Examples with Trainable com.simiacryptus.mindseye.eval.Trainable used on opensource projects

Example 21 with Trainable

use of com.simiacryptus.mindseye.eval.Trainable in project MindsEye by SimiaCryptus.

the class RecursiveSubspace method buildSubspace.

/**
 * Build subspace nn layer.
 *
 * @param subject     the subject
 * @param measurement the measurement
 * @param monitor     the monitor
 * @return the nn layer
 */
@Nullable
public Layer buildSubspace(@Nonnull Trainable subject, @Nonnull PointSample measurement, @Nonnull TrainingMonitor monitor) {
    @Nonnull PointSample origin = measurement.copyFull().backup();
    @Nonnull final DeltaSet<Layer> direction = measurement.delta.scale(-1);
    final double magnitude = direction.getMagnitude();
    if (Math.abs(magnitude) < 1e-10) {
        monitor.log(String.format("Zero gradient: %s", magnitude));
    } else if (Math.abs(magnitude) < 1e-5) {
        monitor.log(String.format("Low gradient: %s", magnitude));
    }
    boolean hasPlaceholders = direction.getMap().entrySet().stream().filter(x -> x.getKey() instanceof PlaceholderLayer).findAny().isPresent();
    List<Layer> deltaLayers = direction.getMap().entrySet().stream().map(x -> x.getKey()).filter(x -> !(x instanceof PlaceholderLayer)).collect(Collectors.toList());
    int size = deltaLayers.size() + (hasPlaceholders ? 1 : 0);
    if (null == weights || weights.length != size)
        weights = new double[size];
    return new LayerBase() {

        @Nonnull
        Layer self = this;

        @Nonnull
        @Override
        public Result eval(Result... array) {
            assertAlive();
            origin.restore();
            IntStream.range(0, deltaLayers.size()).forEach(i -> {
                direction.getMap().get(deltaLayers.get(i)).accumulate(weights[hasPlaceholders ? (i + 1) : i]);
            });
            if (hasPlaceholders) {
                direction.getMap().entrySet().stream().filter(x -> x.getKey() instanceof PlaceholderLayer).distinct().forEach(entry -> entry.getValue().accumulate(weights[0]));
            }
            PointSample measure = subject.measure(monitor);
            double mean = measure.getMean();
            monitor.log(String.format("RecursiveSubspace: %s <- %s", mean, Arrays.toString(weights)));
            direction.addRef();
            return new Result(TensorArray.wrap(new Tensor(mean)), (DeltaSet<Layer> buffer, TensorList data) -> {
                DoubleStream deltaStream = deltaLayers.stream().mapToDouble(layer -> {
                    Delta<Layer> a = direction.getMap().get(layer);
                    Delta<Layer> b = measure.delta.getMap().get(layer);
                    return b.dot(a) / Math.max(Math.sqrt(a.dot(a)), 1e-8);
                });
                if (hasPlaceholders) {
                    deltaStream = DoubleStream.concat(DoubleStream.of(direction.getMap().keySet().stream().filter(x -> x instanceof PlaceholderLayer).distinct().mapToDouble(layer -> {
                        Delta<Layer> a = direction.getMap().get(layer);
                        Delta<Layer> b = measure.delta.getMap().get(layer);
                        return b.dot(a) / Math.max(Math.sqrt(a.dot(a)), 1e-8);
                    }).sum()), deltaStream);
                }
                buffer.get(self, weights).addInPlace(deltaStream.toArray()).freeRef();
            }) {

                @Override
                protected void _free() {
                    measure.freeRef();
                    direction.freeRef();
                }

                @Override
                public boolean isAlive() {
                    return true;
                }
            };
        }

        @Override
        protected void _free() {
            direction.freeRef();
            origin.freeRef();
            super._free();
        }

        @Nonnull
        @Override
        public JsonObject getJson(Map<CharSequence, byte[]> resources, DataSerializer dataSerializer) {
            throw new IllegalStateException();
        }

        @Nullable
        @Override
        public List<double[]> state() {
            return null;
        }
    };
}

Example 22 with Trainable

use of com.simiacryptus.mindseye.eval.Trainable in project MindsEye by SimiaCryptus.

the class RecursiveSubspace method train.

/**
 * Train.
 *
 * @param monitor    the monitor
 * @param macroLayer the macro layer
 */
public void train(@Nonnull TrainingMonitor monitor, Layer macroLayer) {
    @Nonnull BasicTrainable inner = new BasicTrainable(macroLayer);
    // @javax.annotation.Nonnull Tensor tensor = new Tensor();
    @Nonnull ArrayTrainable trainable = new ArrayTrainable(inner, new Tensor[][] { {} });
    inner.freeRef();
    // tensor.freeRef();
    new IterativeTrainer(trainable).setOrientation(new LBFGS()).setLineSearchFactory(n -> new ArmijoWolfeSearch()).setMonitor(new TrainingMonitor() {

        @Override
        public void log(String msg) {
            monitor.log("\t" + msg);
        }
    }).setMaxIterations(getIterations()).setIterationsPerSample(getIterations()).setTerminateThreshold(terminateThreshold).runAndFree();
    trainable.freeRef();
}

Example 23 with Trainable

use of com.simiacryptus.mindseye.eval.Trainable in project MindsEye by SimiaCryptus.

the class StaticRateTest method train.

@Override
public void train(@Nonnull final NotebookOutput log, @Nonnull final Layer network, @Nonnull final Tensor[][] trainingData, final TrainingMonitor monitor) {
    log.code(() -> {
        @Nonnull final SimpleLossNetwork supervisedNetwork = new SimpleLossNetwork(network, new EntropyLossLayer());
        @Nonnull final Trainable trainable = new SampledArrayTrainable(trainingData, supervisedNetwork, 1000);
        return new IterativeTrainer(trainable).setMonitor(monitor).setOrientation(new GradientDescent()).setLineSearchFactory((@Nonnull final CharSequence name) -> new StaticLearningRate(0.001)).setTimeout(3, TimeUnit.MINUTES).setMaxIterations(500).runAndFree();
    });
}

Example 24 with Trainable

use of com.simiacryptus.mindseye.eval.Trainable in project MindsEye by SimiaCryptus.

the class L1NormalizationTest method train.

@Override
public void train(@Nonnull final NotebookOutput log, @Nonnull final Layer network, @Nonnull final Tensor[][] trainingData, final TrainingMonitor monitor) {
    log.code(() -> {
        @Nonnull final SimpleLossNetwork supervisedNetwork = new SimpleLossNetwork(network, new EntropyLossLayer());
        @Nonnull final Trainable trainable = new L12Normalizer(new SampledArrayTrainable(trainingData, supervisedNetwork, 1000)) {

            @Override
            public Layer getLayer() {
                return inner.getLayer();
            }

            @Override
            protected double getL1(final Layer layer) {
                return 1.0;
            }

            @Override
            protected double getL2(final Layer layer) {
                return 0;
            }
        };
        return new IterativeTrainer(trainable).setMonitor(monitor).setTimeout(3, TimeUnit.MINUTES).setMaxIterations(500).runAndFree();
    });
}

Example 25 with Trainable

use of com.simiacryptus.mindseye.eval.Trainable in project MindsEye by SimiaCryptus.

the class SimpleGradientDescentTest method train.

@Override
public void train(@Nonnull final NotebookOutput log, @Nonnull final Layer network, @Nonnull final Tensor[][] trainingData, final TrainingMonitor monitor) {
    log.p("Training a model involves a few different components. First, our model is combined mapCoords a loss function. " + "Then we take that model and combine it mapCoords our training data to define a trainable object. " + "Finally, we use a simple iterative scheme to refine the weights of our model. " + "The final output is the last output value of the loss function when evaluating the last batch.");
    log.code(() -> {
        @Nonnull final SimpleLossNetwork supervisedNetwork = new SimpleLossNetwork(network, new EntropyLossLayer());
        @Nonnull final ArrayList<Tensor[]> trainingList = new ArrayList<>(Arrays.stream(trainingData).collect(Collectors.toList()));
        Collections.shuffle(trainingList);
        @Nonnull final Tensor[][] randomSelection = trainingList.subList(0, 10000).toArray(new Tensor[][] {});
        @Nonnull final Trainable trainable = new ArrayTrainable(randomSelection, supervisedNetwork);
        return new IterativeTrainer(trainable).setMonitor(monitor).setTimeout(3, TimeUnit.MINUTES).setMaxIterations(500).runAndFree();
    });
}