Examples with TrainingMonitor com.simiacryptus.mindseye.opt.TrainingMonitor used on opensource projects

Example 16 with TrainingMonitor

use of com.simiacryptus.mindseye.opt.TrainingMonitor in project MindsEye by SimiaCryptus.

the class OwlQn method orient.

@Nonnull
@Override
public LineSearchCursor orient(final Trainable subject, @Nonnull final PointSample measurement, final TrainingMonitor monitor) {
    @Nonnull final SimpleLineSearchCursor gradient = (SimpleLineSearchCursor) inner.orient(subject, measurement, monitor);
    @Nonnull final DeltaSet<Layer> searchDirection = gradient.direction.copy();
    @Nonnull final DeltaSet<Layer> orthant = new DeltaSet<Layer>();
    for (@Nonnull final Layer layer : getLayers(gradient.direction.getMap().keySet())) {
        final double[] weights = gradient.direction.getMap().get(layer).target;
        @Nullable final double[] delta = gradient.direction.getMap().get(layer).getDelta();
        @Nullable final double[] searchDir = searchDirection.get(layer, weights).getDelta();
        @Nullable final double[] suborthant = orthant.get(layer, weights).getDelta();
        for (int i = 0; i < searchDir.length; i++) {
            final int positionSign = sign(weights[i]);
            final int directionSign = sign(delta[i]);
            suborthant[i] = 0 == positionSign ? directionSign : positionSign;
            searchDir[i] += factor_L1 * (weights[i] < 0 ? -1.0 : 1.0);
            if (sign(searchDir[i]) != directionSign) {
                searchDir[i] = delta[i];
            }
        }
        assert null != searchDir;
    }
    return new SimpleLineSearchCursor(subject, measurement, searchDirection) {

        @Nonnull
        @Override
        public LineSearchPoint step(final double alpha, final TrainingMonitor monitor) {
            origin.weights.stream().forEach(d -> d.restore());
            @Nonnull final DeltaSet<Layer> currentDirection = direction.copy();
            direction.getMap().forEach((layer, buffer) -> {
                if (null == buffer.getDelta())
                    return;
                @Nullable final double[] currentDelta = currentDirection.get(layer, buffer.target).getDelta();
                for (int i = 0; i < buffer.getDelta().length; i++) {
                    final double prevValue = buffer.target[i];
                    final double newValue = prevValue + buffer.getDelta()[i] * alpha;
                    if (sign(prevValue) != 0 && sign(prevValue) != sign(newValue)) {
                        currentDelta[i] = 0;
                        buffer.target[i] = 0;
                    } else {
                        buffer.target[i] = newValue;
                    }
                }
            });
            @Nonnull final PointSample measure = subject.measure(monitor).setRate(alpha);
            return new LineSearchPoint(measure, currentDirection.dot(measure.delta));
        }
    }.setDirectionType("OWL/QN");
}

Example 17 with TrainingMonitor

use of com.simiacryptus.mindseye.opt.TrainingMonitor in project MindsEye by SimiaCryptus.

the class QQN method orient.

@Override
public LineSearchCursor orient(@Nonnull final Trainable subject, @Nonnull final PointSample origin, @Nonnull final TrainingMonitor monitor) {
    inner.addToHistory(origin, monitor);
    final SimpleLineSearchCursor lbfgsCursor = inner.orient(subject, origin, monitor);
    final DeltaSet<Layer> lbfgs = lbfgsCursor.direction;
    @Nonnull final DeltaSet<Layer> gd = origin.delta.scale(-1.0);
    final double lbfgsMag = lbfgs.getMagnitude();
    final double gdMag = gd.getMagnitude();
    if (Math.abs(lbfgsMag - gdMag) / (lbfgsMag + gdMag) > 1e-2) {
        @Nonnull final DeltaSet<Layer> scaledGradient = gd.scale(lbfgsMag / gdMag);
        monitor.log(String.format("Returning Quadratic Cursor %s GD, %s QN", gdMag, lbfgsMag));
        gd.freeRef();
        return new LineSearchCursorBase() {

            @Nonnull
            @Override
            public CharSequence getDirectionType() {
                return CURSOR_NAME;
            }

            @Override
            public DeltaSet<Layer> position(final double t) {
                if (!Double.isFinite(t))
                    throw new IllegalArgumentException();
                return scaledGradient.scale(t - t * t).add(lbfgs.scale(t * t));
            }

            @Override
            public void reset() {
                lbfgsCursor.reset();
            }

            @Nonnull
            @Override
            public LineSearchPoint step(final double t, @Nonnull final TrainingMonitor monitor) {
                if (!Double.isFinite(t))
                    throw new IllegalArgumentException();
                reset();
                position(t).accumulate(1);
                @Nonnull final PointSample sample = subject.measure(monitor).setRate(t);
                // monitor.log(String.format("delta buffers %d %d %d %d %d", sample.delta.apply.size(), origin.delta.apply.size(), lbfgs.apply.size(), gd.apply.size(), scaledGradient.apply.size()));
                inner.addToHistory(sample, monitor);
                @Nonnull final DeltaSet<Layer> tangent = scaledGradient.scale(1 - 2 * t).add(lbfgs.scale(2 * t));
                return new LineSearchPoint(sample, tangent.dot(sample.delta));
            }

            @Override
            public void _free() {
                scaledGradient.freeRef();
                lbfgsCursor.freeRef();
            }
        };
    } else {
        gd.freeRef();
        return lbfgsCursor;
    }
}

Example 18 with TrainingMonitor

use of com.simiacryptus.mindseye.opt.TrainingMonitor 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 19 with TrainingMonitor

use of com.simiacryptus.mindseye.opt.TrainingMonitor 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();
}