Skip to content

Instantly share code, notes, and snippets.

@cogmission

cogmission/PALayer.java

Created August 26, 2016 02:41
Show Gist options
  • Save cogmission/552f5670b1114f1ead65da7ebdc4f59b to your computer and use it in GitHub Desktop.
Save cogmission/552f5670b1114f1ead65da7ebdc4f59b to your computer and use it in GitHub Desktop.
Download ZIP
Prediction Assisted Experiments
Raw
PALayer.java
/* ---------------------------------------------------------------------
* Numenta Platform for Intelligent Computing (NuPIC)
* Copyright (C) 2016, Numenta, Inc. Unless you have an agreement
* with Numenta, Inc., for a separate license for this software code, the
* following terms and conditions apply:
*
* This program is free software: you can redistribute it and/or modify
* it under the terms of the GNU Affero Public License version 3 as
* published by the Free Software Foundation.
*
* This program is distributed in the hope that it will be useful,
* but WITHOUT ANY WARRANTY; without even the implied warranty of
* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.
* See the GNU Affero Public License for more details.
*
* You should have received a copy of the GNU Affero Public License
* along with this program. If not, see http://www.gnu.org/licenses.
*
* http://numenta.org/licenses/
* ---------------------------------------------------------------------
*/
package org.numenta.nupic.network;
import org.numenta.nupic.ComputeCycle;
import org.numenta.nupic.Parameters;
import org.numenta.nupic.algorithms.Anomaly;
import org.numenta.nupic.algorithms.PASpatialPooler;
import org.numenta.nupic.algorithms.SpatialPooler;
import org.numenta.nupic.algorithms.TemporalMemory;
import org.numenta.nupic.encoders.MultiEncoder;
import org.numenta.nupic.model.Cell;
import org.numenta.nupic.model.Column;
/**
* Extension to Prediction-Assisted CLA
*
* PALayer is a Layer which can contain a PASpatialPooler, in which case (via Connections)
* it prepopulates the paSP's overlap vector with a depolarisation value derived from its
* TM's predictive cells. PASpatialPooler adds this vector to the overlaps calculated from the
* feedforward input before doing inhibition. This change pre-biases the paSP to favour columns
* with predictive cells.
*
* Full details at http://arxiv.org/abs/1509.08255
*
* @author David Ray
* @author Fergal Byrne
*/
@SuppressWarnings("unchecked")
public class PALayer<T> extends Layer<T> {
private static final long serialVersionUID = 1L;
/** Set to 0.0 to default to parent behavior */
double paDepolarize = 1.0;
int[] feedForwardActiveColumns;
int verbosity = 0;
/**
* Constructs a new {@code PALayer} which resides in the specified
* {@link Network}
*
* @param n the parent {@link Network}
*/
public PALayer(Network n) {
super(n);
}
/**
* Constructs a new {@code PALayer} which resides in the specified
* {@link Network} and uses the specified {@link Parameters}
*
* @param n the parent {@link Network}
* @param p the parameters object from which to obtain settings
*/
public PALayer(Network n, Parameters p) {
super(n, p);
}
/**
* Constructs a new {@code PALayer} which resides in the specified
* {@link Network} and uses the specified {@link Parameters}, with
* the specified name.
*
* @param name the name specified
* @param n the parent {@link Network}
* @param p the parameters object from which to obtain settings
*/
public PALayer(String name, Network n, Parameters p) {
super(name, n, p);
}
/**
* Manual method of creating a {@code Layer} and specifying its content.
*
* @param params the parameters object from which to obtain settings
* @param e an (optional) encoder providing input
* @param sp an (optional) SpatialPooler
* @param tm an (optional) {@link TemporalMemory}
* @param autoCreateClassifiers flag indicating whether to create {@link CLAClassifier}s
* @param a an (optional) {@link Anomaly} computer.
*/
public PALayer(Parameters params, MultiEncoder e, SpatialPooler sp, TemporalMemory tm, Boolean autoCreateClassifiers, Anomaly a) {
super(params, e, sp, tm, autoCreateClassifiers, a);
}
/**
* Returns paDepolarize (predictive assist per cell) for this {@link PALayer}
*
* @return
*/
public double getPADepolarize() {
return paDepolarize;
}
/**
* Sets paDepolarize {@code PALayer}
*
* @param pa
*/
public void setPADepolarize(double pa) {
paDepolarize = pa;
}
/**
* Returns verbosity level
*
* @return
*/
public int getVerbosity() {
return verbosity;
}
/**
* Sets verbosity level (0 for silent)
*
* @param verbosity
*/
public void setVerbosity(int verbosity) {
this.verbosity = verbosity;
}
/**
* Returns network
*
* @return
*/
public Network getParentNetwork() {
return parentNetwork;
}
/**
* Called internally to invoke the {@link SpatialPooler}
*
* @param input
* @return
*/
protected int[] spatialInput(int[] input) {
if(input == null) {
LOGGER.info("Layer ".concat(getName()).concat(" received null input"));
} else if(input.length < 1) {
LOGGER.info("Layer ".concat(getName()).concat(" received zero length bit vector"));
return input;
} else if(input.length > connections.getNumInputs()) {
if(verbosity > 0) {
System.out.println(input);
}
throw new IllegalArgumentException(String.format("Input size %d > SP's NumInputs %d",input.length, connections.getNumInputs()));
}
spatialPooler.compute(connections, input, feedForwardActiveColumns, sensor == null || sensor.getMetaInfo().isLearn(), isLearn);
return feedForwardActiveColumns;
}
/**
* Called internally to invoke the {@link TemporalMemory}
*
* @param input
* the current input vector
* @param mi
* the current input inference container
* @return
*/
protected int[] temporalInput(int[] input, ManualInput mi) {
int[] sdr = super.temporalInput(input, mi);
ComputeCycle cc = mi.computeCycle;
if(spatialPooler != null && spatialPooler instanceof PASpatialPooler) {
int boosted = 0;
double[] polarization = new double[connections.getNumColumns()];
for(Cell cell : cc.predictiveCells()) {
Column column = cell.getColumn();
if(polarization[column.getIndex()] == 0.0) {
boosted++;
}
polarization[column.getIndex()] += paDepolarize;
if(verbosity >= 2) {
System.out.println(String.format("[%d] = %d", column.getIndex(),(int)paDepolarize));
}
}
if(verbosity >= 1) {
System.out.println(String.format("boosted %d/%d columns", boosted, connections.getNumColumns()));
}
connections.setPAOverlaps(polarization);
}
return sdr;
}
}
Raw
PALayerTest.java
/* ---------------------------------------------------------------------
* Numenta Platform for Intelligent Computing (NuPIC)
* Copyright (C) 2014, Numenta, Inc. Unless you have an agreement
* with Numenta, Inc., for a separate license for this software code, the
* following terms and conditions apply:
*
* This program is free software: you can redistribute it and/or modify
* it under the terms of the GNU Affero Public License version 3 as
* published by the Free Software Foundation.
*
* This program is distributed in the hope that it will be useful,
* but WITHOUT ANY WARRANTY; without even the implied warranty of
* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.
* See the GNU Affero Public License for more details.
*
* You should have received a copy of the GNU Affero Public License
* along with this program. If not, see http://www.gnu.org/licenses.
*
* http://numenta.org/licenses/
* ---------------------------------------------------------------------
*/
package org.numenta.nupic.network;
import static org.junit.Assert.assertEquals;
import static org.junit.Assert.assertFalse;
import static org.junit.Assert.assertNotNull;
import static org.junit.Assert.assertTrue;
import static org.junit.Assert.fail;
import static org.numenta.nupic.algorithms.Anomaly.KEY_MODE;
import static org.numenta.nupic.algorithms.Anomaly.KEY_USE_MOVING_AVG;
import static org.numenta.nupic.algorithms.Anomaly.KEY_WINDOW_SIZE;
import java.io.File;
import java.util.ArrayList;
import java.util.Arrays;
import java.util.HashMap;
import java.util.List;
import java.util.Map;
import org.junit.Ignore;
import org.numenta.nupic.Parameters;
import org.numenta.nupic.Parameters.KEY;
import org.numenta.nupic.SDR;
import org.numenta.nupic.algorithms.Anomaly;
import org.numenta.nupic.algorithms.Anomaly.Mode;
import org.numenta.nupic.algorithms.CLAClassifier;
import org.numenta.nupic.algorithms.TemporalMemory;
import org.numenta.nupic.algorithms.PASpatialPooler;
import org.numenta.nupic.datagen.ResourceLocator;
import org.numenta.nupic.encoders.MultiEncoder;
import org.numenta.nupic.network.sensor.FileSensor;
import org.numenta.nupic.network.sensor.HTMSensor;
import org.numenta.nupic.network.sensor.ObservableSensor;
import org.numenta.nupic.network.sensor.Publisher;
import org.numenta.nupic.network.sensor.Sensor;
import org.numenta.nupic.network.sensor.SensorParams;
import org.numenta.nupic.network.sensor.SensorParams.Keys;
import org.numenta.nupic.util.ArrayUtils;
import org.numenta.nupic.util.MersenneTwister;
import rx.Observable;
import rx.Observer;
import rx.Subscriber;
import rx.functions.Func1;
/**
* Tests the "heart and soul" of the Network API
*
* @author DavidRay
*
*/
public class PALayerTest {
/** Total used for spatial pooler priming tests */
private int TOTAL = 0;
@Ignore
public void testMasking() {
byte algo_content_mask = 0;
// -- Build up mask
algo_content_mask |= PALayer.CLA_CLASSIFIER;
assertEquals(4, algo_content_mask);
algo_content_mask |= PALayer.SPATIAL_POOLER;
assertEquals(5, algo_content_mask);
algo_content_mask |= PALayer.TEMPORAL_MEMORY;
assertEquals(7, algo_content_mask);
algo_content_mask |= PALayer.ANOMALY_COMPUTER;
assertEquals(15, algo_content_mask);
// -- Now Peel Off
algo_content_mask ^= PALayer.ANOMALY_COMPUTER;
assertEquals(7, algo_content_mask);
assertEquals(0, algo_content_mask & PALayer.ANOMALY_COMPUTER);
assertEquals(2, algo_content_mask & PALayer.TEMPORAL_MEMORY);
algo_content_mask ^= PALayer.TEMPORAL_MEMORY;
assertEquals(5, algo_content_mask);
algo_content_mask ^= PALayer.SPATIAL_POOLER;
assertEquals(4, algo_content_mask);
algo_content_mask ^= PALayer.CLA_CLASSIFIER;
assertEquals(0, algo_content_mask);
}
@Ignore
public void testGetAllValues() {
Parameters p = NetworkTestHarness.getParameters().copy();
p = p.union(NetworkTestHarness.getDayDemoTestEncoderParams());
p.set(KEY.RANDOM, new MersenneTwister(42));
MultiEncoder me = MultiEncoder.builder().name("").build();
Layer<Map<String, Object>> l = new PALayer<Map<String, Object>>(p, me, new PASpatialPooler(), new TemporalMemory(), Boolean.TRUE, null);
// Test that we get the expected exception if there hasn't been any processing.
try {
l.getAllValues("dayOfWeek", 1);
fail();
}catch(Exception e) {
assertEquals("Predictions not available. Either classifiers unspecified or inferencing has not yet begun.", e.getMessage());
}
l.subscribe(new Observer<Inference>() {
@Override public void onCompleted() {}
@Override public void onError(Throwable e) { System.out.println("error: " + e.getMessage()); e.printStackTrace();}
@Override
public void onNext(Inference i) {
assertNotNull(i);
assertEquals(36, i.getSDR().length);
}
});
// Now push some fake data through so that "onNext" is called above
Map<String, Object> multiInput = new HashMap<>();
multiInput.put("dayOfWeek", 0.0);
l.compute(multiInput);
Object[] values = l.getAllValues("dayOfWeek", 1);
assertNotNull(values);
assertTrue(values.length == 1);
assertEquals(0.0D, values[0]);
}
@Ignore
public void testConstructors() {
Parameters p = NetworkTestHarness.getParameters().copy();
Network n = new Network("test",p);
PALayer<?> l = new PALayer<>(n, p);
assertTrue(n == l.getParentNetwork());
}
@Ignore
public void testPolariseAndVerbosity() {
Parameters p = NetworkTestHarness.getParameters().copy();
Network n = new Network("test",p);
PALayer<?> l = new PALayer<>(n, p);
l.setPADepolarize(1.0);
assertTrue(1.0 == l.getPADepolarize());
l.setVerbosity(1);
assertTrue(1 == l.getVerbosity());
}
@Ignore
public void testSpatialInput() {
Parameters p = NetworkTestHarness.getParameters().copy();
p.set(KEY.INPUT_DIMENSIONS, new int[]{3});
Network n = new Network("test",p);
PALayer<?> l = new PALayer<>(n, p);
l.setVerbosity(2);
int[] result;
try {
result = l.spatialInput(null);
} catch(Exception e) {
assertTrue(e instanceof NullPointerException);
}
result = l.spatialInput(new int[] {});
assertTrue(result.length == 0);
try {
result = l.spatialInput(new int[] {1, 0, 1, 0, 0});
} catch(Exception e) {
assertTrue(e instanceof IllegalArgumentException);
}
}
@Ignore
public void testResetMethod() {
Parameters p = NetworkTestHarness.getParameters().copy();
Layer<?> l = Network.createPALayer("l1", p).add(new TemporalMemory());
try {
l.reset();
assertTrue(l.hasTemporalMemory());
}catch(Exception e) {
fail();
}
Layer<?> l2 = Network.createPALayer("l2", p).add(new PASpatialPooler());
try {
l2.reset();
assertFalse(l2.hasTemporalMemory());
}catch(Exception e) {
fail();
}
}
@Ignore
public void testResetRecordNum() {
Parameters p = NetworkTestHarness.getParameters().copy();
@SuppressWarnings("unchecked")
Layer<int[]> l = (PALayer<int[]>)Network.createPALayer("l1", p).add(new TemporalMemory());
l.subscribe(new Observer<Inference>() {
@Override public void onCompleted() {}
@Override public void onError(Throwable e) { e.printStackTrace(); }
@Override public void onNext(Inference output) {
// System.out.println("output = " + Arrays.toString(output.getSDR()));
}
});
l.compute(new int[] { 2,3,4 });
l.compute(new int[] { 2,3,4 });
assertEquals(1, l.getRecordNum());
l.resetRecordNum();
assertEquals(0, l.getRecordNum());
}
boolean isHalted = false;
@Ignore
public void testHalt() {
Sensor<File> sensor = Sensor.create(
FileSensor::create,
SensorParams.create(
Keys::path, "", ResourceLocator.path("rec-center-hourly-small.csv")));
Parameters p = NetworkTestHarness.getParameters().copy();
p = p.union(NetworkTestHarness.getHotGymTestEncoderParams());
p.set(KEY.RANDOM, new MersenneTwister(42));
p.set(KEY.AUTO_CLASSIFY, Boolean.TRUE);
HTMSensor<File> htmSensor = (HTMSensor<File>)sensor;
Network n = Network.create("test network", p);
final Layer<int[]> l = new PALayer<>(n);
l.add(htmSensor);
l.subscribe(new Observer<Inference>() {
@Override public void onCompleted() {
assertTrue(l.isHalted());
isHalted = true;
}
@Override public void onError(Throwable e) { e.printStackTrace(); }
@Override public void onNext(Inference output) {}
});
l.start();
try {
l.halt();
l.getLayerThread().join();
assertTrue(isHalted);
}catch(Exception e) {
e.printStackTrace();
fail();
}
}
int trueCount = 0;
@Ignore
public void testReset() {
Sensor<File> sensor = Sensor.create(
FileSensor::create,
SensorParams.create(
Keys::path, "", ResourceLocator.path("rec-center-hourly-4reset.csv")));
Parameters p = NetworkTestHarness.getParameters().copy();
p = p.union(NetworkTestHarness.getHotGymTestEncoderParams());
p.set(KEY.RANDOM, new MersenneTwister(42));
p.set(KEY.AUTO_CLASSIFY, Boolean.TRUE);
HTMSensor<File> htmSensor = (HTMSensor<File>)sensor;
Network n = Network.create("test network", p);
final Layer<int[]> l = new PALayer<>(n);
l.add(htmSensor);
l.subscribe(new Observer<Inference>() {
@Override public void onCompleted() {}
@Override public void onError(Throwable e) { e.printStackTrace(); }
@Override public void onNext(Inference output) {
if(l.getSensor().getMetaInfo().isReset()) {
trueCount++;
}
}
});
l.start();
try {
l.getLayerThread().join();
assertEquals(3, trueCount);
}catch(Exception e) {
e.printStackTrace();
fail();
}
}
int seqResetCount = 0;
@Ignore
public void testSequenceChangeReset() {
Sensor<File> sensor = Sensor.create(
FileSensor::create,
SensorParams.create(
Keys::path, "", ResourceLocator.path("rec-center-hourly-4seqReset.csv")));
Parameters p = NetworkTestHarness.getParameters().copy();
p = p.union(NetworkTestHarness.getHotGymTestEncoderParams());
p.set(KEY.RANDOM, new MersenneTwister(42));
p.set(KEY.AUTO_CLASSIFY, Boolean.TRUE);
HTMSensor<File> htmSensor = (HTMSensor<File>)sensor;
Network n = Network.create("test network", p);
final Layer<int[]> l = new PALayer<>(n);
l.add(htmSensor);
l.subscribe(new Observer<Inference>() {
@Override public void onCompleted() {}
@Override public void onError(Throwable e) { e.printStackTrace(); }
@Override public void onNext(Inference output) {
if(l.getSensor().getMetaInfo().isReset()) {
seqResetCount++;
}
}
});
l.start();
try {
l.getLayerThread().join();
assertEquals(3, seqResetCount);
}catch(Exception e) {
e.printStackTrace();
fail();
}
}
@Ignore
public void testLayerWithObservableInput() {
Publisher manual = Publisher.builder()
.addHeader("timestamp,consumption")
.addHeader("datetime,float")
.addHeader("B")
.build();
Sensor<ObservableSensor<String[]>> sensor = Sensor.create(
ObservableSensor::create,
SensorParams.create(
Keys::obs, new Object[] {"name", manual}));
Parameters p = NetworkTestHarness.getParameters().copy();
p = p.union(NetworkTestHarness.getHotGymTestEncoderParams());
p.set(KEY.RANDOM, new MersenneTwister(42));
p.set(KEY.AUTO_CLASSIFY, Boolean.TRUE);
HTMSensor<ObservableSensor<String[]>> htmSensor = (HTMSensor<ObservableSensor<String[]>>)sensor;
Network n = Network.create("test network", p);
final Layer<int[]> l = new PALayer<>(n);
l.add(htmSensor);
l.subscribe(new Observer<Inference>() {
int idx = 0;
@Override public void onCompleted() {}
@Override public void onError(Throwable e) { e.printStackTrace(); }
@Override public void onNext(Inference output) {
switch(idx) {
case 0: assertEquals("[0, 0, 1, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 0, 0, 1, 1, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 1]", Arrays.toString(output.getSDR()));
break;
case 1: assertEquals("[0, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 0, 0, 0, 0, 0, 1, 0, 0, 1, 1, 1, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1]", Arrays.toString(output.getSDR()));
break;
case 2: assertEquals("[1, 0, 0, 0, 0, 0, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 0, 0, 0, 0, 1, 0, 0, 1, 1, 1, 1, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0]", Arrays.toString(output.getSDR()));
break;
}
++idx;
}
});
l.start();
try {
String[] entries = {
"7/2/10 0:00,21.2",
"7/2/10 1:00,34.0",
"7/2/10 2:00,40.4",
};
// Send inputs through the observable
for(String s : entries) {
manual.onNext(s);
}
Thread.sleep(100);
}catch(Exception e) {
e.printStackTrace();
fail();
}
}
@Ignore
public void testLayerWithObservableInputIntegerArray() {
Publisher manual = Publisher.builder()
.addHeader("sdr_in")
.addHeader("darr")
.addHeader("B")
.build();
Sensor<ObservableSensor<String[]>> sensor = Sensor.create(
ObservableSensor::create,
SensorParams.create(
Keys::obs, new Object[] {"name", manual}));
Parameters p = Parameters.getAllDefaultParameters();
p = p.union(getArrayTestParams());
p.set(KEY.RANDOM, new MersenneTwister(42));
HTMSensor<ObservableSensor<String[]>> htmSensor = (HTMSensor<ObservableSensor<String[]>>)sensor;
Network n = Network.create("test network", p);
final Layer<int[]> l = new PALayer<>(n);
l.add(htmSensor);
String input = "[0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 0, 0, 0, 0, 0, 0, 0, 0, 1, 1, 0, 0, 0, 0, 0, 0, 0, 1, 1, 1, 0, 0, 0, 0, 0, 0, "
+ "1, 1, 1, 1, 0, 0, 0, 0, 0, 1, 1, 1, 1, 1, 0, 0, 0, 0, 1, 1, 1, 1, 1, 1, 0, 0, 0, 0, 1, 1, 1, 1, 1, 0, 0, 0, 0, 1, 1, 1, 1, 1, 1, 0, 0, 0, "
+ "0, 1, 1, 1, 1, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, "
+ "0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, "
+ "0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, "
+ "0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, "
+ "1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 0, 0, 0, 0, 0, 0, 0, 0, "
+ "0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 1, 1, 1, 1, 1, 1, 1, "
+ "1, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, "
+ "0, 0, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 0, 0, 0, 0, 0, 0, "
+ "0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 1, 1, 1, 1, 1, "
+ "1, 1, 1, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, "
+ "0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, "
+ "0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, "
+ "0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, "
+ "0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, "
+ "0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, "
+ "0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, "
+ "0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, "
+ "0, 0, 0, 0, 0, 0, 0, 0, 0, 0]";
l.subscribe(new Observer<Inference>() {
@Override public void onCompleted() {}
@Override public void onError(Throwable e) { e.printStackTrace(); }
@Override public void onNext(Inference output) {
assertEquals(input, Arrays.toString((int[])output.getLayerInput()));
}
});
l.start();
try {
String[] entries = {
input
};
// Send inputs through the observable
for(String s : entries) {
manual.onNext(s);
manual.onComplete();
}
l.getLayerThread().join();
}catch(Exception e) {
e.printStackTrace();
fail();
}
}
@Ignore
public void testLayerWithGenericObservable() {
Parameters p = NetworkTestHarness.getParameters().copy();
p.set(KEY.RANDOM, new MersenneTwister(42));
int[][] inputs = new int[7][8];
inputs[0] = new int[] { 1, 1, 0, 0, 0, 0, 0, 1 };
inputs[1] = new int[] { 1, 1, 1, 0, 0, 0, 0, 0 };
inputs[2] = new int[] { 0, 1, 1, 1, 0, 0, 0, 0 };
inputs[3] = new int[] { 0, 0, 1, 1, 1, 0, 0, 0 };
inputs[4] = new int[] { 0, 0, 0, 1, 1, 1, 0, 0 };
inputs[5] = new int[] { 0, 0, 0, 0, 1, 1, 1, 0 };
inputs[6] = new int[] { 0, 0, 0, 0, 0, 1, 1, 1 };
final int[] expected0 = new int[] { 0, 1, 0, 0, 0, 1, 0, 1, 0, 0, 0, 1, 1, 0, 0, 0, 1, 0, 0, 0 };
final int[] expected1 = new int[] { 0, 1, 0, 0, 0, 1, 0, 0, 0, 0, 0, 1, 1, 1, 0, 0, 0, 0, 0, 0 };
Func1<ManualInput, ManualInput> addedFunc = l -> {
return l.customObject("Interposed: " + Arrays.toString(l.getSDR()));
};
Network n = Network.create("Generic Test", p)
.add(Network.createRegion("R1")
.add(Network.createPALayer("L1", p)
.add(addedFunc)
.add(new PASpatialPooler())));
@SuppressWarnings("unchecked")
Layer<int[]> l = (PALayer<int[]>)n.lookup("R1").lookup("L1");
l.subscribe(new Observer<Inference>() {
int test = 0;
@Override public void onCompleted() {}
@Override public void onError(Throwable e) { e.printStackTrace(); }
@Override
public void onNext(Inference i) {
if(test == 0) {
assertTrue(Arrays.equals(expected0, i.getSDR()));//
assertEquals("Interposed: [0, 1, 0, 0, 0, 1, 0, 1, 0, 0, 0, 1, 1, 0, 0, 0, 1, 0, 0, 0]", i.getCustomObject());
}
if(test == 1) {
assertTrue(Arrays.equals(expected1, i.getSDR()));
assertEquals("Interposed: [0, 1, 0, 0, 0, 1, 0, 0, 0, 0, 0, 1, 1, 1, 0, 0, 0, 0, 0, 0]", i.getCustomObject());
}
++test;
}
});
// SHOULD RECEIVE BOTH
// Now push some fake data through so that "onNext" is called above
l.compute(inputs[0]);
l.compute(inputs[1]);
}
@Ignore
public void testBasicSetupEncoder_UsingSubscribe() {
Parameters p = NetworkTestHarness.getParameters().copy();
p = p.union(NetworkTestHarness.getDayDemoTestEncoderParams());
p.set(KEY.RANDOM, new MersenneTwister(42));
MultiEncoder me = MultiEncoder.builder().name("").build();
Layer<Map<String, Object>> l = new PALayer<>(p, me, null, null, null, null);
final int[][] expected = new int[7][8];
expected[0] = new int[] { 1, 1, 0, 0, 0, 0, 0, 1 };
expected[1] = new int[] { 1, 1, 1, 0, 0, 0, 0, 0 };
expected[2] = new int[] { 0, 1, 1, 1, 0, 0, 0, 0 };
expected[3] = new int[] { 0, 0, 1, 1, 1, 0, 0, 0 };
expected[4] = new int[] { 0, 0, 0, 1, 1, 1, 0, 0 };
expected[5] = new int[] { 0, 0, 0, 0, 1, 1, 1, 0 };
expected[6] = new int[] { 0, 0, 0, 0, 0, 1, 1, 1 };
l.subscribe(new Observer<Inference>() {
int seq = 0;
@Override public void onCompleted() {}
@Override public void onError(Throwable e) { e.printStackTrace(); }
@Override public void onNext(Inference output) {
assertTrue(Arrays.equals(expected[seq++], output.getSDR()));
}
});
Map<String, Object> inputs = new HashMap<String, Object>();
for(double i = 0;i < 7;i++) {
inputs.put("dayOfWeek", i);
l.compute(inputs);
}
}
@Ignore
public void testBasicSetupEncoder_UsingObserve() {
Parameters p = NetworkTestHarness.getParameters().copy();
p = p.union(NetworkTestHarness.getDayDemoTestEncoderParams());
p.set(KEY.RANDOM, new MersenneTwister(42));
MultiEncoder me = MultiEncoder.builder().name("").build();
Layer<Map<String, Object>> l = new PALayer<>(p, me, null, null, null, null);
final int[][] expected = new int[7][8];
expected[0] = new int[] { 1, 1, 0, 0, 0, 0, 0, 1 };
expected[1] = new int[] { 1, 1, 1, 0, 0, 0, 0, 0 };
expected[2] = new int[] { 0, 1, 1, 1, 0, 0, 0, 0 };
expected[3] = new int[] { 0, 0, 1, 1, 1, 0, 0, 0 };
expected[4] = new int[] { 0, 0, 0, 1, 1, 1, 0, 0 };
expected[5] = new int[] { 0, 0, 0, 0, 1, 1, 1, 0 };
expected[6] = new int[] { 0, 0, 0, 0, 0, 1, 1, 1 };
Observable<Inference> o = l.observe();
o.subscribe(new Observer<Inference>() {
int seq = 0;
@Override public void onCompleted() {}
@Override public void onError(Throwable e) { e.printStackTrace(); }
@Override public void onNext(Inference output) {
assertTrue(Arrays.equals(expected[seq++], output.getSDR()));
}
});
Map<String, Object> inputs = new HashMap<String, Object>();
for(double i = 0;i < 7;i++) {
inputs.put("dayOfWeek", i);
l.compute(inputs);
}
}
@Ignore
public void testBasicSetupEncoder_AUTO_MODE() {
Sensor<File> sensor = Sensor.create(
FileSensor::create,
SensorParams.create(
Keys::path, "", ResourceLocator.path("rec-center-hourly-small.csv")));
Parameters p = NetworkTestHarness.getParameters().copy();
p = p.union(NetworkTestHarness.getHotGymTestEncoderParams());
p.set(KEY.RANDOM, new MersenneTwister(42));
p.set(KEY.AUTO_CLASSIFY, Boolean.TRUE);
HTMSensor<File> htmSensor = (HTMSensor<File>)sensor;
Network n = Network.create("test network", p);
Layer<int[]> l = new PALayer<>(n);
l.add(htmSensor);
final int[][] expected = new int[][] {
{ 0, 0, 1, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 0, 0, 1, 1, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 1},
{ 0, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 0, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 0, 0, 1, 1, 1, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1},
{ 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 0, 0, 0, 0, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 0, 0, 0, 0, 1, 0, 0, 1, 1, 1, 1, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0},
{ 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 0, 0, 0, 0, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 0, 0, 0, 0, 1, 0, 0, 0, 1, 1, 1, 1, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0},
{ 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 0, 0, 0, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 0, 0, 0, 0, 1, 0, 0, 0, 0, 0, 1, 1, 1, 1, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0},
{ 0, 0, 1, 0, 1, 0, 0, 0, 0, 0, 0, 0, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 0, 0, 0, 0, 0, 0, 1, 1, 1, 1, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0},
{ 0, 1, 0, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 0, 0, 0, 0, 0, 0, 0, 1, 0, 0, 0, 0, 0, 0, 0, 1, 1, 1, 1, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0},
{ 0, 0, 1, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 0, 0, 0, 0, 0, 0, 0, 0, 1, 1, 1, 1, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0},
{ 0, 0, 0, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 1, 1, 1, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0},
{ 0, 0, 0, 0, 0, 1, 0, 0, 1, 0, 0, 0, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 1, 1, 1, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0},
{ 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 0, 1, 0, 0, 0, 0, 0, 0, 0, 0, 1, 0, 0, 0, 0, 0, 0, 0, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 1, 1, 1, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0},
{ 0, 0, 0, 0, 1, 0, 0, 0, 0, 0, 0, 0, 1, 0, 0, 0, 0, 0, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 1, 1, 1, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0},
{ 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 0, 0, 0, 0, 0, 0, 0, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 1, 1, 1, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0},
{ 0, 0, 0, 1, 0, 0, 0, 0, 0, 1, 0, 0, 0, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 1, 1, 1, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0},
{ 0, 0, 0, 0, 1, 0, 0, 0, 0, 0, 1, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 1, 1, 1, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0},
{ 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 0, 0, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 0, 0, 0, 0, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 1, 1, 1, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0},
{ 0, 0, 0, 0, 0, 0, 1, 0, 0, 0, 0, 1, 0, 0, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 1, 1, 1, 1, 0, 0, 0, 0, 0, 0, 0}
};
///////////////////////////////////////////////////////
// Test with 2 subscribers //
///////////////////////////////////////////////////////
l.observe().subscribe(new Observer<Inference>() {
int seq = 0;
@Override public void onCompleted() {}
@Override public void onError(Throwable e) { e.printStackTrace(); }
@Override public void onNext(Inference output) {
// System.out.println(" seq = " + seq + ", recNum = " + output.getRecordNum() + ", expected = " + Arrays.toString(expected[seq]));
// System.out.println(" seq = " + seq + ", recNum = " + output.getRecordNum() + ", output = " + Arrays.toString(output.getSDR()));
assertTrue(Arrays.equals(expected[seq], output.getSDR()));
seq++;
}
});
l.observe().subscribe(new Observer<Inference>() {
int seq2 = 0;
@Override public void onCompleted() {}
@Override public void onError(Throwable e) { e.printStackTrace(); }
@Override public void onNext(Inference output) {
// System.out.println(" seq = " + seq2 + ", recNum = " + output.getRecordNum() + ", expected = " + Arrays.toString(expected[seq2]));
// System.out.println(" seq = " + seq2 + ", recNum = " + output.getRecordNum() + ", output = " + Arrays.toString(output.getSDR()));
assertTrue(Arrays.equals(expected[seq2], output.getSDR()));
seq2++;
}
});
l.start();
try {
l.getLayerThread().join();
}catch(Exception e) {
e.printStackTrace();
}
}
/**
* Temporary test to test basic sequence mechanisms
*/
@Ignore
public void testBasicSetup_SpatialPooler_MANUAL_MODE() {
Parameters p = NetworkTestHarness.getParameters().copy();
p.set(KEY.RANDOM, new MersenneTwister(42));
int[][] inputs = new int[7][8];
inputs[0] = new int[] { 1, 1, 0, 0, 0, 0, 0, 1 };
inputs[1] = new int[] { 1, 1, 1, 0, 0, 0, 0, 0 };
inputs[2] = new int[] { 0, 1, 1, 1, 0, 0, 0, 0 };
inputs[3] = new int[] { 0, 0, 1, 1, 1, 0, 0, 0 };
inputs[4] = new int[] { 0, 0, 0, 1, 1, 1, 0, 0 };
inputs[5] = new int[] { 0, 0, 0, 0, 1, 1, 1, 0 };
inputs[6] = new int[] { 0, 0, 0, 0, 0, 1, 1, 1 };
final int[] expected0 = new int[] { 0, 1, 0, 0, 0, 1, 0, 1, 0, 0, 0, 1, 1, 0, 0, 0, 1, 0, 0, 0 };
final int[] expected1 = new int[] { 0, 1, 0, 0, 0, 1, 0, 0, 0, 0, 0, 1, 1, 1, 0, 0, 0, 0, 0, 0 };
Layer<int[]> l = new PALayer<>(p, null, new PASpatialPooler(), null, null, null);
l.subscribe(new Observer<Inference>() {
int test = 0;
@Override public void onCompleted() {}
@Override public void onError(Throwable e) { e.printStackTrace(); }
@Override
public void onNext(Inference spatialPoolerOutput) {
if(test == 0) {
assertTrue(Arrays.equals(expected0, spatialPoolerOutput.getSDR()));
}
if(test == 1) {
assertTrue(Arrays.equals(expected1, spatialPoolerOutput.getSDR()));
}
++test;
}
});
// Now push some fake data through so that "onNext" is called above
l.compute(inputs[0]);
l.compute(inputs[1]);
}
/**
* Temporary test to test basic sequence mechanisms
*/
@Ignore
public void testBasicSetup_SpatialPooler_AUTO_MODE() {
Sensor<File> sensor = Sensor.create(
FileSensor::create,
SensorParams.create(
Keys::path, "", ResourceLocator.path("days-of-week.csv")));
Parameters p = NetworkTestHarness.getParameters().copy();
p = p.union(NetworkTestHarness.getDayDemoTestEncoderParams());
p.set(KEY.RANDOM, new MersenneTwister(42));
p.set(KEY.AUTO_CLASSIFY, Boolean.TRUE);
HTMSensor<File> htmSensor = (HTMSensor<File>)sensor;
Network n = Network.create("test network", p);
Layer<int[]> l = new PALayer<>(n);
l.add(htmSensor).add(new PASpatialPooler());
final int[] expected0 = new int[] { 0, 1, 0, 0, 0, 1, 0, 1, 0, 0, 0, 1, 1, 0, 0, 0, 1, 0, 0, 0 };
final int[] expected1 = new int[] { 0, 1, 0, 0, 0, 1, 0, 0, 0, 0, 0, 1, 1, 1, 0, 0, 0, 0, 0, 0 };
l.subscribe(new Observer<Inference>() {
int test = 0;
@Override public void onCompleted() {}
@Override public void onError(Throwable e) { e.printStackTrace(); }
@Override
public void onNext(Inference spatialPoolerOutput) {
if(test == 0) {
assertTrue(Arrays.equals(expected0, spatialPoolerOutput.getSDR()));
}
if(test == 1) {
assertTrue(Arrays.equals(expected1, spatialPoolerOutput.getSDR()));
}
++test;
}
});
l.start();
try {
l.getLayerThread().join();
}catch(Exception e) {
e.printStackTrace();
}
}
/**
* Temporary test to test basic sequence mechanisms
*/
@Ignore
public void testBasicSetup_TemporalMemory_MANUAL_MODE() {
Parameters p = NetworkTestHarness.getParameters().copy();
p.set(KEY.RANDOM, new MersenneTwister(42));
final int[] input1 = new int[] { 0, 1, 0, 0, 0, 1, 0, 1, 0, 0, 0, 1, 1, 0, 0, 0, 1, 0, 0, 0 };
final int[] input2 = new int[] { 0, 1, 0, 0, 0, 1, 0, 0, 0, 0, 0, 1, 1, 1, 0, 0, 0, 0, 0, 0 };
final int[] input3 = new int[] { 0, 0, 1, 1, 0, 0, 0, 0, 0, 0, 0, 1, 1, 1, 1, 0, 0, 0, 0, 0 };
final int[] input4 = new int[] { 0, 0, 1, 1, 0, 0, 0, 0, 1, 1, 0, 0, 1, 0, 0, 0, 0, 1, 1, 0 };
final int[] input5 = new int[] { 0, 0, 1, 1, 0, 0, 0, 0, 1, 0, 0, 0, 1, 0, 0, 0, 0, 1, 1, 0 };
final int[] input6 = new int[] { 0, 0, 1, 0, 0, 0, 0, 1, 1, 1, 0, 0, 0, 0, 0, 0, 0, 1, 1, 1 };
final int[] input7 = new int[] { 0, 1, 0, 0, 0, 0, 0, 1, 1, 1, 0, 0, 0, 0, 0, 0, 0, 1, 1, 0 };
final int[][] inputs = { input1, input2, input3, input4, input5, input6, input7 };
Layer<int[]> l = new PALayer<>(p, null, null, new TemporalMemory(), null, null);
int timeUntilStable = 600;
l.subscribe(new Observer<Inference>() {
int test = 0;
int seq = 0;
@Override public void onCompleted() {}
@Override public void onError(Throwable e) { e.printStackTrace(); }
@Override
public void onNext(Inference output) {
if(seq / 7 >= timeUntilStable) {
// System.out.println("seq: " + (seq) + " --> " + (test) + " output = " + Arrays.toString(output.getSDR()) +
// ", \t\t\t\t cols = " + Arrays.toString(SDR.asColumnIndices(output.getSDR(), l.getConnections().getCellsPerColumn())));
assertTrue(output.getSDR().length >= 8);
}
if(test == 6) test = 0;
else test++;
seq++;
}
});
// Now push some warm up data through so that "onNext" is called above
for(int j = 0;j < timeUntilStable;j++) {
for(int i = 0;i < inputs.length;i++) {
l.compute(inputs[i]);
}
}
for(int j = 0;j < 2;j++) {
for(int i = 0;i < inputs.length;i++) {
l.compute(inputs[i]);
}
}
}
@Ignore
public void testBasicSetup_SPandTM() {
Parameters p = NetworkTestHarness.getParameters().copy();
p.set(KEY.RANDOM, new MersenneTwister(42));
int[][] inputs = new int[7][8];
inputs[0] = new int[] { 1, 1, 0, 0, 0, 0, 0, 1 };
inputs[1] = new int[] { 1, 1, 1, 0, 0, 0, 0, 0 };
inputs[2] = new int[] { 0, 1, 1, 1, 0, 0, 0, 0 };
inputs[3] = new int[] { 0, 0, 1, 1, 1, 0, 0, 0 };
inputs[4] = new int[] { 0, 0, 0, 1, 1, 1, 0, 0 };
inputs[5] = new int[] { 0, 0, 0, 0, 1, 1, 1, 0 };
inputs[6] = new int[] { 0, 0, 0, 0, 0, 1, 1, 1 };
Layer<int[]> l = new PALayer<>(p, null, new PASpatialPooler(), new TemporalMemory(), null, null);
l.subscribe(new Observer<Inference>() {
@Override public void onCompleted() {}
@Override public void onError(Throwable e) {}
@Override
public void onNext(Inference i) {
assertNotNull(i);
assertEquals(0, i.getSDR().length);
}
});
// Now push some fake data through so that "onNext" is called above
l.compute(inputs[0]);
l.compute(inputs[1]);
}
@Ignore
public void testSpatialPoolerPrimerDelay() {
Parameters p = NetworkTestHarness.getParameters().copy();
p.set(KEY.RANDOM, new MersenneTwister(42));
int[][] inputs = new int[7][8];
inputs[0] = new int[] { 1, 1, 0, 0, 0, 0, 0, 1 };
inputs[1] = new int[] { 1, 1, 1, 0, 0, 0, 0, 0 };
inputs[2] = new int[] { 0, 1, 1, 1, 0, 0, 0, 0 };
inputs[3] = new int[] { 0, 0, 1, 1, 1, 0, 0, 0 };
inputs[4] = new int[] { 0, 0, 0, 1, 1, 1, 0, 0 };
inputs[5] = new int[] { 0, 0, 0, 0, 1, 1, 1, 0 };
inputs[6] = new int[] { 0, 0, 0, 0, 0, 1, 1, 1 };
final int[] expected0 = new int[] { 0, 1, 0, 0, 0, 1, 0, 1, 0, 0, 0, 1, 1, 0, 0, 0, 1, 0, 0, 0 };
final int[] expected1 = new int[] { 0, 1, 0, 0, 0, 1, 0, 0, 0, 0, 0, 1, 1, 1, 0, 0, 0, 0, 0, 0 };
// First test without prime directive :-P
Layer<int[]> l = new PALayer<>(p, null, new PASpatialPooler(), null, null, null);
l.subscribe(new Observer<Inference>() {
int test = 0;
@Override public void onCompleted() {}
@Override public void onError(Throwable e) { e.printStackTrace(); }
@Override
public void onNext(Inference i) {
if(test == 0) {
assertTrue(Arrays.equals(expected0, i.getSDR()));
}
if(test == 1) {
assertTrue(Arrays.equals(expected1, i.getSDR()));
}
++test;
}
});
// SHOULD RECEIVE BOTH
// Now push some fake data through so that "onNext" is called above
l.compute(inputs[0]);
l.compute(inputs[1]);
// --------------------------------------------------------------------------------------------
// NOW TEST WITH prime directive
p.set(KEY.RANDOM, new MersenneTwister(42)); // due to static RNG we have to reset the sequence
p.set(KEY.SP_PRIMER_DELAY, 1);
Layer<int[]> l2 = new PALayer<>(p, null, new PASpatialPooler(), null, null, null);
l2.subscribe(new Observer<Inference>() {
@Override public void onCompleted() {}
@Override public void onError(Throwable e) { e.printStackTrace(); }
@Override
public void onNext(Inference i) {
// should be one and only onNext() called
assertTrue(Arrays.equals(expected1, i.getSDR()));
}
});
// SHOULD RECEIVE BOTH
// Now push some fake data through so that "onNext" is called above
l2.compute(inputs[0]);
l2.compute(inputs[1]);
}
/**
* Simple test to verify data gets passed through the {@link CLAClassifier}
* configured within the chain of components.
*/
@Ignore
public void testBasicClassifierSetup() {
Parameters p = NetworkTestHarness.getParameters().copy();
p = p.union(NetworkTestHarness.getDayDemoTestEncoderParams());
p.set(KEY.RANDOM, new MersenneTwister(42));
MultiEncoder me = MultiEncoder.builder().name("").build();
Layer<Map<String, Object>> l = new PALayer<>(p, me, new PASpatialPooler(), new TemporalMemory(), Boolean.TRUE, null);
l.subscribe(new Observer<Inference>() {
@Override public void onCompleted() {}
@Override public void onError(Throwable e) { System.out.println("error: " + e.getMessage()); e.printStackTrace();}
@Override
public void onNext(Inference i) {
assertNotNull(i);
assertEquals(36, i.getSDR().length);
}
});
// Now push some fake data through so that "onNext" is called above
Map<String, Object> multiInput = new HashMap<>();
multiInput.put("dayOfWeek", 0.0);
l.compute(multiInput);
}
/**
* The {@link PASpatialPooler} sometimes needs to have data run through it
* prior to passing the data on to subsequent algorithmic components. This
* tests the ability to specify exactly the number of input records for
* the PASpatialPooler to consume before passing records on.
*/
@Ignore
public void testMoreComplexSpatialPoolerPriming() {
final int PRIME_COUNT = 35;
final int NUM_CYCLES = 20;
final int INPUT_GROUP_COUNT = 7; // Days of Week
TOTAL = 0;
Parameters p = NetworkTestHarness.getParameters().copy();
p = p.union(NetworkTestHarness.getDayDemoTestEncoderParams());
p.set(KEY.RANDOM, new MersenneTwister(42));
p.set(KEY.SP_PRIMER_DELAY, PRIME_COUNT);
MultiEncoder me = MultiEncoder.builder().name("").build();
Layer<Map<String, Object>> l = new PALayer<>(p, me, new PASpatialPooler(), new TemporalMemory(), Boolean.TRUE, null);
l.subscribe(new Observer<Inference>() {
@Override public void onCompleted() {}
@Override public void onError(Throwable e) { System.out.println("error: " + e.getMessage()); e.printStackTrace();}
@Override
public void onNext(Inference i) {
assertNotNull(i);
TOTAL++;
}
});
// Now push some fake data through so that "onNext" is called above
Map<String, Object> multiInput = new HashMap<>();
for(int i = 0;i < NUM_CYCLES;i++) {
for(double j = 0;j < INPUT_GROUP_COUNT;j++) {
multiInput.put("dayOfWeek", j);
l.compute(multiInput);
}
}
// Assert we can accurately specify how many inputs to "prime" the spatial pooler
// and subtract that from the total input to get the total entries sent through
// the event chain from bottom to top.
assertEquals((NUM_CYCLES * INPUT_GROUP_COUNT) - PRIME_COUNT, TOTAL);
}
/**
* Tests the ability for multiple subscribers to receive copies of
* a given {@link PALayer}'s computed values.
*/
@Ignore
public void test2ndAndSubsequentSubscribersPossible() {
final int PRIME_COUNT = 35;
final int NUM_CYCLES = 50;
final int INPUT_GROUP_COUNT = 7; // Days of Week
TOTAL = 0;
Parameters p = NetworkTestHarness.getParameters().copy();
p = p.union(NetworkTestHarness.getDayDemoTestEncoderParams());
p.set(KEY.RANDOM, new MersenneTwister(42));
p.set(KEY.SP_PRIMER_DELAY, PRIME_COUNT);
MultiEncoder me = MultiEncoder.builder().name("").build();
Layer<Map<String, Object>> l = new PALayer<>(p, me, new PASpatialPooler(), new TemporalMemory(), Boolean.TRUE, null);
int[][] inputs = new int[7][8];
inputs[0] = new int[] { 1, 1, 0, 0, 0, 0, 0, 1 };
inputs[1] = new int[] { 1, 1, 1, 0, 0, 0, 0, 0 };
inputs[2] = new int[] { 0, 1, 1, 1, 0, 0, 0, 0 };
inputs[3] = new int[] { 0, 0, 1, 1, 1, 0, 0, 0 };
inputs[4] = new int[] { 0, 0, 0, 1, 1, 1, 0, 0 };
inputs[5] = new int[] { 0, 0, 0, 0, 1, 1, 1, 0 };
inputs[6] = new int[] { 0, 0, 0, 0, 0, 1, 1, 1 };
l.subscribe(new Observer<Inference>() {
@Override public void onCompleted() {}
@Override public void onError(Throwable e) { System.out.println("error: " + e.getMessage()); e.printStackTrace();}
@Override
public void onNext(Inference i) {
assertNotNull(i);
TOTAL++;
}
});
l.subscribe(new Observer<Inference>() {
@Override public void onCompleted() {}
@Override public void onError(Throwable e) { System.out.println("error: " + e.getMessage()); e.printStackTrace();}
@Override
public void onNext(Inference i) {
assertNotNull(i);
TOTAL++;
}
});
l.subscribe(new Observer<Inference>() {
@Override public void onCompleted() {}
@Override public void onError(Throwable e) { System.out.println("error: " + e.getMessage()); e.printStackTrace();}
@Override
public void onNext(Inference i) {
assertNotNull(i);
TOTAL++;
}
});
// Now push some fake data through so that "onNext" is called above
Map<String, Object> multiInput = new HashMap<>();
for(int i = 0;i < NUM_CYCLES;i++) {
for(double j = 0;j < INPUT_GROUP_COUNT;j++) {
multiInput.put("dayOfWeek", j);
l.compute(multiInput);
}
}
int NUM_SUBSCRIBERS = 3;
// Assert we can accurately specify how many inputs to "prime" the spatial pooler
// and subtract that from the total input to get the total entries sent through
// the event chain from bottom to top.
assertEquals( ((NUM_CYCLES * INPUT_GROUP_COUNT) - PRIME_COUNT) * NUM_SUBSCRIBERS, TOTAL);
}
@Ignore
public void testGetAllPredictions() {
final int PRIME_COUNT = 35;
final int NUM_CYCLES = 120;
final int INPUT_GROUP_COUNT = 7; // Days of Week
TOTAL = 0;
Parameters p = NetworkTestHarness.getParameters().copy();
p = p.union(NetworkTestHarness.getDayDemoTestEncoderParams());
p.set(KEY.RANDOM, new MersenneTwister(42));
p.set(KEY.SP_PRIMER_DELAY, PRIME_COUNT);
final int cellsPerColumn = (int)p.get(KEY.CELLS_PER_COLUMN);
assertTrue(cellsPerColumn > 0);
MultiEncoder me = MultiEncoder.builder().name("").build();
final PALayer<Map<String, Object>> l = new PALayer<>(p, me, new PASpatialPooler(), new TemporalMemory(), Boolean.TRUE, null);
l.setPADepolarize(0.0);
l.subscribe(new Observer<Inference>() {
@Override public void onCompleted() {}
@Override public void onError(Throwable e) { System.out.println("error: " + e.getMessage()); e.printStackTrace();}
@Override
public void onNext(Inference i) {
assertNotNull(i);
TOTAL++;
// if(l.getPreviousPredictiveCells() != null) {
//UNCOMMENT TO VIEW STABILIZATION OF PREDICTED FIELDS
// System.out.println("recordNum: " + i.getRecordNum() + " Day: " + ((Map<String, Object>)i.getLayerInput()).get("dayOfWeek") + " - " +
// Arrays.toString(ArrayUtils.where(l.getFeedForwardActiveColumns(), ArrayUtils.WHERE_1)) +
// " - " + Arrays.toString(SDR.cellsAsColumnIndices(l.getPreviousPredictiveCells(), cellsPerColumn)));
// }
}
});
// Now push some fake data through so that "onNext" is called above
Map<String, Object> multiInput = new HashMap<>();
for(int i = 0;i < NUM_CYCLES;i++) {
for(double j = 0;j < INPUT_GROUP_COUNT;j++) {
multiInput.put("dayOfWeek", j);
l.compute(multiInput);
}
l.reset();
}
// Assert we can accurately specify how many inputs to "prime" the spatial pooler
// and subtract that from the total input to get the total entries sent through
// the event chain from bottom to top.
assertEquals((NUM_CYCLES * INPUT_GROUP_COUNT) - PRIME_COUNT, TOTAL);
double[] all = l.getAllPredictions("dayOfWeek", 1);
double highestVal = Double.NEGATIVE_INFINITY;
int highestIdx = -1;
int i = 0;
for(double d : all) {
if(d > highestVal) {
highestIdx = i;
highestVal = d;
}
i++;
}
assertEquals(highestIdx, l.getMostProbableBucketIndex("dayOfWeek", 1));
assertEquals(7, l.getAllPredictions("dayOfWeek", 1).length);
assertTrue(Arrays.equals(
ArrayUtils.where(l.getFeedForwardActiveColumns(), ArrayUtils.WHERE_1),
SDR.cellsAsColumnIndices(l.getPreviousPredictiveCells(), cellsPerColumn)));
}
/**
* Test that a given layer can return an {@link Observable} capable of
* service multiple subscribers.
*/
@Ignore
public void testObservableRetrieval() {
Parameters p = NetworkTestHarness.getParameters().copy();
p = p.union(NetworkTestHarness.getDayDemoTestEncoderParams());
p.set(KEY.RANDOM, new MersenneTwister(42));
MultiEncoder me = MultiEncoder.builder().name("").build();
final Layer<Map<String, Object>> l = new PALayer<>(p, me, new PASpatialPooler(), new TemporalMemory(), Boolean.TRUE, null);
final List<int[]> emissions = new ArrayList<int[]>();
Observable<Inference> o = l.observe();
o.subscribe(new Subscriber<Inference>() {
@Override public void onCompleted() {}
@Override public void onError(Throwable e) { System.out.println("error: " + e.getMessage()); e.printStackTrace();}
@Override public void onNext(Inference i) {
emissions.add(l.getFeedForwardActiveColumns());
}
});
o.subscribe(new Subscriber<Inference>() {
@Override public void onCompleted() {}
@Override public void onError(Throwable e) { System.out.println("error: " + e.getMessage()); e.printStackTrace();}
@Override public void onNext(Inference i) {
emissions.add(l.getFeedForwardActiveColumns());
}
});
o.subscribe(new Subscriber<Inference>() {
@Override public void onCompleted() {}
@Override public void onError(Throwable e) { System.out.println("error: " + e.getMessage()); e.printStackTrace();}
@Override public void onNext(Inference i) {
emissions.add(l.getFeedForwardActiveColumns());
}
});
Map<String, Object> multiInput = new HashMap<>();
multiInput.put("dayOfWeek", 0.0);
l.compute(multiInput);
assertEquals(3, emissions.size());
int[] e1 = emissions.get(0);
for(int[] ia : emissions) {
assertTrue(ia == e1);//test same object propagated
}
}
/**
* Simple test to verify data gets passed through the {@link CLAClassifier}
* configured within the chain of components.
*/
boolean flowReceived = false;
@Ignore
public void testFullLayerFluentAssembly() {
Parameters p = NetworkTestHarness.getParameters().copy();
p = p.union(NetworkTestHarness.getHotGymTestEncoderParams());
p.set(KEY.RANDOM, new MersenneTwister(42));
p.set(KEY.COLUMN_DIMENSIONS, new int[] { 2048 });
p.set(KEY.POTENTIAL_RADIUS, 200);
p.set(KEY.INHIBITION_RADIUS, 50);
p.set(KEY.GLOBAL_INHIBITION, true);
// System.out.println(p);
Map<String, Object> params = new HashMap<>();
params.put(KEY_MODE, Mode.PURE);
params.put(KEY_WINDOW_SIZE, 3);
params.put(KEY_USE_MOVING_AVG, true);
Anomaly anomalyComputer = Anomaly.create(params);
Layer<?> l = Network.createPALayer("TestLayer", p)
.alterParameter(KEY.AUTO_CLASSIFY, true)
.add(anomalyComputer)
.add(new TemporalMemory())
.add(new PASpatialPooler())
.add(Sensor.create(
FileSensor::create,
SensorParams.create(
Keys::path, "", ResourceLocator.path("rec-center-hourly-small.csv"))));
l.getConnections().printParameters();
l.subscribe(new Observer<Inference>() {
@Override public void onCompleted() {}
@Override public void onError(Throwable e) { System.out.println("error: " + e.getMessage()); e.printStackTrace();}
@Override
public void onNext(Inference i) {
if(flowReceived) return; // No need to set this value multiple times
flowReceived = i.getClassifiers().size() == 2 &&
i.getClassifiers().get("timestamp") != null &&
i.getClassifiers().get("consumption") != null;
}
});
l.start();
try {
l.getLayerThread().join();
assertTrue(flowReceived);
}catch(Exception e) {
e.printStackTrace();
}
}
private Parameters getArrayTestParams() {
Map<String, Map<String, Object>> fieldEncodings = setupMap(
null,
884, // n
0, // w
0, 0, 0, 0, null, null, null,
"sdr_in", "darr", "SDRPassThroughEncoder");
Parameters p = Parameters.empty();
p.set(KEY.FIELD_ENCODING_MAP, fieldEncodings);
return p;
}
private Map<String, Map<String, Object>> setupMap(
Map<String, Map<String, Object>> map,
int n, int w, double min, double max, double radius, double resolution, Boolean periodic,
Boolean clip, Boolean forced, String fieldName, String fieldType, String encoderType) {
if(map == null) {
map = new HashMap<String, Map<String, Object>>();
}
Map<String, Object> inner = null;
if((inner = map.get(fieldName)) == null) {
map.put(fieldName, inner = new HashMap<String, Object>());
}
inner.put("n", n);
inner.put("w", w);
inner.put("minVal", min);
inner.put("maxVal", max);
inner.put("radius", radius);
inner.put("resolution", resolution);
if(periodic != null) inner.put("periodic", periodic);
if(clip != null) inner.put("clip", clip);
if(forced != null) inner.put("forced", forced);
if(fieldName != null) inner.put("fieldName", fieldName);
if(fieldType != null) inner.put("fieldType", fieldType);
if(encoderType != null) inner.put("encoderType", encoderType);
return map;
}
}
Raw
PASpatialPooler.java
/* ---------------------------------------------------------------------
* Numenta Platform for Intelligent Computing (NuPIC)
* Copyright (C) 2014, Numenta, Inc. Unless you have an agreement
* with Numenta, Inc., for a separate license for this software code, the
* following terms and conditions apply:
*
* This program is free software: you can redistribute it and/or modify
* it under the terms of the GNU Affero Public License version 3 as
* published by the Free Software Foundation.
*
* This program is distributed in the hope that it will be useful,
* but WITHOUT ANY WARRANTY; without even the implied warranty of
* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.
* See the GNU Affero Public License for more details.
*
* You should have received a copy of the GNU Affero Public License
* along with this program. If not, see http://www.gnu.org/licenses.
*
* http://numenta.org/licenses/
* ---------------------------------------------------------------------
*/
package org.numenta.nupic.algorithms;
import org.numenta.nupic.Connections;
import org.numenta.nupic.util.ArrayUtils;
/**
* Subclasses {@link SpatialPooler} to perform Prediction-Assisted CLA
*
* @author David Ray
* @author Fergal Byrne
*
*/
public class PASpatialPooler extends SpatialPooler {
private static final long serialVersionUID = 1L;
/**
* This function determines each column's overlap with the current input
* vector. The overlap of a column is the number of synapses for that column
* that are connected (permanence value is greater than '_synPermConnected')
* to input bits which are turned on. Overlap values that are lower than
* the 'stimulusThreshold' are ignored. The implementation takes advantage of
* the SpraseBinaryMatrix class to perform this calculation efficiently.
*
* @param c the {@link Connections} memory encapsulation
* @param inputVector an input array of 0's and 1's that comprises the input to
* the spatial pooler.
* @return
*/
public int[] calculateOverlap(Connections c, int[] inputVector) {
int[] overlaps = new int[c.getNumColumns()];
c.getConnectedCounts().rightVecSumAtNZ(inputVector, overlaps);
int[] paOverlaps = ArrayUtils.toIntArray(c.getPAOverlaps());
overlaps = ArrayUtils.i_add(paOverlaps, overlaps);
ArrayUtils.lessThanXThanSetToY(overlaps, (int)c.getStimulusThreshold(), 0);
return overlaps;
}
}
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment