Package gov.sandia.cognition.statistics.distribution

Examples of gov.sandia.cognition.statistics.distribution.UnivariateGaussian

450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
      LogitMixParticle priorParticle, ObservedValue<Vector, Matrix> data) {

    final LogitMixParticle updatedParticle = priorParticle.clone();
    final KalmanFilter filter = updatedParticle.getRegressionFilter();

    final UnivariateGaussian evComponent = updatedParticle.EVcomponent;

    final boolean isOne = !data.getObservedValue().isZero();

    final int smplLowerIdx = DiscreteSamplingUtil.sampleIndexFromProportions(
        getRandom(), updatedParticle.getComponentLikelihoods());
//    final int smplLowerIdx = DiscreteSamplingUtil.sampleIndexFromProbabilities(
//        getRandom(), this.evDistribution.getPriorWeights());

    final UnivariateGaussian partComponent =
        this.evDistribution.getDistributions().get(smplLowerIdx);

    final double dsampledAugResponse = sampleAugResponse(
        updatedParticle.getPriorPredMean(),
        updatedParticle.getPriorPredCov(), isOne,
        partComponent);

    // TODO we should've already set this, so it might be redundant.
    filter.setMeasurementCovariance(
        MatrixFactory.getDefault().copyArray(new double[][] {{
          evComponent.getVariance() + partComponent.getVariance()}}));

    final Vector sampledAugResponse =
        VectorFactory.getDefault().copyValues(
            dsampledAugResponse
            - evComponent.getMean().doubleValue()
            - partComponent.getMean().doubleValue());

    updatedParticle.setAugResponseSample(sampledAugResponse);

    final MultivariateGaussian posteriorState = updatedParticle.getLinearState().clone();
    filter.update(posteriorState, sampledAugResponse);
View Full Code Here
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
  protected GaussianArTransitionState propagate(
      GaussianArTransitionState prevState, int predClass, ObservedValue<Double,Void> data) {
    /*
     * Perform the filtering step
     */
    final UnivariateGaussian priorDist = prevState.getSuffStat();
    final double priorPredMean = a[predClass] * priorDist.getMean();
    final double priorPredCov = a[predClass] * a[predClass] * priorDist.getVariance() + sigma2[predClass];
   
    final StandardHMM<Double> newHmm = prevState.getHmm();

    final GaussianArTransitionState newTransState =
        new GaussianArTransitionState(prevState, newHmm,
            predClass, data, null);

    newTransState.setPriorPredSuffStats(new UnivariateGaussian(priorPredMean, priorPredCov));
   
    return newTransState;
  }
View Full Code Here
116
117
118
119
120
121
122
123
124
125
126
127
      final double postCovariance = 1d/(1d/priorPredCov + 1d/sigma_y2);
      final double postMean = (priorPredMean/priorPredCov
          + priorParameter.getObservation().getObservedValue()/sigma_y2)
          * postCovariance;

      final UnivariateGaussian postDist =
          new UnivariateGaussian(postMean, postCovariance);
 
      final GaussianArTransitionState postState =
          priorParameter.clone();
      postState.setSuffStat(postDist);

View Full Code Here
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
      final double upperVar = particle.getPriorPredCov();
      double logLikelihood = Double.NEGATIVE_INFINITY;
      double[] componentLikelihoods = new double[10];
      for (int i = 0; i < 10; i++) {

        final UnivariateGaussian partComponent =
            this.evDistribution.getDistributions().get(i);

        // TODO FIXME: do this in log scale...
        double partLogLik =
            ExtStatisticsUtils.normalCdf(0d,
            upperMean - partComponent.getMean(),
            Math.sqrt(upperVar + partComponent.getVariance()),
            true);
        if (observation.getObservedValue().getElement(0) > 0d) {
//        if (!observation.getObservedValue().isZero()) {
          partLogLik = LogMath.subtract(0d, partLogLik);
        }
View Full Code Here
131
132
133
134
135
136
137
138
139
140
141
      for (int i = 0; i < numParticles; i++) {
       
        final MultivariateGaussian initialPriorState = initialPrior.clone();
        final KalmanFilter kf = this.initialFilter.clone();
        final int componentId = this.rng.nextInt(10);
        final UnivariateGaussian evDist = this.evDistribution.
            getDistributions().get(componentId);
       
        final LogitMixParticle particleMvgDPDist =
            new LogitMixParticle(null,
                kf, initialPriorState, evDist);
View Full Code Here
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
          predictiveParticle.setPreviousParticle(particle);
          predictiveParticle.setBetaSample(betaMean);
          predictiveParticle.setLinearState(predictivePrior);

          final UnivariateGaussian componentDist =
              filter.evDistribution.getDistributions().get(j);

          predictiveParticle.setEVcomponent(componentDist);
         
          /*
           * Update the observed data for the regression component.
           */
          predictiveParticle.getRegressionFilter().getModel().setC(F);

          final Matrix compVar = MatrixFactory.getDefault().copyArray(
              new double[][] {{componentDist.getVariance()}});
          predictiveParticle.getRegressionFilter().setMeasurementCovariance(compVar);
         
          final double compPredPriorObsMean =
               F.times(betaMean).getElement(0)
               + componentDist.getMean();
          final double compPredPriorObsCov =
               F.times(predictivePrior.getCovariance()).times(F.transpose()).getElement(0, 0)
               + componentDist.getVariance();
          predictiveParticle.setPriorPredMean(compPredPriorObsMean);
          predictiveParticle.setPriorPredCov(compPredPriorObsCov);

          final double logLikelihood =
              filter.updater.computeLogLikelihood(predictiveParticle, data);
View Full Code Here
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
          predictiveParticle.setPreviousParticle(particle);
          predictiveParticle.setBetaSample(betaSample);
          predictiveParticle.setAugResponseSample(sampledAugResponse);
          predictiveParticle.setLinearState(predictivePrior);

          final UnivariateGaussian componentDist =
              this.evDistribution.getDistributions().get(j);

          predictiveParticle.setEVcomponent(componentDist);
         
          /*
           * Update the observed data for the regression component.
           */
          predictiveParticle.getRegressionFilter().getModel().setC(F);

          // TODO would be great to have a 1x1 matrix class here...
          final Matrix compVar = MatrixFactory.getDefault().copyArray(
              new double[][] {{componentDist.getVariance()}});
          predictiveParticle.getRegressionFilter().setMeasurementCovariance(compVar);
         
          final double compPredPriorObsMean =
               F.times(betaSample).getElement(0)
               + componentDist.getMean();
          final double compPredPriorObsCov =
               F.times(predictivePrior.getCovariance()).times(F.transpose()).getElement(0, 0)
               + componentDist.getVariance();
          predictiveParticle.setPriorPredMean(compPredPriorObsMean);
          predictiveParticle.setPriorPredCov(compPredPriorObsCov);

          final double logLikelihood =
              this.updater.computeLogLikelihood(predictiveParticle, data);
View Full Code Here
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
  private LogitMixParticle sufficientStatUpdate(
      LogitMixParticle priorParticle, ObservedValue<Vector, Matrix> data) {
    final LogitMixParticle updatedParticle = priorParticle.clone();
   
    final KalmanFilter filter = updatedParticle.getRegressionFilter();
    final UnivariateGaussian evComponent = updatedParticle.EVcomponent;
    // TODO we should've already set this, so it might be redundant.
    filter.setMeasurementCovariance(
        MatrixFactory.getDefault().copyArray(new double[][] {{
          evComponent.getVariance()}}));

    final Vector sampledAugResponse = priorParticle.getAugResponseSample();
    final Vector diffAugResponse =
        sampledAugResponse.minus(VectorFactory.getDefault().copyArray(
        new double[] {
            evComponent.getMean().doubleValue()
            }));
    final MultivariateGaussian posteriorState = updatedParticle.getLinearState().clone();
    filter.update(posteriorState,
        diffAugResponse);
    updatedParticle.setLinearState(posteriorState);
View Full Code Here
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
      final double upperMean = particle.getPriorPredMean();
      final double upperVar = particle.getPriorPredCov();
      double logLikelihood = Double.NEGATIVE_INFINITY;
      for (int i = 0; i < 10; i++) {

        final UnivariateGaussian partComponent =
            this.evDistribution.getDistributions().get(i);

        // TODO FIXME: do this in log scale...
        double partLogLik =
            ExtStatisticsUtils.normalCdf(0d,
            upperMean - partComponent.getMean(),
            Math.sqrt(upperVar + partComponent.getVariance()),
            true);
        if (!observation.getObservedValue().isZero()) {
          partLogLik = LogMath.subtract(0d, partLogLik);
        }
View Full Code Here
110
111
112
113
114
115
116
117
118
119
120
      for (int i = 0; i < numParticles; i++) {
       
        final MultivariateGaussian initialPriorState = initialPrior.clone();
        final KalmanFilter kf = this.initialFilter.clone();
        final int componentId = this.rng.nextInt(10);
        final UnivariateGaussian evDist = this.evDistribution.
            getDistributions().get(componentId);
       
        final LogitMixParticle particleMvgDPDist =
            new LogitMixParticle(null,
                kf, initialPriorState, evDist);
View Full Code Here
TOP

Related Classes of gov.sandia.cognition.statistics.distribution.UnivariateGaussian

  • plm.hmm.gaussian.GaussianArHmmMkfFilter
  • plm.hmm.gaussian.GaussianArHmmMkfFilterTest
  • plm.hmm.gaussian.GaussianArHmmPlFilter
  • plm.hmm.gaussian.GaussianArHmmPlFilter$GaussianArHmmPlUpdater
  • plm.logit.fruehwirth.LogitFSWFFilter
  • plm.logit.fruehwirth.LogitFSWFFilter$LogitFSWFUpdater
  • plm.logit.fruehwirth.LogitParRBCWFFilter
  • plm.logit.fruehwirth.LogitParRBCWFFilter$LogitParRBCWFUpdater
  • plm.logit.fruehwirth.LogitParRBCWFFilter$PrepAndWeightParticles
  • plm.logit.fruehwirth.LogitRBCWFFilter
    • Copyright © 2018 www.massapicom. All rights reserved.
    All source code are property of their respective owners. Java is a trademark of Sun Microsystems, Inc and owned by ORACLE Inc. Contact coftware#gmail.com.