gnu/packages/machine-learning.scm

   1 ;;; GNU Guix --- Functional package management for GNU
   2 ;;; Copyright © 2015, 2016, 2017, 2018, 2019 Ricardo Wurmus <rekado@elephly.net>
   3 ;;; Copyright © 2016 Efraim Flashner <efraim@flashner.co.il>
   4 ;;; Copyright © 2016, 2017 Marius Bakke <mbakke@fastmail.com>
   5 ;;; Copyright © 2016 Hartmut Goebel <h.goebel@crazy-compilers.com>
   6 ;;; Copyright © 2018 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 ;;;
  14 ;;; This file is part of GNU Guix.
  15 ;;;
  16 ;;; GNU Guix is free software; you can redistribute it and/or modify it
  17 ;;; under the terms of the GNU General Public License as published by
  18 ;;; the Free Software Foundation; either version 3 of the License, or (at
  19 ;;; your option) any later version.
  20 ;;;
  21 ;;; GNU Guix is distributed in the hope that it will be useful, but
  22 ;;; WITHOUT ANY WARRANTY; without even the implied warranty of
  23 ;;; MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
  24 ;;; GNU General Public License for more details.
  25 ;;;
  26 ;;; You should have received a copy of the GNU General Public License
  27 ;;; along with GNU Guix.  If not, see <http://www.gnu.org/licenses/>.
  28
  29 (define-module (gnu packages machine-learning)
  30   #:use-module ((guix licenses) #:prefix license:)
  31   #:use-module (guix packages)
  32   #:use-module (guix utils)
  33   #:use-module (guix download)
  34   #:use-module (guix svn-download)
  35   #:use-module (guix build-system cmake)
  36   #:use-module (guix build-system gnu)
  37   #:use-module (guix build-system ocaml)
  38   #:use-module (guix build-system python)
  39   #:use-module (guix build-system r)
  40   #:use-module (guix git-download)
  41   #:use-module (gnu packages)
  42   #:use-module (gnu packages algebra)
  43   #:use-module (gnu packages audio)
  44   #:use-module (gnu packages autotools)
  45   #:use-module (gnu packages base)
  46   #:use-module (gnu packages boost)
  47   #:use-module (gnu packages check)
  48   #:use-module (gnu packages compression)
  49   #:use-module (gnu packages cran)
  50   #:use-module (gnu packages dejagnu)
  51   #:use-module (gnu packages gcc)
  52   #:use-module (gnu packages glib)
  53   #:use-module (gnu packages gstreamer)
  54   #:use-module (gnu packages image)
  55   #:use-module (gnu packages linux)
  56   #:use-module (gnu packages maths)
  57   #:use-module (gnu packages mpi)
  58   #:use-module (gnu packages ocaml)
  59   #:use-module (gnu packages onc-rpc)
  60   #:use-module (gnu packages perl)
  61   #:use-module (gnu packages pkg-config)
  62   #:use-module (gnu packages python)
  63   #:use-module (gnu packages python-xyz)
  64   #:use-module (gnu packages statistics)
  65   #:use-module (gnu packages swig)
  66   #:use-module (gnu packages xml)
  67   #:use-module (gnu packages xorg))
  68
  69 (define-public fann
  70   ;; The last release is >100 commits behind, so we package from git.
  71   (let ((commit "d71d54788bee56ba4cf7522801270152da5209d7"))
  72     (package
  73       (name "fann")
  74       (version (string-append "2.2.0-1." (string-take commit 8)))
  75       (source (origin
  76                 (method git-fetch)
  77                 (uri (git-reference
  78                       (url "https://github.com/libfann/fann.git")
  79                       (commit commit)))
  80                 (file-name (string-append name "-" version "-checkout"))
  81                 (sha256
  82                  (base32
  83                   "0ibwpfrjs6q2lijs8slxjgzb2llcl6rk3v2ski4r6215g5jjhg3x"))))
  84       (build-system cmake-build-system)
  85       (arguments
  86        `(#:phases
  87          (modify-phases %standard-phases
  88            (replace 'check
  89              (lambda* (#:key outputs #:allow-other-keys)
  90                (let* ((out (assoc-ref outputs "out")))
  91                  (with-directory-excursion (string-append (getcwd) "/tests")
  92                    (invoke "./fann_tests"))))))))
  93       (home-page "http://leenissen.dk/fann/wp/")
  94       (synopsis "Fast Artificial Neural Network")
  95       (description
  96        "FANN is a free open source neural network library, which implements
  97 multilayer artificial neural networks in C with support for both fully
  98 connected and sparsely connected networks.")
  99       (license license:lgpl2.1))))
 100
 101 (define-public libsvm
 102   (package
 103     (name "libsvm")
 104     (version "3.22")
 105     (source
 106      (origin
 107        (method url-fetch)
 108        (uri (string-append "https://www.csie.ntu.edu.tw/~cjlin/libsvm/"
 109                            name "-" version ".tar.gz"))
 110        (sha256
 111         (base32
 112          "0zd7s19y5vb7agczl6456bn45cj1y64739sslaskw1qk7dywd0bd"))))
 113     (build-system gnu-build-system)
 114     (arguments
 115      `(#:tests? #f ;no "check" target
 116        #:phases (modify-phases %standard-phases
 117                   (delete 'configure)
 118                   (replace
 119                    'install             ; no ‘install’ target
 120                    (lambda* (#:key outputs #:allow-other-keys)
 121                      (let* ((out (assoc-ref outputs "out"))
 122                             (bin (string-append out "/bin/")))
 123                        (mkdir-p bin)
 124                        (for-each (lambda (file)
 125                                    (copy-file file (string-append bin file)))
 126                                  '("svm-train"
 127                                    "svm-predict"
 128                                    "svm-scale")))
 129                      #t)))))
 130     (home-page "http://www.csie.ntu.edu.tw/~cjlin/libsvm/")
 131     (synopsis "Library for Support Vector Machines")
 132     (description
 133      "LIBSVM is a machine learning library for support vector
 134 classification, (C-SVC, nu-SVC), regression (epsilon-SVR, nu-SVR) and
 135 distribution estimation (one-class SVM).  It supports multi-class
 136 classification.")
 137     (license license:bsd-3)))
 138
 139 (define-public python-libsvm
 140   (package (inherit libsvm)
 141     (name "python-libsvm")
 142     (build-system gnu-build-system)
 143     (arguments
 144      `(#:tests? #f ;no "check" target
 145        #:make-flags '("-C" "python")
 146        #:phases
 147        (modify-phases %standard-phases
 148          (delete 'configure)
 149          (replace
 150           'install                      ; no ‘install’ target
 151           (lambda* (#:key inputs outputs #:allow-other-keys)
 152             (let ((site (string-append (assoc-ref outputs "out")
 153                                        "/lib/python"
 154                                        (string-take
 155                                         (string-take-right
 156                                          (assoc-ref inputs "python") 5) 3)
 157                                        "/site-packages/")))
 158               (substitute* "python/svm.py"
 159                 (("../libsvm.so.2") "libsvm.so.2"))
 160               (mkdir-p site)
 161               (for-each (lambda (file)
 162                           (copy-file file (string-append site (basename file))))
 163                         (find-files "python" "\\.py"))
 164               (copy-file "libsvm.so.2"
 165                          (string-append site "libsvm.so.2")))
 166             #t)))))
 167     (inputs
 168      `(("python" ,python)))
 169     (synopsis "Python bindings of libSVM")))
 170
 171 (define-public ghmm
 172   ;; The latest release candidate is several years and a couple of fixes have
 173   ;; been published since.  This is why we download the sources from the SVN
 174   ;; repository.
 175   (let ((svn-revision 2341))
 176     (package
 177       (name "ghmm")
 178       (version (string-append "0.9-rc3-0." (number->string svn-revision)))
 179       (source (origin
 180                 (method svn-fetch)
 181                 (uri (svn-reference
 182                       (url "http://svn.code.sf.net/p/ghmm/code/trunk")
 183                       (revision svn-revision)))
 184                 (file-name (string-append name "-" version))
 185                 (sha256
 186                  (base32
 187                   "0qbq1rqp94l530f043qzp8aw5lj7dng9wq0miffd7spd1ff638wq"))))
 188       (build-system gnu-build-system)
 189       (arguments
 190        `(#:imported-modules (,@%gnu-build-system-modules
 191                              (guix build python-build-system))
 192          #:phases
 193          (modify-phases %standard-phases
 194            (add-after 'unpack 'enter-dir
 195              (lambda _ (chdir "ghmm") #t))
 196            (delete 'check)
 197            (add-after 'install 'check
 198              (assoc-ref %standard-phases 'check))
 199            (add-before 'check 'fix-PYTHONPATH
 200              (lambda* (#:key inputs outputs #:allow-other-keys)
 201                (let ((python-version ((@@ (guix build python-build-system)
 202                                            get-python-version)
 203                                       (assoc-ref inputs "python"))))
 204                  (setenv "PYTHONPATH"
 205                          (string-append (getenv "PYTHONPATH")
 206                                         ":" (assoc-ref outputs "out")
 207                                         "/lib/python" python-version
 208                                         "/site-packages")))
 209                #t))
 210            (add-after 'enter-dir 'fix-runpath
 211              (lambda* (#:key outputs #:allow-other-keys)
 212                (substitute* "ghmmwrapper/setup.py"
 213                  (("^(.*)extra_compile_args = \\[" line indent)
 214                   (string-append indent
 215                                  "extra_link_args = [\"-Wl,-rpath="
 216                                  (assoc-ref outputs "out") "/lib\"],\n"
 217                                  line
 218                                  "\"-Wl,-rpath="
 219                                  (assoc-ref outputs "out")
 220                                  "/lib\", ")))
 221                #t))
 222            (add-after 'enter-dir 'disable-broken-tests
 223              (lambda _
 224                (substitute* "tests/Makefile.am"
 225                  ;; GHMM_SILENT_TESTS is assumed to be a command.
 226                  (("TESTS_ENVIRONMENT.*") "")
 227                  ;; Do not build broken tests.
 228                  (("chmm .*") "")
 229                  (("read_fa .*") "")
 230                  (("mcmc .*") "")
 231                  (("label_higher_order_test.*$")
 232                   "label_higher_order_test\n"))
 233
 234                ;; These Python unittests are broken as there is no gato.
 235                ;; See https://sourceforge.net/p/ghmm/support-requests/3/
 236                (substitute* "ghmmwrapper/ghmmunittests.py"
 237                  (("^(.*)def (testNewXML|testMultipleTransitionClasses|testNewXML)"
 238                    line indent)
 239                   (string-append indent
 240                                  "@unittest.skip(\"Disabled by Guix\")\n"
 241                                  line)))
 242                #t))
 243            (add-after 'disable-broken-tests 'autogen
 244              (lambda _
 245                (invoke "bash" "autogen.sh"))))))
 246       (inputs
 247        `(("python" ,python-2) ; only Python 2 is supported
 248          ("libxml2" ,libxml2)))
 249       (native-inputs
 250        `(("pkg-config" ,pkg-config)
 251          ("dejagnu" ,dejagnu)
 252          ("swig" ,swig)
 253          ("autoconf" ,autoconf)
 254          ("automake" ,automake)
 255          ("libtool" ,libtool)))
 256       (home-page "http://ghmm.org")
 257       (synopsis "Hidden Markov Model library")
 258       (description
 259        "The General Hidden Markov Model library (GHMM) is a C library with
 260 additional Python bindings implementing a wide range of types of @dfn{Hidden
 261 Markov Models} (HMM) and algorithms: discrete, continuous emissions, basic
 262 training, HMM clustering, HMM mixtures.")
 263       (license license:lgpl2.0+))))
 264
 265 (define-public mcl
 266   (package
 267     (name "mcl")
 268     (version "14.137")
 269     (source (origin
 270               (method url-fetch)
 271               (uri (string-append
 272                     "http://micans.org/mcl/src/mcl-"
 273                     (string-replace-substring version "." "-")
 274                     ".tar.gz"))
 275               (sha256
 276                (base32
 277                 "15xlax3z31lsn62vlg94hkm75nm40q4679amnfg13jm8m2bnhy5m"))))
 278     (build-system gnu-build-system)
 279     (arguments
 280      `(#:configure-flags (list "--enable-blast")))
 281     (inputs
 282      `(("perl" ,perl)))
 283     (home-page "http://micans.org/mcl/")
 284     (synopsis "Clustering algorithm for graphs")
 285     (description
 286      "The MCL algorithm is short for the @dfn{Markov Cluster Algorithm}, a
 287 fast and scalable unsupervised cluster algorithm for graphs (also known as
 288 networks) based on simulation of (stochastic) flow in graphs.")
 289     ;; In the LICENCE file and web page it says "The software is licensed
 290     ;; under the GNU General Public License, version 3.", but in several of
 291     ;; the source code files it suggests GPL3 or later.
 292     ;; http://listserver.ebi.ac.uk/pipermail/mcl-users/2016/000376.html
 293     (license license:gpl3)))
 294
 295 (define-public ocaml-mcl
 296   (package
 297     (name "ocaml-mcl")
 298     (version "12-068oasis4")
 299     (source
 300      (origin
 301        (method url-fetch)
 302        (uri (string-append
 303              "https://github.com/fhcrc/mcl/archive/"
 304              version ".tar.gz"))
 305        (file-name (string-append name "-" version ".tar.gz"))
 306        (sha256
 307         (base32
 308          "1l5jbhwjpsj38x8b9698hfpkv75h8hn3kj0gihjhn8ym2cwwv110"))))
 309     (build-system ocaml-build-system)
 310     (arguments
 311      `(#:ocaml ,ocaml-4.02
 312        #:findlib ,ocaml4.02-findlib
 313        #:phases
 314        (modify-phases %standard-phases
 315          (add-before 'configure 'patch-paths
 316            (lambda _
 317              (substitute* "configure"
 318                (("SHELL = /bin/sh") (string-append "SHELL = "(which "sh"))))
 319              (substitute* "setup.ml"
 320                (("LDFLAGS=-fPIC")
 321                 (string-append "LDFLAGS=-fPIC\"; \"SHELL=" (which "sh"))))
 322              #t)))))
 323     (home-page "https://github.com/fhcrc/mcl")
 324     (synopsis "OCaml wrappers around MCL")
 325     (description
 326      "This package provides OCaml bindings for the MCL graph clustering
 327 algorithm.")
 328     (license license:gpl3)))
 329
 330 (define-public randomjungle
 331   (package
 332     (name "randomjungle")
 333     (version "2.1.0")
 334     (source
 335      (origin
 336        (method url-fetch)
 337        (uri (string-append
 338              "https://www.imbs.uni-luebeck.de/fileadmin/files/Software"
 339              "/randomjungle/randomjungle-" version ".tar_.gz"))
 340        (patches (search-patches "randomjungle-disable-static-build.patch"))
 341        (sha256
 342         (base32
 343          "12c8rf30cla71swx2mf4ww9mfd8jbdw5lnxd7dxhyw1ygrvg6y4w"))))
 344     (build-system gnu-build-system)
 345     (arguments
 346      `(#:configure-flags
 347        (list "--disable-static"
 348              (string-append "--with-boost="
 349                             (assoc-ref %build-inputs "boost")))
 350        #:phases
 351        (modify-phases %standard-phases
 352          (add-before
 353           'configure 'set-CXXFLAGS
 354           (lambda _
 355             (setenv "CXXFLAGS" "-fpermissive ")
 356             #t)))))
 357     (inputs
 358      `(("boost" ,boost)
 359        ("gsl" ,gsl)
 360        ("libxml2" ,libxml2)
 361        ("zlib" ,zlib)))
 362     (native-inputs
 363      `(("gfortran" ,gfortran)
 364        ("gfortran:lib" ,gfortran "lib")))
 365     ;; Non-portable assembly instructions are used so building fails on
 366     ;; platforms other than x86_64 or i686.
 367     (supported-systems '("x86_64-linux" "i686-linux"))
 368     (home-page "https://www.imbs.uni-luebeck.de/forschung/software/details.html#c224")
 369     (synopsis "Implementation of the Random Forests machine learning method")
 370     (description
 371      "Random Jungle is an implementation of Random Forests.  It is supposed to
 372 analyse high dimensional data.  In genetics, it can be used for analysing big
 373 Genome Wide Association (GWA) data.  Random Forests is a powerful machine
 374 learning method.  Most interesting features are variable selection, missing
 375 value imputation, classifier creation, generalization error estimation and
 376 sample proximities between pairs of cases.")
 377     (license license:gpl3+)))
 378
 379 (define-public openfst
 380   (package
 381     (name "openfst")
 382     (version "1.7.1")
 383     (source (origin
 384               (method url-fetch)
 385               (uri (string-append "http://www.openfst.org/twiki/pub/FST/"
 386                                   "FstDownload/openfst-" version ".tar.gz"))
 387               (sha256
 388                (base32
 389                 "0x9wfcqd8hq4h349s7j77sr60h8xjdfshqw1m3a2n6z5bdr9qkm1"))))
 390     (build-system gnu-build-system)
 391     (home-page "http://www.openfst.org")
 392     (synopsis "Library for weighted finite-state transducers")
 393     (description "OpenFst is a library for constructing, combining,
 394 optimizing, and searching weighted finite-state transducers (FSTs).")
 395     (license license:asl2.0)))
 396
 397 (define-public shogun
 398   (package
 399     (name "shogun")
 400     (version "6.1.3")
 401     (source
 402      (origin
 403        (method url-fetch)
 404        (uri (string-append
 405              "ftp://shogun-toolbox.org/shogun/releases/"
 406              (version-major+minor version)
 407              "/sources/shogun-" version ".tar.bz2"))
 408        (sha256
 409         (base32
 410          "1rn9skm3nw6hr7mr3lgp2gfqhi7ii0lyxck7qmqnf8avq349s5jp"))
 411        (modules '((guix build utils)
 412                   (ice-9 rdelim)))
 413        (snippet
 414         '(begin
 415            ;; Remove non-free sources and files referencing them
 416            (for-each delete-file
 417                      (find-files "src/shogun/classifier/svm/"
 418                                  "SVMLight\\.(cpp|h)"))
 419            (for-each delete-file
 420                      (find-files "examples/undocumented/libshogun/"
 421                                  (string-append
 422                                   "(classifier_.*svmlight.*|"
 423                                   "evaluation_cross_validation_locked_comparison).cpp")))
 424            ;; Remove non-free functions.
 425            (define (delete-ifdefs file)
 426              (with-atomic-file-replacement file
 427                (lambda (in out)
 428                  (let loop ((line (read-line in 'concat))
 429                             (skipping? #f))
 430                    (if (eof-object? line)
 431                        #t
 432                        (let ((skip-next?
 433                               (or (and skipping?
 434                                        (not (string-prefix?
 435                                              "#endif //USE_SVMLIGHT" line)))
 436                                   (string-prefix?
 437                                    "#ifdef USE_SVMLIGHT" line))))
 438                          (when (or (not skipping?)
 439                                    (and skipping? (not skip-next?)))
 440                            (display line out))
 441                          (loop (read-line in 'concat) skip-next?)))))))
 442            (for-each delete-ifdefs
 443                      (append
 444                       (find-files "src/shogun/classifier/mkl"
 445                                   "^MKLClassification\\.cpp")
 446                       (find-files "src/shogun/classifier/svm"
 447                                   "^SVMLightOneClass\\.(cpp|h)")
 448                       (find-files "src/shogun/multiclass"
 449                                   "^ScatterSVM\\.(cpp|h)")
 450                       (find-files "src/shogun/kernel/"
 451                                   "^(Kernel|CombinedKernel|ProductKernel)\\.(cpp|h)")
 452                       (find-files "src/shogun/regression/svr"
 453                                   "^(MKLRegression|SVRLight)\\.(cpp|h)")
 454                       (find-files "src/shogun/transfer/domain_adaptation"
 455                                   "^DomainAdaptationSVM\\.(cpp|h)")))
 456            #t))))
 457     (build-system cmake-build-system)
 458     (arguments
 459      '(#:tests? #f ;no check target
 460        #:phases
 461        (modify-phases %standard-phases
 462          (add-after 'unpack 'delete-broken-symlinks
 463            (lambda _
 464              (for-each delete-file '("applications/arts/data"
 465                                      "applications/asp/data"
 466                                      "applications/easysvm/data"
 467                                      "applications/msplicer/data"
 468                                      "applications/ocr/data"
 469                                      "examples/meta/data"
 470                                      "examples/undocumented/data"))
 471              #t))
 472          (add-after 'unpack 'change-R-target-path
 473            (lambda* (#:key outputs #:allow-other-keys)
 474              (substitute* '("src/interfaces/r/CMakeLists.txt"
 475                             "examples/meta/r/CMakeLists.txt")
 476                (("\\$\\{R_COMPONENT_LIB_PATH\\}")
 477                 (string-append (assoc-ref outputs "out")
 478                                "/lib/R/library/")))
 479              #t))
 480          (add-after 'unpack 'fix-octave-modules
 481            (lambda* (#:key outputs #:allow-other-keys)
 482              (substitute* "src/interfaces/octave/CMakeLists.txt"
 483                (("^include_directories\\(\\$\\{OCTAVE_INCLUDE_DIRS\\}")
 484                 "include_directories(${OCTAVE_INCLUDE_DIRS} ${OCTAVE_INCLUDE_DIRS}/octave")
 485                ;; change target directory
 486                (("\\$\\{OCTAVE_OCT_LOCAL_API_FILE_DIR\\}")
 487                 (string-append (assoc-ref outputs "out")
 488                                "/share/octave/packages")))
 489              (substitute* '("src/interfaces/octave/swig_typemaps.i"
 490                             "src/interfaces/octave/sg_print_functions.cpp")
 491                ;; "octave/config.h" and "octave/oct-obj.h" deprecated in Octave.
 492                (("octave/config\\.h") "octave/octave-config.h")
 493                (("octave/oct-obj.h") "octave/ovl.h"))
 494              #t))
 495          (add-after 'unpack 'move-rxcpp
 496            (lambda* (#:key inputs #:allow-other-keys)
 497              (let ((rxcpp-dir "shogun/third-party/rxcpp"))
 498                (mkdir-p rxcpp-dir)
 499                (install-file (assoc-ref inputs "rxcpp") rxcpp-dir)
 500                #t)))
 501          (add-before 'build 'set-HOME
 502            ;; $HOME needs to be set at some point during the build phase
 503            (lambda _ (setenv "HOME" "/tmp") #t)))
 504        #:configure-flags
 505        (list "-DCMAKE_BUILD_WITH_INSTALL_RPATH=TRUE"
 506              "-DUSE_SVMLIGHT=OFF" ;disable proprietary SVMLIGHT
 507              "-DBUILD_META_EXAMPLES=OFF" ;requires unpackaged ctags
 508              ;;"-DINTERFACE_JAVA=ON" ;requires unpackaged jblas
 509              ;;"-DINTERFACE_RUBY=ON" ;requires unpackaged ruby-narray
 510              ;;"-DINTERFACE_PERL=ON" ;"FindPerlLibs" does not exist
 511              ;;"-DINTERFACE_LUA=ON"  ;fails because lua doesn't build pkgconfig file
 512              "-DINTERFACE_OCTAVE=ON"
 513              "-DINTERFACE_PYTHON=ON"
 514              "-DINTERFACE_R=ON")))
 515     (inputs
 516      `(("python" ,python)
 517        ("numpy" ,python-numpy)
 518        ("r-minimal" ,r-minimal)
 519        ("octave" ,octave-cli)
 520        ("swig" ,swig)
 521        ("eigen" ,eigen)
 522        ("hdf5" ,hdf5)
 523        ("atlas" ,atlas)
 524        ("arpack" ,arpack-ng)
 525        ("lapack" ,lapack)
 526        ("glpk" ,glpk)
 527        ("libxml2" ,libxml2)
 528        ("lzo" ,lzo)
 529        ("zlib" ,zlib)))
 530     (native-inputs
 531      `(("pkg-config" ,pkg-config)
 532        ("rxcpp" ,rxcpp)))
 533     ;; Non-portable SSE instructions are used so building fails on platforms
 534     ;; other than x86_64.
 535     (supported-systems '("x86_64-linux"))
 536     (home-page "http://shogun-toolbox.org/")
 537     (synopsis "Machine learning toolbox")
 538     (description
 539      "The Shogun Machine learning toolbox provides a wide range of unified and
 540 efficient Machine Learning (ML) methods.  The toolbox seamlessly allows to
 541 combine multiple data representations, algorithm classes, and general purpose
 542 tools.  This enables both rapid prototyping of data pipelines and extensibility
 543 in terms of new algorithms.")
 544     (license license:gpl3+)))
 545
 546 (define-public rxcpp
 547   (package
 548     (name "rxcpp")
 549     (version "4.0.0")
 550     (source
 551      (origin
 552        (method url-fetch)
 553        (uri (string-append "https://github.com/ReactiveX/RxCpp/archive/v"
 554                            version ".tar.gz"))
 555        (sha256
 556         (base32
 557          "0y2isr8dy2n1yjr9c5570kpc9lvdlch6jv0jvw000amwn5d3krsh"))
 558        (file-name (string-append name "-" version ".tar.gz"))))
 559     (build-system cmake-build-system)
 560     (arguments
 561      `(#:phases
 562        (modify-phases %standard-phases
 563          (add-after 'unpack 'remove-werror
 564            (lambda _
 565              (substitute* (find-files ".")
 566                (("-Werror") ""))
 567              #t))
 568          (replace 'check
 569            (lambda _
 570              (invoke "ctest"))))))
 571     (native-inputs
 572      `(("catch" ,catch-framework)))
 573     (home-page "http://reactivex.io/")
 574     (synopsis "Reactive Extensions for C++")
 575     (description
 576      "The Reactive Extensions for C++ (RxCpp) is a library of algorithms for
 577 values-distributed-in-time.  ReactiveX is a library for composing asynchronous
 578 and event-based programs by using observable sequences.
 579
 580 It extends the observer pattern to support sequences of data and/or events and
 581 adds operators that allow you to compose sequences together declaratively while
 582 abstracting away concerns about things like low-level threading,
 583 synchronization, thread-safety, concurrent data structures, and non-blocking
 584 I/O.")
 585     (license license:asl2.0)))
 586
 587 (define-public r-adaptivesparsity
 588   (package
 589     (name "r-adaptivesparsity")
 590     (version "1.6")
 591     (source (origin
 592               (method url-fetch)
 593               (uri (cran-uri "AdaptiveSparsity" version))
 594               (sha256
 595                (base32
 596                 "0imr5m8mll9j6n4icsv6z9rl5kbnwsp9wvzrg7n90nnmcxq2cz91"))))
 597     (properties
 598      `((upstream-name . "AdaptiveSparsity")))
 599     (build-system r-build-system)
 600     (arguments
 601      `(#:phases
 602        (modify-phases %standard-phases
 603          (add-after 'unpack 'link-against-armadillo
 604            (lambda _
 605              (substitute* "src/Makevars"
 606                (("PKG_LIBS=" prefix)
 607                 (string-append prefix "-larmadillo"))))))))
 608     (propagated-inputs
 609      `(("r-mass" ,r-mass)
 610        ("r-matrix" ,r-matrix)
 611        ("r-rcpp" ,r-rcpp)
 612        ("r-rcpparmadillo" ,r-rcpparmadillo)))
 613     (inputs
 614      `(("armadillo" ,armadillo)))
 615     (home-page "https://cran.r-project.org/web/packages/AdaptiveSparsity")
 616     (synopsis "Adaptive sparsity models")
 617     (description
 618      "This package implements the Figueiredo machine learning algorithm for
 619 adaptive sparsity and the Wong algorithm for adaptively sparse gaussian
 620 geometric models.")
 621     (license license:lgpl3+)))
 622
 623 (define-public r-kernlab
 624   (package
 625     (name "r-kernlab")
 626     (version "0.9-27")
 627     (source
 628      (origin
 629        (method url-fetch)
 630        (uri (cran-uri "kernlab" version))
 631        (sha256
 632         (base32
 633          "1m0xqf6gyvwayz7w3c83y32ayvnlz0jicj8ijk808zq9sh7dbbgn"))))
 634     (build-system r-build-system)
 635     (home-page "https://cran.r-project.org/web/packages/kernlab")
 636     (synopsis "Kernel-based machine learning tools")
 637     (description
 638      "This package provides kernel-based machine learning methods for
 639 classification, regression, clustering, novelty detection, quantile regression
 640 and dimensionality reduction.  Among other methods @code{kernlab} includes
 641 Support Vector Machines, Spectral Clustering, Kernel PCA, Gaussian Processes
 642 and a QP solver.")
 643     (license license:gpl2)))
 644
 645 (define-public dlib
 646   (package
 647     (name "dlib")
 648     (version "19.7")
 649     (source (origin
 650               (method url-fetch)
 651               (uri (string-append
 652                     "http://dlib.net/files/dlib-" version ".tar.bz2"))
 653               (sha256
 654                (base32
 655                 "1mljz02kwkrbggyncxv5fpnyjdybw2qihaacb3js8yfkw12vwpc2"))
 656               (modules '((guix build utils)))
 657               (snippet
 658                '(begin
 659                   ;; Delete ~13MB of bundled dependencies.
 660                   (delete-file-recursively "dlib/external")
 661                   (delete-file-recursively "docs/dlib/external")
 662                   #t))))
 663     (build-system cmake-build-system)
 664     (arguments
 665      `(#:phases
 666        (modify-phases %standard-phases
 667          (add-after 'unpack 'disable-asserts
 668            (lambda _
 669              ;; config.h recommends explicitly enabling or disabling asserts
 670              ;; when building as a shared library. By default neither is set.
 671              (substitute* "dlib/config.h"
 672                (("^//#define DLIB_DISABLE_ASSERTS") "#define DLIB_DISABLE_ASSERTS"))
 673              #t))
 674          (add-after 'disable-asserts 'disable-failing-tests
 675            (lambda _
 676              ;; One test times out on MIPS, so we need to disable it.
 677              ;; Others are flaky on some platforms.
 678              (let* ((system ,(or (%current-target-system)
 679                                  (%current-system)))
 680                     (disabled-tests (cond
 681                                      ((string-prefix? "mips64" system)
 682                                       '("object_detector" ; timeout
 683                                         "data_io"))
 684                                      ((string-prefix? "armhf" system)
 685                                       '("learning_to_track"))
 686                                      ((string-prefix? "i686" system)
 687                                       '("optimization"))
 688                                      (else '()))))
 689                (for-each
 690                 (lambda (test)
 691                   (substitute* "dlib/test/makefile"
 692                     (((string-append "SRC \\+= " test "\\.cpp")) "")))
 693                 disabled-tests)
 694                #t)))
 695          (replace 'check
 696            (lambda _
 697              ;; No test target, so we build and run the unit tests here.
 698              (let ((test-dir (string-append "../dlib-" ,version "/dlib/test")))
 699                (with-directory-excursion test-dir
 700                  (invoke "make" "-j" (number->string (parallel-job-count)))
 701                  (invoke "./dtest" "--runall"))
 702                #t)))
 703          (add-after 'install 'delete-static-library
 704            (lambda* (#:key outputs #:allow-other-keys)
 705              (delete-file (string-append (assoc-ref outputs "out")
 706                                          "/lib/libdlib.a"))
 707              #t)))))
 708     (native-inputs
 709      `(("pkg-config" ,pkg-config)
 710        ;; For tests.
 711        ("libnsl" ,libnsl)))
 712     (inputs
 713      `(("giflib" ,giflib)
 714        ("lapack" ,lapack)
 715        ("libjpeg" ,libjpeg)
 716        ("libpng" ,libpng)
 717        ("libx11" ,libx11)
 718        ("openblas" ,openblas)
 719        ("zlib" ,zlib)))
 720     (synopsis
 721      "Toolkit for making machine learning and data analysis applications in C++")
 722     (description
 723      "Dlib is a modern C++ toolkit containing machine learning algorithms and
 724 tools.  It is used in both industry and academia in a wide range of domains
 725 including robotics, embedded devices, mobile phones, and large high performance
 726 computing environments.")
 727     (home-page "http://dlib.net")
 728     (license license:boost1.0)))
 729
 730 (define-public python-scikit-learn
 731   (package
 732     (name "python-scikit-learn")
 733     (version "0.20.1")
 734     (source
 735      (origin
 736        (method git-fetch)
 737        (uri (git-reference
 738              (url "https://github.com/scikit-learn/scikit-learn.git")
 739              (commit version)))
 740        (file-name (git-file-name name version))
 741        (sha256
 742         (base32
 743          "0fkhwg3xn1s7ln9q1szq6kwc4jhwvjh8w4kmv9wcrqy7cq3lbv0d"))))
 744     (build-system python-build-system)
 745     (arguments
 746      `(#:phases
 747        (modify-phases %standard-phases
 748          (add-after 'build 'build-ext
 749            (lambda _ (invoke "python" "setup.py" "build_ext" "--inplace") #t))
 750          (replace 'check
 751            (lambda _
 752              ;; Restrict OpenBLAS threads to prevent segfaults while testing!
 753              (setenv "OPENBLAS_NUM_THREADS" "1")
 754
 755              ;; Some tests require write access to $HOME.
 756              (setenv "HOME" "/tmp")
 757
 758              (invoke "pytest" "sklearn" "-m" "not network")))
 759          ;; FIXME: This fails with permission denied
 760          (delete 'reset-gzip-timestamps))))
 761     (inputs
 762      `(("openblas" ,openblas)))
 763     (native-inputs
 764      `(("python-pytest" ,python-pytest)
 765        ("python-pandas" ,python-pandas) ;for tests
 766        ("python-cython" ,python-cython)))
 767     (propagated-inputs
 768      `(("python-numpy" ,python-numpy)
 769        ("python-scipy" ,python-scipy)))
 770     (home-page "http://scikit-learn.org/")
 771     (synopsis "Machine Learning in Python")
 772     (description
 773      "Scikit-learn provides simple and efficient tools for data mining and
 774 data analysis.")
 775     (license license:bsd-3)))
 776
 777 (define-public python2-scikit-learn
 778   (package-with-python2 python-scikit-learn))
 779
 780 (define-public python-autograd
 781   (let* ((commit "442205dfefe407beffb33550846434baa90c4de7")
 782          (revision "0")
 783          (version (git-version "0.0.0" revision commit)))
 784     (package
 785       (name "python-autograd")
 786       (home-page "https://github.com/HIPS/autograd")
 787       (source (origin
 788                 (method git-fetch)
 789                 (uri (git-reference
 790                       (url home-page)
 791                       (commit commit)))
 792                 (sha256
 793                  (base32
 794                   "189sv2xb0mwnjawa9z7mrgdglc1miaq93pnck26r28fi1jdwg0z4"))
 795                 (file-name (git-file-name name version))))
 796       (version version)
 797       (build-system python-build-system)
 798       (native-inputs
 799        `(("python-nose" ,python-nose)
 800          ("python-pytest" ,python-pytest)))
 801       (propagated-inputs
 802        `(("python-future" ,python-future)
 803          ("python-numpy" ,python-numpy)))
 804       (arguments
 805        `(#:phases (modify-phases %standard-phases
 806                     (replace 'check
 807                       (lambda _
 808                         (invoke "py.test" "-v"))))))
 809       (synopsis "Efficiently computes derivatives of NumPy code")
 810       (description "Autograd can automatically differentiate native Python and
 811 NumPy code.  It can handle a large subset of Python's features, including loops,
 812 ifs, recursion and closures, and it can even take derivatives of derivatives
 813 of derivatives.  It supports reverse-mode differentiation
 814 (a.k.a. backpropagation), which means it can efficiently take gradients of
 815 scalar-valued functions with respect to array-valued arguments, as well as
 816 forward-mode differentiation, and the two can be composed arbitrarily.  The
 817 main intended application of Autograd is gradient-based optimization.")
 818       (license license:expat))))
 819
 820 (define-public python2-autograd
 821   (package-with-python2 python-autograd))
 822
 823 (define-public lightgbm
 824   (package
 825     (name "lightgbm")
 826     (version "2.0.12")
 827     (source (origin
 828               (method url-fetch)
 829               (uri (string-append
 830                     "https://github.com/Microsoft/LightGBM/archive/v"
 831                     version ".tar.gz"))
 832               (sha256
 833                (base32
 834                 "132zf0yk0545mg72hyzxm102g3hpb6ixx9hnf8zd2k55gas6cjj1"))
 835               (file-name (string-append name "-" version ".tar.gz"))))
 836     (native-inputs
 837      `(("python-pytest" ,python-pytest)
 838        ("python-nose" ,python-nose)))
 839     (inputs
 840      `(("openmpi" ,openmpi)))
 841     (propagated-inputs
 842      `(("python-numpy" ,python-numpy)
 843        ("python-scipy" ,python-scipy)))
 844     (arguments
 845      `(#:configure-flags
 846        '("-DUSE_MPI=ON")
 847        #:phases
 848        (modify-phases %standard-phases
 849          (replace 'check
 850            (lambda* (#:key outputs #:allow-other-keys)
 851              (with-directory-excursion ,(string-append "../LightGBM-" version)
 852                (invoke "pytest" "tests/c_api_test/test_.py")))))))
 853     (build-system cmake-build-system)
 854     (home-page "https://github.com/Microsoft/LightGBM")
 855     (synopsis "Gradient boosting framework based on decision tree algorithms")
 856     (description "LightGBM is a gradient boosting framework that uses tree
 857 based learning algorithms.  It is designed to be distributed and efficient with
 858 the following advantages:
 859
 860 @itemize
 861 @item Faster training speed and higher efficiency
 862 @item Lower memory usage
 863 @item Better accuracy
 864 @item Parallel and GPU learning supported (not enabled in this package)
 865 @item Capable of handling large-scale data
 866 @end itemize\n")
 867     (license license:expat)))
 868
 869 (define-public vowpal-wabbit
 870   ;; Language bindings not included.
 871   (package
 872     (name "vowpal-wabbit")
 873     (version "8.5.0")
 874     (source (origin
 875               (method url-fetch)
 876               (uri (string-append
 877                     "https://github.com/JohnLangford/vowpal_wabbit/archive/"
 878                     version ".tar.gz"))
 879               (sha256
 880                (base32
 881                 "0clp2kb7rk5sckhllxjr5a651awf4s8dgzg4659yh4hf5cqnf0gr"))
 882               (file-name (string-append name "-" version ".tar.gz"))))
 883     (inputs
 884      `(("boost" ,boost)
 885        ("zlib" ,zlib)))
 886     (arguments
 887      `(#:configure-flags
 888        (list (string-append "--with-boost="
 889                             (assoc-ref %build-inputs "boost")))))
 890     (build-system gnu-build-system)
 891     (home-page "https://github.com/JohnLangford/vowpal_wabbit")
 892     (synopsis "Fast machine learning library for online learning")
 893     (description "Vowpal Wabbit is a machine learning system with techniques
 894 such as online, hashing, allreduce, reductions, learning2search, active, and
 895 interactive learning.")
 896     (license license:bsd-3)))
 897
 898 (define-public python2-fastlmm
 899   (package
 900     (name "python2-fastlmm")
 901     (version "0.2.21")
 902     (source
 903      (origin
 904        (method url-fetch)
 905        (uri (pypi-uri "fastlmm" version ".zip"))
 906        (sha256
 907         (base32
 908          "1q8c34rpmwkfy3r4d5172pzdkpfryj561897z9r3x22gq7813x1m"))))
 909     (build-system python-build-system)
 910     (arguments
 911      `(#:tests? #f ; some test files are missing
 912        #:python ,python-2)) ; only Python 2.7 is supported
 913     (propagated-inputs
 914      `(("python2-numpy" ,python2-numpy)
 915        ("python2-scipy" ,python2-scipy)
 916        ("python2-matplotlib" ,python2-matplotlib)
 917        ("python2-pandas" ,python2-pandas)
 918        ("python2-scikit-learn" ,python2-scikit-learn)
 919        ("python2-pysnptools" ,python2-pysnptools)))
 920     (native-inputs
 921      `(("unzip" ,unzip)
 922        ("python2-cython" ,python2-cython)
 923        ("python2-mock" ,python2-mock)
 924        ("python2-nose" ,python2-nose)))
 925     (home-page "http://research.microsoft.com/en-us/um/redmond/projects/mscompbio/fastlmm/")
 926     (synopsis "Perform genome-wide association studies on large data sets")
 927     (description
 928      "FaST-LMM, which stands for Factored Spectrally Transformed Linear Mixed
 929 Models, is a program for performing both single-SNP and SNP-set genome-wide
 930 association studies (GWAS) on extremely large data sets.")
 931     (license license:asl2.0)))
 932
 933 ;; There have been no proper releases yet.
 934 (define-public kaldi
 935   (let ((commit "2f95609f0bb085bd3a1dc5eb0a39f3edea59e606")
 936         (revision "1"))
 937     (package
 938       (name "kaldi")
 939       (version (git-version "0" revision commit))
 940       (source (origin
 941                 (method git-fetch)
 942                 (uri (git-reference
 943                       (url "https://github.com/kaldi-asr/kaldi.git")
 944                       (commit commit)))
 945                 (file-name (git-file-name name version))
 946                 (sha256
 947                  (base32
 948                   "082qh3pfi7hvncylp4xsmkfahbd7gb0whdfa4rwrx7fxk9rdh3kz"))))
 949       (build-system gnu-build-system)
 950       (arguments
 951        `(#:test-target "test"
 952          #:phases
 953          (modify-phases %standard-phases
 954            (add-after 'unpack 'chdir
 955              (lambda _ (chdir "src") #t))
 956            (replace 'configure
 957              (lambda* (#:key build system inputs outputs #:allow-other-keys)
 958                (when (not (or (string-prefix? "x86_64" system)
 959                               (string-prefix? "i686" system)))
 960                  (substitute* "makefiles/linux_openblas.mk"
 961                    (("-msse -msse2") "")))
 962                (substitute* "makefiles/default_rules.mk"
 963                  (("/bin/bash") (which "bash")))
 964                (substitute* "Makefile"
 965                  (("ext_depend: check_portaudio")
 966                   "ext_depend:"))
 967                (substitute* '("online/Makefile"
 968                               "onlinebin/Makefile"
 969                               "gst-plugin/Makefile")
 970                  (("../../tools/portaudio/install")
 971                   (assoc-ref inputs "portaudio")))
 972
 973                ;; This `configure' script doesn't support variables passed as
 974                ;; arguments, nor does it support "prefix".
 975                (let ((out (assoc-ref outputs "out"))
 976                      (openblas (assoc-ref inputs "openblas"))
 977                      (openfst (assoc-ref inputs "openfst")))
 978                  (substitute* "configure"
 979                    (("check_for_slow_expf;") "")
 980                    ;; This affects the RPATH and also serves as the installation
 981                    ;; directory.
 982                    (("KALDILIBDIR=`pwd`/lib")
 983                     (string-append "KALDILIBDIR=" out "/lib")))
 984                  (mkdir-p out) ; must exist
 985                  (setenv "CONFIG_SHELL" (which "bash"))
 986                  (setenv "OPENFST_VER" ,(package-version openfst))
 987                  (invoke "./configure"
 988                          "--use-cuda=no"
 989                          "--shared"
 990                          (string-append "--openblas-root=" openblas)
 991                          (string-append "--fst-root=" openfst)))))
 992            (add-after 'build 'build-ext-and-gstreamer-plugin
 993              (lambda _
 994                (invoke "make" "-C" "online" "depend")
 995                (invoke "make" "-C" "online")
 996                (invoke "make" "-C" "onlinebin" "depend")
 997                (invoke "make" "-C" "onlinebin")
 998                (invoke "make" "-C" "gst-plugin" "depend")
 999                (invoke "make" "-C" "gst-plugin")
1000                #t))
1001            ;; TODO: also install the executables.
1002            (replace 'install
1003              (lambda* (#:key outputs #:allow-other-keys)
1004                (let* ((out (assoc-ref outputs "out"))
1005                       (lib (string-append out "/lib")))
1006                  (mkdir-p lib)
1007                  (install-file "gst-plugin/libgstonlinegmmdecodefaster.so" lib)
1008                  #t))))))
1009       (inputs
1010        `(("alsa-lib" ,alsa-lib)
1011          ("gfortran" ,gfortran "lib")
1012          ("glib" ,glib)
1013          ("gstreamer" ,gstreamer)
1014          ("jack" ,jack-1)
1015          ("openblas" ,openblas)
1016          ("openfst" ,openfst)
1017          ("portaudio" ,portaudio)
1018          ("python" ,python)))
1019       (native-inputs
1020        `(("glib" ,glib "bin")             ; glib-genmarshal
1021          ("grep" ,grep)
1022          ("sed" ,sed)
1023          ("pkg-config" ,pkg-config)
1024          ("which" ,which)))
1025       (home-page "https://kaldi-asr.org/")
1026       (synopsis "Speech recognition toolkit")
1027       (description "Kaldi is an extensible toolkit for speech recognition
1028 written in C++.")
1029       (license license:asl2.0))))