mle-diag-gmm-test.cc 14 KB
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403 404 405 406 407 408 409 410 411 412
// gmm/mle-diag-gmm-test.cc

// Copyright 2009-2011  Georg Stemmer;  Jan Silovsky;  Saarland University;
//                      Microsoft Corporation;  Yanmin Qian

// See ../../COPYING for clarification regarding multiple authors
//
// Licensed under the Apache License, Version 2.0 (the "License");
// you may not use this file except in compliance with the License.
// You may obtain a copy of the License at
//
//  http://www.apache.org/licenses/LICENSE-2.0
//
// THIS CODE IS PROVIDED *AS IS* BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY
// KIND, EITHER EXPRESS OR IMPLIED, INCLUDING WITHOUT LIMITATION ANY IMPLIED
// WARRANTIES OR CONDITIONS OF TITLE, FITNESS FOR A PARTICULAR PURPOSE,
// MERCHANTABLITY OR NON-INFRINGEMENT.
// See the Apache 2 License for the specific language governing permissions and
// limitations under the License.

#include "gmm/diag-gmm.h"
#include "gmm/diag-gmm-normal.h"
#include "gmm/mle-diag-gmm.h"
#include "util/kaldi-io.h"

using namespace kaldi;

void TestComponentAcc(const DiagGmm &gmm, const Matrix<BaseFloat> &feats) {
  MleDiagGmmOptions config;
  AccumDiagGmm est_atonce;    // updates all components
  AccumDiagGmm est_compwise;  // updates single components

  // Initialize estimators
  est_atonce.Resize(gmm.NumGauss(), gmm.Dim(), kGmmAll);
  est_atonce.SetZero(kGmmAll);
  est_compwise.Resize(gmm.NumGauss(),
      gmm.Dim(), kGmmAll);
  est_compwise.SetZero(kGmmAll);

  // accumulate estimators
  for (int32 i = 0; i < feats.NumRows(); i++) {
    est_atonce.AccumulateFromDiag(gmm, feats.Row(i), 1.0F);
    Vector<BaseFloat> post(gmm.NumGauss());
    gmm.ComponentPosteriors(feats.Row(i), &post);
    for (int32 m = 0; m < gmm.NumGauss(); m++) {
      est_compwise.AccumulateForComponent(feats.Row(i), m, post(m));
    }
  }

  DiagGmm gmm_atonce;    // model with all components accumulated together
  DiagGmm gmm_compwise;  // model with each component accumulated separately
  gmm_atonce.Resize(gmm.NumGauss(), gmm.Dim());
  gmm_compwise.Resize(gmm.NumGauss(), gmm.Dim());

  MleDiagGmmUpdate(config, est_atonce, kGmmAll, &gmm_atonce, NULL, NULL);
  MleDiagGmmUpdate(config, est_compwise, kGmmAll, &gmm_compwise, NULL, NULL);

  // the two ways of updating should result in the same model
  double loglike0 = 0.0;
  double loglike1 = 0.0;
  double loglike2 = 0.0;
  for (int32 i = 0; i < feats.NumRows(); i++) {
    loglike0 += static_cast<double>(gmm.LogLikelihood(feats.Row(i)));
    loglike1 += static_cast<double>(gmm_atonce.LogLikelihood(feats.Row(i)));
    loglike2 += static_cast<double>(gmm_compwise.LogLikelihood(feats.Row(i)));
  }

  std::cout << "Per-frame log-likelihood before update = "
      << (loglike0/feats.NumRows()) << '\n';
  std::cout << "Per-frame log-likelihood (accumulating at once) = "
      << (loglike1/feats.NumRows()) << '\n';
  std::cout << "Per-frame log-likelihood (accumulating component-wise) = "
      << (loglike2/feats.NumRows()) << '\n';

  AssertEqual(loglike1, loglike2, 1.0e-6);

  if (est_atonce.NumGauss() != gmm.NumGauss()) {
    KALDI_WARN << "Unable to pass test_update_flags() test because of "
      "component removal during Update() call (this is normal)";
    return;
  } else {
    KALDI_ASSERT(loglike1 >= loglike0 - (std::abs(loglike1)+std::abs(loglike0))*1.0e-06);
    KALDI_ASSERT(loglike2 >= loglike0 - (std::abs(loglike2)+std::abs(loglike0))*1.0e-06);
  }
}

void test_flags_driven_update(const DiagGmm &gmm,
                              const Matrix<BaseFloat> &feats,
                              GmmFlagsType flags) {
  MleDiagGmmOptions config;
  AccumDiagGmm est_gmm_allp;   // updates all params
  // let's trust that all-params update works
  AccumDiagGmm est_gmm_somep;  // updates params indicated by flags

  // warm-up estimators
  est_gmm_allp.Resize(gmm.NumGauss(),
    gmm.Dim(), kGmmAll);
  est_gmm_allp.SetZero(kGmmAll);
  est_gmm_somep.Resize(gmm.NumGauss(),
    gmm.Dim(), flags);
  est_gmm_somep.SetZero(flags);

  // accumulate estimators
  for (int32 i = 0; i < feats.NumRows(); i++) {
    est_gmm_allp.AccumulateFromDiag(gmm, feats.Row(i), 1.0F);
    est_gmm_somep.AccumulateFromDiag(gmm, feats.Row(i), 1.0F);
  }

  DiagGmm gmm_all_update;   // model with all params updated
  DiagGmm gmm_some_update;  // model with some params updated
  gmm_all_update.CopyFromDiagGmm(gmm);   // init with orig. model
  gmm_some_update.CopyFromDiagGmm(gmm);  // init with orig. model

  MleDiagGmmUpdate(config, est_gmm_allp, kGmmAll, &gmm_all_update, NULL, NULL);
  MleDiagGmmUpdate(config, est_gmm_somep, flags, &gmm_some_update, NULL, NULL);

  if (est_gmm_allp.NumGauss() != gmm.NumGauss()) {
    KALDI_WARN << "Unable to pass test_update_flags() test because of "
      "component removal during Update() call (this is normal)";
    return;
  }

  // now back-off the gmm_all_update params that were not updated
  // in gmm_some_update to orig.
  if (~flags & kGmmWeights)
    gmm_all_update.SetWeights(gmm.weights());
  if (~flags & kGmmMeans) {
    Matrix<BaseFloat> means(gmm.NumGauss(), gmm.Dim());
    gmm.GetMeans(&means);
    gmm_all_update.SetMeans(means);
  }
  if (~flags & kGmmVariances) {
    Matrix<BaseFloat> vars(gmm.NumGauss(), gmm.Dim());
    gmm.GetVars(&vars);
    vars.InvertElements();
    gmm_all_update.SetInvVars(vars);
  }
  gmm_all_update.ComputeGconsts();

  // now both models gmm_all_update, gmm_all_update have the same params updated
  // compute loglike for models for check
  double loglike0 = 0.0;
  double loglike1 = 0.0;
  double loglike2 = 0.0;
  for (int32 i = 0; i < feats.NumRows(); i++) {
    loglike0 += static_cast<double>(
      gmm.LogLikelihood(feats.Row(i)));
    loglike1 += static_cast<double>(
      gmm_all_update.LogLikelihood(feats.Row(i)));
    loglike2 += static_cast<double>(
      gmm_some_update.LogLikelihood(feats.Row(i)));
  }
  if ((flags & kGmmVariances) && !(flags & kGmmMeans))
    return;  // Don't run the test as the variance update gives a different
  // answer if you don't update the mean.

  AssertEqual(loglike1, loglike2, 1.0e-6);
}

void
test_io(const DiagGmm &gmm, const AccumDiagGmm &est_gmm, bool binary,
        const Matrix<BaseFloat> &feats) {
  std::cout << "Testing I/O, binary = " << binary << '\n';

  est_gmm.Write(Output("tmp_stats", binary).Stream(), binary);

  bool binary_in;
  AccumDiagGmm est_gmm2;
  est_gmm2.Resize(est_gmm.NumGauss(),
    est_gmm.Dim(), kGmmAll);
  Input ki("tmp_stats", &binary_in);
  est_gmm2.Read(ki.Stream(), binary_in, false);  // not adding

  Input ki2("tmp_stats", &binary_in);
  est_gmm2.Read(ki2.Stream(), binary_in, true);  // adding

  est_gmm2.Scale(0.5, kGmmAll);
    // 0.5 -> make it same as what it would have been if we read just once.
    // [may affect it due to removal of components with small counts].

  MleDiagGmmOptions config;
  DiagGmm gmm1;
  DiagGmm gmm2;
  gmm1.CopyFromDiagGmm(gmm);
  gmm2.CopyFromDiagGmm(gmm);
  MleDiagGmmUpdate(config, est_gmm, est_gmm.Flags(), &gmm1, NULL, NULL);
  MleDiagGmmUpdate(config, est_gmm2, est_gmm2.Flags(), &gmm2, NULL, NULL);

  BaseFloat loglike1 = 0.0;
  BaseFloat loglike2 = 0.0;
  for (int32 i = 0; i < feats.NumRows(); i++) {
    loglike1 += gmm1.LogLikelihood(feats.Row(i));
    loglike2 += gmm2.LogLikelihood(feats.Row(i));
  }

  AssertEqual(loglike1, loglike2, 1.0e-6);

  unlink("tmp_stats");
}

void
UnitTestEstimateDiagGmm() {
  size_t dim = 15;  // dimension of the gmm
  size_t nMix = 9;  // number of mixtures in the data
  size_t maxiterations = 20;  // number of iterations for estimation

  // maximum number of densities in the GMM
  // larger than the number of mixtures in the data
  // so that we can test the removal of unseen components
  int32 maxcomponents = 10;

  // generate random feature vectors
  Matrix<BaseFloat> means_f(nMix, dim), vars_f(nMix, dim);
  // first, generate random mean and variance vectors
  for (size_t m = 0; m < nMix; m++) {
    for (size_t d= 0; d < dim; d++) {
      means_f(m, d) = kaldi::RandGauss()*100.0F;
      vars_f(m, d) = Exp(kaldi::RandGauss())*1000.0F+ 1.0F;
    }
//    std::cout << "Gauss " << m << ": Mean = " << means_f.Row(m) << '\n'
//        << "Vars = " << vars_f.Row(m) << '\n';
  }
  // second, generate 1000 feature vectors for each of the mixture components
  size_t counter = 0, multiple = 200;
  Matrix<BaseFloat> feats(nMix*multiple, dim);
  for (size_t m = 0; m < nMix; m++) {
    for (size_t i = 0; i < multiple; i++) {
      for (size_t d = 0; d < dim; d++) {
        feats(counter, d) = means_f(m, d) + kaldi::RandGauss() *
            std::sqrt(vars_f(m, d));
      }
      counter++;
    }
  }
  // Compute the global mean and variance
  Vector<BaseFloat> mean_acc(dim);
  Vector<BaseFloat> var_acc(dim);
  Vector<BaseFloat> featvec(dim);
  for (size_t i = 0; i < counter; i++) {
    featvec.CopyRowFromMat(feats, i);
    mean_acc.AddVec(1.0, featvec);
    featvec.ApplyPow(2.0);
    var_acc.AddVec(1.0, featvec);
  }
  mean_acc.Scale(1.0F/counter);
  var_acc.Scale(1.0F/counter);
  var_acc.AddVec2(-1.0, mean_acc);
//  std::cout << "Mean acc = " << mean_acc << '\n' << "Var acc = "
//      << var_acc << '\n';

  // write the feature vectors to a file
  //  std::ofstream of("tmpfeats");
  //  of.precision(10);
  //  of << feats;
  //  of.close();

  // now generate randomly initial values for the GMM
  Vector<BaseFloat> weights(1);
  Matrix<BaseFloat> means(1, dim), vars(1, dim), invvars(1, dim);
  for (size_t d= 0; d < dim; d++) {
    means(0, d) = kaldi::RandGauss()*100.0F;
    vars(0, d) = Exp(kaldi::RandGauss()) *10.0F + 1e-5F;
  }
  weights(0) = 1.0F;
  invvars.CopyFromMat(vars);
  invvars.InvertElements();

  // new GMM
  DiagGmm *gmm = new DiagGmm();
  gmm->Resize(1, dim);
  gmm->SetWeights(weights);
  gmm->SetInvVarsAndMeans(invvars, means);
  gmm->ComputeGconsts();

  {
    KALDI_LOG << "Testing natural<>normal conversion";
    DiagGmmNormal ngmm(*gmm);
    DiagGmm rgmm;
    rgmm.Resize(1, dim);
    ngmm.CopyToDiagGmm(&rgmm);

    // check contents
    KALDI_ASSERT(ApproxEqual(weights(0), 1.0F, 1e-6));
    KALDI_ASSERT(ApproxEqual(gmm->weights()(0), rgmm.weights()(0), 1e-6));
    for (int32 d = 0; d < dim; d++) {
      KALDI_ASSERT(ApproxEqual(means.Row(0)(d), ngmm.means_.Row(0)(d), 1e-6));
      KALDI_ASSERT(ApproxEqual(1./invvars.Row(0)(d), ngmm.vars_.Row(0)(d), 1e-6));
      KALDI_ASSERT(ApproxEqual(gmm->means_invvars().Row(0)(d), rgmm.means_invvars().Row(0)(d), 1e-6));
      KALDI_ASSERT(ApproxEqual(gmm->inv_vars().Row(0)(d), rgmm.inv_vars().Row(0)(d), 1e-6));
    }
    KALDI_LOG << "OK";
  }

  AccumDiagGmm est_gmm;
//  var_acc.Scale(0.1);
//  est_gmm.config_.p_variance_floor_vector = &var_acc;

  MleDiagGmmOptions  config;
  config.min_variance = 0.01;
  GmmFlagsType flags = kGmmAll;  // Should later try reducing this.

  est_gmm.Resize(gmm->NumGauss(), gmm->Dim(), flags);

  // iterate
  size_t iteration = 0;
  float lastloglike = 0.0;
  int32 lastloglike_nM = 0;

  while (iteration < maxiterations) {
    Vector<BaseFloat> featvec(dim);
    est_gmm.Resize(gmm->NumGauss(), gmm->Dim(), flags);
    est_gmm.SetZero(flags);
    double loglike = 0.0;
    for (size_t i = 0; i < counter; i++) {
      featvec.CopyRowFromMat(feats, i);
      loglike += static_cast<double>(est_gmm.AccumulateFromDiag(*gmm,
        featvec, 1.0F));
    }
    std::cout << "Loglikelihood before iteration " << iteration << " : "
        << std::scientific << loglike << " number of components: "
        << gmm->NumGauss() << '\n';

    // every 5th iteration check loglike change and update lastloglike
    if (iteration % 5 == 0) {
      // likelihood should be increasing on the long term
      if ((iteration > 0) && (gmm->NumGauss() >= lastloglike_nM)) {
        KALDI_ASSERT(loglike - lastloglike >= -1.0);
      }
      lastloglike = loglike;
      lastloglike_nM = gmm->NumGauss();
    }

    // binary write
    est_gmm.Write(Output("tmp_stats", true).Stream(), true);

    // binary read
    bool binary_in;
    Input ki("tmp_stats", &binary_in);
    est_gmm.Read(ki.Stream(), binary_in, false);  // false = not adding.

    BaseFloat obj, count;
    MleDiagGmmUpdate(config, est_gmm, flags, gmm, &obj, &count);
    KALDI_LOG <<"ML objective function change = " << (obj/count)
              << " per frame, over " << (count) << " frames.";

    if ((iteration % 3 == 1) && (gmm->NumGauss() * 2 <= maxcomponents)) {
      gmm->Split(gmm->NumGauss() * 2, 0.001);
    }

    if (iteration == 5) {  // run following tests with not too overfitted model
      std::cout << "Testing flags-driven updates" << '\n';
      test_flags_driven_update(*gmm, feats, kGmmAll);
      test_flags_driven_update(*gmm, feats, kGmmWeights);
      test_flags_driven_update(*gmm, feats, kGmmMeans);
      test_flags_driven_update(*gmm, feats, kGmmVariances);
      test_flags_driven_update(*gmm, feats, kGmmWeights | kGmmMeans);
      std::cout << "Testing component-wise accumulation" << '\n';
      TestComponentAcc(*gmm, feats);
    }

    iteration++;
  }

  {  // I/O tests
    GmmFlagsType flags_all = kGmmAll;
    est_gmm.Resize(gmm->NumGauss(),
      gmm->Dim(), flags_all);
    est_gmm.SetZero(flags_all);
    float loglike = 0.0;
    for (size_t i = 0; i < counter; i++) {
      loglike += est_gmm.AccumulateFromDiag(*gmm, feats.Row(i), 1.0F);
    }
    test_io(*gmm, est_gmm, false, feats);  // ASCII mode
    test_io(*gmm, est_gmm, true, feats);   // Binary mode
  }

  { // Test multi-threaded update.
    GmmFlagsType flags_all = kGmmAll;
    est_gmm.Resize(gmm->NumGauss(),
      gmm->Dim(), flags_all);
    est_gmm.SetZero(flags_all);

    Vector<BaseFloat> weights(counter);
    for (size_t i = 0; i < counter; i++)
      weights(i) = 0.5 + 0.1 * (Rand() % 10);


    float loglike = 0.0;
    for (size_t i = 0; i < counter; i++) {
      loglike += weights(i) *
          est_gmm.AccumulateFromDiag(*gmm, feats.Row(i), weights(i));
    }
    AccumDiagGmm est_gmm2(*gmm, flags_all);
    int32 num_threads = 2;
    float loglike2 =
        est_gmm2.AccumulateFromDiagMultiThreaded(*gmm, feats, weights, num_threads);
    AssertEqual(loglike, loglike2);
    est_gmm.AssertEqual(est_gmm2);
  }


  delete gmm;

  unlink("tmp_stats");
}

int main() {
  // repeat the test five times
  for (int i = 0; i < 2; i++)
    UnitTestEstimateDiagGmm();
  std::cout << "Test OK.\n";
}