package hex.deeplearning;
import hex.ScoringInfo;
import hex.deeplearning.DeepLearningModel.DeepLearningParameters;
import hex.deeplearning.DeepLearningModel.DeepLearningParameters.Activation;
import hex.deeplearning.DeepLearningModel.DeepLearningParameters.InitialWeightDistribution;
import hex.deeplearning.DeepLearningModel.DeepLearningParameters.Loss;
import hex.genmodel.GenModel;
import org.junit.Assert;
import org.junit.BeforeClass;
import org.junit.Ignore;
import org.junit.Test;
import water.DKV;
import water.Key;
import water.TestUtil;
import water.fvec.Frame;
import water.fvec.Vec;
import water.util.ArrayUtils;
import water.util.Log;
import water.util.MathUtils;
import water.util.RandomUtils;
import java.util.Random;
public class DeepLearningIrisTest extends TestUtil {
static final String PATH = "smalldata/iris/iris.csv";
Frame _train, _test;
@BeforeClass() public static void setup() { stall_till_cloudsize(1); }
// Default run is the short run
@Test public void run() throws Exception { runFraction(0.05f); }
private void compareVal(double a, double b, double abseps, double releps) {
if( !MathUtils.compare(a,b,abseps,releps) ) // Complex test does not fit JUnit Assert very well
Assert.assertEquals("Not equal: ", a, b, 0.0); // always fails if we get here, and prints nice msg
}
void runFraction(float fraction) {
long seed0 = 0xDECAF;
int num_runs = 0;
Frame frame = null;
try {
frame = parse_test_file(Key.make("iris.hex"),PATH);
for (int repeat = 0; repeat < 5; ++repeat) {
// Testing different things
// Note: Microsoft reference implementation is only for Tanh + MSE.
// Note: Rectifier and MCE are implemented by H2O.ai (trivial).
// Note: Initial weight distributions are copied, but what is tested is the stability behavior.
Activation[] activations = {Activation.Tanh, Activation.Rectifier};
Loss[] losses = {Loss.Quadratic, Loss.CrossEntropy};
InitialWeightDistribution[] dists = {
InitialWeightDistribution.Normal,
InitialWeightDistribution.Uniform,
InitialWeightDistribution.UniformAdaptive
};
final long seed = seed0 + repeat;
Random rng = new Random(seed);
double[] initial_weight_scales = {1e-4 + rng.nextDouble()};
double[] holdout_ratios = {0.1 + rng.nextDouble() * 0.8};
double[] momenta = {rng.nextDouble() * 0.99};
int[] hiddens = {1, 2 + rng.nextInt(50)};
int[] epochs = {1, 2 + rng.nextInt(50)};
double[] rates = {0.01, 1e-5 + rng.nextDouble() * .1};
for (Activation activation : activations) {
for (Loss loss : losses) {
for (InitialWeightDistribution dist : dists) {
for (double scale : initial_weight_scales) {
for (double holdout_ratio : holdout_ratios) {
for (double momentum : momenta) {
for (int hidden : hiddens) {
for (int epoch : epochs) {
for (double rate : rates) {
DeepLearningModel mymodel = null;
Frame trainPredict = null;
Frame testPredict = null;
try {
num_runs++;
if (fraction < rng.nextFloat()) continue;
Log.info("");
Log.info("STARTING.");
Log.info("Running with " + activation.name() + " activation function and " + loss.name() + " loss function.");
Log.info("Initialization with " + dist.name() + " distribution and " + scale + " scale, holdout ratio " + holdout_ratio);
Log.info("Using " + hidden + " hidden layer neurons and momentum: " + momentum);
Log.info("Using seed " + seed);
Random rand;
int trial = 0;
do {
Log.info("Trial #" + ++trial);
if (_train != null) _train.delete();
if (_test != null) _test .delete();
rand = RandomUtils.getRNG(seed);
double[][] rows = new double[(int) frame.numRows()][frame.numCols()];
String[] names = new String[frame.numCols()];
for (int c = 0; c < frame.numCols(); c++) {
names[c] = "ColumnName" + c;
for (int r = 0; r < frame.numRows(); r++)
rows[r][c] = frame.vecs()[c].at(r);
}
for (int i = rows.length - 1; i >= 0; i--) {
int shuffle = rand.nextInt(i + 1);
double[] row = rows[shuffle];
rows[shuffle] = rows[i];
rows[i] = row;
}
int limit = (int) (frame.numRows() * holdout_ratio);
_train = ArrayUtils.frame(names, water.util.ArrayUtils.subarray(rows, 0, limit));
_test = ArrayUtils.frame(names, water.util.ArrayUtils.subarray(rows, limit, (int) frame.numRows() - limit));
// Must have all output classes in training
// data (since that's what the reference
// implementation has hardcoded). But count
// of classes is not known unless we visit
// all the response data - force that now.
String respname = _train.lastVecName();
Vec resp = _train.lastVec().toCategoricalVec();
_train.remove(respname).remove();
_train.add(respname, resp);
DKV.put(_train);
Vec vresp = _test.lastVec().toCategoricalVec();
_test.remove(respname).remove();
_test.add(respname, vresp);
DKV.put(_test);
}
while( _train.lastVec().cardinality() < 3);
// use the same seed for the reference implementation
DeepLearningMLPReference ref = new DeepLearningMLPReference();
ref.init(activation, RandomUtils.getRNG(seed), holdout_ratio, hidden);
DeepLearningParameters p = new DeepLearningParameters();
p._train = _train._key;
p._response_column = _train.lastVecName();
assert _train.lastVec().isCategorical();
p._ignored_columns = null;
p._seed = seed;
p._hidden = new int[]{hidden};
p._adaptive_rate = false;
p._rho = 0;
p._epsilon = 0;
p._rate = rate / (1 - momentum); //adapt to (1-m) correction that's done inside (only for constant momentum!)
p._activation = activation;
p._max_w2 = Float.POSITIVE_INFINITY;
p._input_dropout_ratio = 0;
p._rate_annealing = 0; //do not change - not implemented in reference
p._l1 = 0;
p._loss = loss;
p._l2 = 0;
p._momentum_stable = momentum; //reference only supports constant momentum
p._momentum_start = p._momentum_stable; //do not change - not implemented in reference
p._momentum_ramp = 0; //do not change - not implemented in reference
p._initial_weight_distribution = dist;
p._initial_weight_scale = scale;
p._valid = null;
p._quiet_mode = true;
p._fast_mode = false; //to be the same as reference
// p._fast_mode = true; //to be the same as old NeuralNet code
p._nesterov_accelerated_gradient = false; //to be the same as reference
// p._nesterov_accelerated_gradient = true; //to be the same as old NeuralNet code
p._train_samples_per_iteration = 0; //sync once per period
p._ignore_const_cols = false;
p._shuffle_training_data = false;
p._classification_stop = -1; //don't stop early -> need to compare against reference, which doesn't stop either
p._force_load_balance = false; //keep just 1 chunk for reproducibility
p._overwrite_with_best_model = false;
p._replicate_training_data = false;
p._mini_batch_size = 1;
p._single_node_mode = true;
p._epochs = 0;
p._elastic_averaging = false;
mymodel = new DeepLearning(p).trainModel().get();
p._epochs = epoch;
Neurons[] neurons = DeepLearningTask.makeNeuronsForTraining(mymodel.model_info());
// use the same random weights for the reference implementation
Neurons l = neurons[1];
for (int o = 0; o < l._a[0].size(); o++) {
for (int i = 0; i < l._previous._a[0].size(); i++) {
// System.out.println("initial weight[" + o + "]=" + l._w[o * l._previous._a.length + i]);
ref._nn.ihWeights[i][o] = l._w.get(o, i);
}
ref._nn.hBiases[o] = l._b.get(o);
// System.out.println("initial bias[" + o + "]=" + l._b[o]);
}
l = neurons[2];
for (int o = 0; o < l._a[0].size(); o++) {
for (int i = 0; i < l._previous._a[0].size(); i++) {
// System.out.println("initial weight[" + o + "]=" + l._w[o * l._previous._a.length + i]);
ref._nn.hoWeights[i][o] = l._w.get(o, i);
}
ref._nn.oBiases[o] = l._b.get(o);
// System.out.println("initial bias[" + o + "]=" + l._b[o]);
}
// Train the Reference
ref.train((int) p._epochs, rate, p._momentum_stable, loss, seed);
// Train H2O
mymodel.delete();
DeepLearning dl = new DeepLearning(p);
mymodel = dl.trainModel().get();
Assert.assertTrue(mymodel.model_info().get_processed_total() == epoch * dl.train().numRows());
/**
* Tolerances (should ideally be super tight -> expect the same double/float precision math inside both algos)
*/
final double abseps = 1e-6;
final double releps = 1e-6;
/**
* Compare weights and biases in hidden layer
*/
neurons = DeepLearningTask.makeNeuronsForTesting(mymodel.model_info()); //link the weights to the neurons, for easy access
l = neurons[1];
for (int o = 0; o < l._a[0].size(); o++) {
for (int i = 0; i < l._previous._a[0].size(); i++) {
double a = ref._nn.ihWeights[i][o];
double b = l._w.get(o, i);
compareVal(a, b, abseps, releps);
// System.out.println("weight[" + o + "]=" + b);
}
double ba = ref._nn.hBiases[o];
double bb = l._b.get(o);
compareVal(ba, bb, abseps, releps);
}
Log.info("Weights and biases for hidden layer: PASS");
/**
* Compare weights and biases for output layer
*/
l = neurons[2];
for (int o = 0; o < l._a[0].size(); o++) {
for (int i = 0; i < l._previous._a[0].size(); i++) {
double a = ref._nn.hoWeights[i][o];
double b = l._w.get(o, i);
compareVal(a, b, abseps, releps);
}
double ba = ref._nn.oBiases[o];
double bb = l._b.get(o);
compareVal(ba, bb, abseps, releps);
}
Log.info("Weights and biases for output layer: PASS");
/**
* Compare predictions
* Note: Reference and H2O each do their internal data normalization,
* so we must use their "own" test data, which is assumed to be created correctly.
*/
// H2O predictions
Frame fpreds = mymodel.score(_test); //[0] is label, [1]...[4] are the probabilities
try {
for (int i = 0; i < _test.numRows(); ++i) {
// Reference predictions
double[] xValues = new double[neurons[0]._a[0].size()];
System.arraycopy(ref._testData[i], 0, xValues, 0, xValues.length);
double[] ref_preds = ref._nn.ComputeOutputs(xValues);
// find the label
// do the same as H2O here (compare float values and break ties based on row number)
double[] preds = new double[ref_preds.length + 1];
for (int j = 0; j < ref_preds.length; ++j) preds[j + 1] = ref_preds[j];
preds[0] = GenModel.getPrediction(preds, null, xValues, 0.5);
// compare predicted label
Assert.assertTrue(preds[0] == (int) fpreds.vecs()[0].at(i));
// // compare predicted probabilities
// for (int j=0; j<ref_preds.length; ++j) {
// compareVal((float)(ref_preds[j]), fpreds.vecs()[1+j].at(i), abseps, releps);
// }
}
} finally {
if (fpreds != null) fpreds.delete();
}
Log.info("Predicted values: PASS");
/**
* Compare (self-reported) scoring
*/
final double trainErr = ref._nn.Accuracy(ref._trainData);
final double testErr = ref._nn.Accuracy(ref. _testData);
trainPredict = mymodel.score(_train);
testPredict = mymodel.score(_test );
hex.ModelMetrics mmtrain = hex.ModelMetrics.getFromDKV(mymodel,_train);
hex.ModelMetrics mmtest = hex.ModelMetrics.getFromDKV(mymodel,_test );
final double myTrainErr = mmtrain.cm().err();
final double myTestErr = mmtest .cm().err();
Log.info("H2O training error : " + myTrainErr * 100 + "%, test error: " + myTestErr * 100 + "%");
Log.info("REF training error : " + trainErr * 100 + "%, test error: " + testErr * 100 + "%");
compareVal(trainErr, myTrainErr, abseps, releps);
compareVal( testErr, myTestErr, abseps, releps);
Log.info("Scoring: PASS");
// get the actual best error on training data
float best_err = Float.MAX_VALUE;
for (ScoringInfo e : mymodel.scoring_history()) {
DeepLearningScoringInfo err = (DeepLearningScoringInfo) e;
best_err = Math.min(best_err, (float) (Double.isNaN(err.scored_train._classError) ? best_err : err.scored_train._classError)); //multi-class classification
}
Log.info("Actual best error : " + best_err * 100 + "%.");
// this is enabled by default
if (p._overwrite_with_best_model) {
Frame bestPredict = null;
try {
bestPredict = mymodel.score(_train);
hex.ModelMetrics mmbest = hex.ModelMetrics.getFromDKV(mymodel,_train);
final double bestErr = mmbest.cm().err();
Log.info("Best_model's error : " + bestErr * 100 + "%.");
compareVal(bestErr, best_err, abseps, releps);
} finally {
if (bestPredict != null) bestPredict.delete();
}
}
Log.info("Parameters combination " + num_runs + ": PASS");
} finally {
// cleanup
if (mymodel != null) {
mymodel.delete();
}
if (_train != null) _train.delete();
if (_test != null) _test.delete();
if (trainPredict != null) trainPredict.delete();
if (testPredict != null) testPredict.delete();
}
}
}
}
}
}
}
}
}
}
}
} catch(Throwable t) {
t.printStackTrace();
throw new RuntimeException(t);
} finally {
if (frame != null) frame.delete();
}
}
public static class Long extends DeepLearningIrisTest {
@Test
@Ignore
public void run() throws Exception { runFraction(0.1f); }
}
public static class Short extends DeepLearningIrisTest {
@Test @Ignore public void run() throws Exception { runFraction(0.05f); }
}
}