-inputs
1894      `(("python-absl-py" ,python-absl-py)
1895        ("python-astor" ,python-astor)
1896        ("python-gast" ,python-gast)
1897        ("python-grpcio" ,python-grpcio)
1898        ("python-numpy" ,python-numpy)
1899        ("python-protobuf" ,python-protobuf-3.6)
1900        ("python-six" ,python-six)
1901        ("python-termcolo" ,python-termcolor)
1902        ("python-wheel" ,python-wheel)))
1903     (inputs
1904      `(("c-ares" ,c-ares)
1905        ("eigen" ,eigen-for-tensorflow)
1906        ("gemmlowp" ,gemmlowp-for-tensorflow)
1907        ("lmdb" ,lmdb)
1908        ("libjpeg" ,libjpeg-turbo)
1909        ("libpng" ,libpng)
1910        ("giflib" ,giflib)
1911        ("grpc" ,grpc-1.16.1 "static")
1912        ("grpc:bin" ,grpc-1.16.1)
1913        ("jsoncpp" ,jsoncpp-for-tensorflow)
1914        ("snappy" ,snappy)
1915        ("sqlite" ,sqlite)
1916        ("protobuf" ,protobuf-3.6)
1917        ("python" ,python-wrapper)
1918        ("zlib" ,zlib)))
1919     (home-page "https://tensorflow.org")
1920     (synopsis "Machine learning framework")
1921     (description
1922      "TensorFlow is a flexible platform for building and training machine
1923 learning models.  It provides a library for high performance numerical
1924 computation and includes high level Python APIs, including both a sequential
1925 API for beginners that allows users to build models quickly by plugging
1926 together building blocks and a subclassing API with an imperative style for
1927 advanced research.")
1928     (license license:asl2.0)))
1929
1930 (define-public python-iml
1931   (package
1932     (name "python-iml")
1933     (version "0.6.2")
1934     (source
1935      (origin
1936        (method url-fetch)
1937        (uri (pypi-uri "iml" version))
1938        (sha256
1939         (base32
1940          "1k8szlpm19rcwcxdny9qdm3gmaqq8akb4xlvrzyz8c2d679aak6l"))))
1941     (build-system python-build-system)
1942     (propagated-inputs
1943      `(("ipython" ,python-ipython)
1944        ("numpy" ,python-numpy)
1945        ("pandas" ,python-pandas)
1946        ("scipy" ,python-scipy)))
1947     (native-inputs
1948      `(("nose" ,python-nose)))
1949     (home-page "https://github.com/interpretable-ml/iml")
1950     (synopsis "Interpretable Machine Learning (iML) package")
1951     (description "Interpretable ML (iML) is a set of data type objects,
1952 visualizations, and interfaces that can be used by any method designed to
1953 explain the predictions of machine learning models (or really the output of
1954 any function).  It currently contains the interface and IO code from the Shap
1955 project, and it will potentially also do the same for the Lime project.")
1956     (license license:expat)))
1957
1958 (define-public python-keras-applications
1959   (package
1960     (name "python-keras-applications")
1961     (version "1.0.8")
1962     (source
1963      (origin
1964        (method url-fetch)
1965        (uri (pypi-uri "Keras_Applications" version))
1966        (sha256
1967         (base32
1968          "1rcz31ca4axa6kzhjx4lwqxbg4wvlljkj8qj9a7p9sfd5fhzjyam"))))
1969     (build-system python-build-system)
1970     ;; The tests require Keras, but this package is needed to build Keras.
1971     (arguments '(#:tests? #f))
1972     (propagated-inputs
1973      `(("python-h5py" ,python-h5py)
1974        ("python-numpy" ,python-numpy)))
1975     (native-inputs
1976      `(("python-pytest" ,python-pytest)
1977        ("python-pytest-cov" ,python-pytest-cov)
1978        ("python-pytest-pep8" ,python-pytest-pep8)
1979        ("python-pytest-xdist" ,python-pytest-xdist)))
1980     (home-page "https://github.com/keras-team/keras-applications")
1981     (synopsis "Reference implementations of popular deep learning models")
1982     (description
1983      "This package provides reference implementations of popular deep learning
1984 models for use with the Keras deep learning framework.")
1985     (license license:expat)))
1986
1987 (define-public python-keras-preprocessing
1988   (package
1989     (name "python-keras-preprocessing")
1990     (version "1.1.0")
1991     (source
1992      (origin
1993        (method url-fetch)
1994        (uri (pypi-uri "Keras_Preprocessing" version))
1995        (sha256
1996         (base32
1997          "1r98nm4k1svsqjyaqkfk23i31bl1kcfcyp7094yyj3c43phfp3as"))))
1998     (build-system python-build-system)
1999     (propagated-inputs
2000      `(("python-numpy" ,python-numpy)
2001        ("python-six" ,python-six)))
2002     (native-inputs
2003      `(("python-pandas" ,python-pandas)
2004        ("python-pillow" ,python-pillow)
2005        ("python-pytest" ,python-pytest)
2006        ("python-pytest-cov" ,python-pytest-cov)
2007        ("python-pytest-xdist" ,python-pytest-xdist)
2008        ("tensorflow" ,tensorflow)))
2009     (home-page "https://github.com/keras-team/keras-preprocessing/")
2010     (synopsis "Data preprocessing and augmentation for deep learning models")
2011     (description
2012      "Keras Preprocessing is the data preprocessing and data augmentation
2013 module of the Keras deep learning library.  It provides utilities for working
2014 with image data, text data, and sequence data.")
2015     (license license:expat)))
2016
2017 (define-public python-keras
2018   (package
2019     (name "python-keras")
2020     (version "2.2.4")
2021     (source
2022      (origin
2023        (method url-fetch)
2024        (uri (pypi-uri "Keras" version))
2025        (patches (search-patches "python-keras-integration-test.patch"))
2026        (sha256
2027         (base32
2028          "1j8bsqzh49vjdxy6l1k4iwax5vpjzniynyd041xjavdzvfii1dlh"))))
2029     (build-system python-build-system)
2030     (arguments
2031      `(#:phases
2032        (modify-phases %standard-phases
2033          (add-after 'unpack 'remove-tests-for-unavailable-features
2034            (lambda _
2035              (delete-file "keras/backend/theano_backend.py")
2036              (delete-file "keras/backend/cntk_backend.py")
2037              (delete-file "tests/keras/backend/backend_test.py")
2038
2039              ;; FIXME: This doesn't work because Tensorflow is missing the
2040              ;; coder ops library.
2041              (delete-file "tests/keras/test_callbacks.py")
2042              #t))
2043          (replace 'check
2044            (lambda _
2045              ;; These tests attempt to download data files from the internet.
2046              (delete-file "tests/integration_tests/test_datasets.py")
2047              (delete-file "tests/integration_tests/imagenet_utils_test.py")
2048              ;; Backport https://github.com/keras-team/keras/pull/12479.
2049              (substitute* "tests/keras/engine/test_topology.py"
2050                (("np.ones\\(\\(3, 2\\)\\)")
2051                 "1."))
2052              (invoke "python" "-m" "pytest"
2053                      ;; The following test fail only in the build container;
2054                      ;; skip it.
2055                      "-k" "not test_selu"))))))
2056     (propagated-inputs
2057      `(("python-h5py" ,python-h5py)
2058        ("python-keras-applications" ,python-keras-applications)
2059        ("python-keras-preprocessing" ,python-keras-preprocessing)
2060        ("python-numpy" ,python-numpy)
2061        ("python-pydot" ,python-pydot)
2062        ("python-pyyaml" ,python-pyyaml)
2063        ("python-scipy" ,python-scipy)
2064        ("python-six" ,python-six)
2065        ("tensorflow" ,tensorflow)
2066        ("graphviz" ,graphviz)))
2067     (native-inputs
2068      `(("python-pandas" ,python-pandas)
2069        ("python-pytest" ,python-pytest)
2070        ("python-pytest-cov" ,python-pytest-cov)
2071        ("python-pytest-pep8" ,python-pytest-pep8)
2072        ("python-pytest-timeout" ,python-pytest-timeout)
2073        ("python-pytest-xdist" ,python-pytest-xdist)
2074        ("python-sphinx" ,python-sphinx)
2075        ("python-requests" ,python-requests)))
2076     (home-page "https://github.com/keras-team/keras")
2077     (synopsis "High-level deep learning framework")
2078     (description "Keras is a high-level neural networks API, written in Python
2079 and capable of running on top of TensorFlow.  It was developed with a focus on
2080 enabling fast experimentation.  Use Keras if you need a deep learning library
2081 that:
2082
2083 @itemize
2084 @item Allows for easy and fast prototyping (through user friendliness,
2085   modularity, and extensibility).
2086 @item Supports both convolutional networks and recurrent networks, as well as
2087   combinations of the two.
2088 @item Runs seamlessly on CPU and GPU.
2089 @end itemize\n")
2090     (license license:expat)))
2091
2092 (define-public gloo
2093   (let ((version "0.0.0") ; no proper version tag
2094         (commit "ca528e32fea9ca8f2b16053cff17160290fc84ce")
2095         (revision "0"))
2096     (package
2097       (name "gloo")
2098       (version (git-version version revision commit))
2099       (source
2100        (origin
2101          (method git-fetch)
2102          (uri (git-reference
2103                (url "https://github.com/facebookincubator/gloo")
2104                (commit commit)))
2105          (file-name (git-file-name name version))
2106          (sha256
2107           (base32
2108            "1q9f80zy75f6njrzrqkmhc0g3qxs4gskr7ns2jdqanxa2ww7a99w"))))
2109       (build-system cmake-build-system)
2110       (native-inputs
2111        `(("googletest" ,googletest)))
2112       (arguments
2113        `(#:configure-flags '("-DBUILD_TEST=1")
2114          #:phases
2115          (modify-phases %standard-phases
2116            (replace 'check
2117              (lambda _
2118                (invoke "make" "gloo_test")
2119                #t)))))
2120       (synopsis "Collective communications library")
2121       (description
2122        "Gloo is a collective communications library.  It comes with a
2123 number of collective algorithms useful for machine learning applications.
2124 These include a barrier, broadcast, and allreduce.")
2125       (home-page "https://github.com/facebookincubator/gloo")
2126       (license license:bsd-3))))
2127
2128 (define-public python-umap-learn
2129   (package
2130     (name "python-umap-learn")
2131     (version "0.3.10")
2132     (source
2133      (origin
2134        (method url-fetch)
2135        (uri (pypi-uri "umap-learn" version))
2136        (sha256
2137         (base32
2138          "02ada2yy6km6zgk2836kg1c97yrcpalvan34p8c57446finnpki1"))))
2139     (build-system python-build-system)
2140     (native-inputs
2141      `(("python-joblib" ,python-joblib)
2142        ("python-nose" ,python-nose)))
2143     (propagated-inputs
2144      `(("python-numba" ,python-numba)
2145        ("python-numpy" ,python-numpy)
2146        ("python-scikit-learn" ,python-scikit-learn)
2147        ("python-scipy" ,python-scipy)))
2148     (home-page "https://github.com/lmcinnes/umap")
2149     (synopsis
2150      "Uniform Manifold Approximation and Projection")
2151     (description
2152      "Uniform Manifold Approximation and Projection is a dimension reduction
2153 technique that can be used for visualisation similarly to t-SNE, but also for
2154 general non-linear dimension reduction.")
2155     (license license:bsd-3)))