gaoyunhaii commented on a change in pull request #28: URL: https://github.com/apache/flink-ml/pull/28#discussion_r771186431
########## File path: flink-ml-lib/src/main/java/org/apache/flink/ml/classification/logisticregression/LogisticRegression.java ########## @@ -0,0 +1,458 @@ +/* + * Licensed to the Apache Software Foundation (ASF) under one + * or more contributor license agreements. See the NOTICE file + * distributed with this work for additional information + * regarding copyright ownership. The ASF licenses this file + * to you under the Apache License, Version 2.0 (the + * "License"); you may not use this file except in compliance + * with the License. You may obtain a copy of the License at + * + * http://www.apache.org/licenses/LICENSE-2.0 + * + * Unless required by applicable law or agreed to in writing, software + * distributed under the License is distributed on an "AS IS" BASIS, + * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. + * See the License for the specific language governing permissions and + * limitations under the License. + */ + +package org.apache.flink.ml.classification.logisticregression; + +import org.apache.flink.api.common.state.ListState; +import org.apache.flink.api.common.state.ListStateDescriptor; +import org.apache.flink.api.common.typeinfo.BasicTypeInfo; +import org.apache.flink.api.common.typeinfo.PrimitiveArrayTypeInfo; +import org.apache.flink.api.common.typeinfo.TypeInformation; +import org.apache.flink.api.java.tuple.Tuple2; +import org.apache.flink.iteration.DataStreamList; +import org.apache.flink.iteration.IterationBody; +import org.apache.flink.iteration.IterationBodyResult; +import org.apache.flink.iteration.IterationConfig; +import org.apache.flink.iteration.IterationListener; +import org.apache.flink.iteration.Iterations; +import org.apache.flink.iteration.ReplayableDataStreamList; +import org.apache.flink.iteration.operator.OperatorStateUtils; +import org.apache.flink.ml.api.Estimator; +import org.apache.flink.ml.common.datastream.DataStreamUtils; +import org.apache.flink.ml.common.feature.LabeledPointWithWeight; +import org.apache.flink.ml.common.iteration.TerminateOnMaxIterOrTol; +import org.apache.flink.ml.linalg.BLAS; +import org.apache.flink.ml.linalg.DenseVector; +import org.apache.flink.ml.param.Param; +import org.apache.flink.ml.util.ParamUtils; +import org.apache.flink.ml.util.ReadWriteUtils; +import org.apache.flink.runtime.state.StateInitializationContext; +import org.apache.flink.runtime.state.StateSnapshotContext; +import org.apache.flink.streaming.api.datastream.DataStream; +import org.apache.flink.streaming.api.datastream.SingleOutputStreamOperator; +import org.apache.flink.streaming.api.environment.StreamExecutionEnvironment; +import org.apache.flink.streaming.api.operators.AbstractStreamOperator; +import org.apache.flink.streaming.api.operators.BoundedOneInput; +import org.apache.flink.streaming.api.operators.OneInputStreamOperator; +import org.apache.flink.streaming.api.operators.TwoInputStreamOperator; +import org.apache.flink.streaming.runtime.streamrecord.StreamRecord; +import org.apache.flink.table.api.Table; +import org.apache.flink.table.api.bridge.java.StreamTableEnvironment; +import org.apache.flink.table.api.internal.TableImpl; +import org.apache.flink.util.Collector; +import org.apache.flink.util.OutputTag; +import org.apache.flink.util.Preconditions; + +import org.apache.commons.collections.IteratorUtils; + +import java.io.IOException; +import java.util.ArrayList; +import java.util.Arrays; +import java.util.HashMap; +import java.util.List; +import java.util.Map; +import java.util.Random; + +/** + * An Estimator which implements the logistic regression algorithm. + * + * <p>See https://en.wikipedia.org/wiki/Logistic_regression. + */ +public class LogisticRegression + implements Estimator<LogisticRegression, LogisticRegressionModel>, + LogisticRegressionParams<LogisticRegression> { + + private final Map<Param<?>, Object> paramMap = new HashMap<>(); + + public LogisticRegression() { + ParamUtils.initializeMapWithDefaultValues(paramMap, this); + } + + @Override + public Map<Param<?>, Object> getParamMap() { + return paramMap; + } + + @Override + public void save(String path) throws IOException { + ReadWriteUtils.saveMetadata(this, path); + } + + public static LogisticRegression load(StreamExecutionEnvironment env, String path) + throws IOException { + return ReadWriteUtils.loadStageParam(path); + } + + @Override + @SuppressWarnings({"rawTypes", "ConstantConditions"}) + public LogisticRegressionModel fit(Table... inputs) { + Preconditions.checkArgument(inputs.length == 1); + String classificationType = getMultiClass(); + Preconditions.checkArgument( + "auto".equals(classificationType) || "binomial".equals(classificationType), + "Multinomial classification is not supported yet. Supported options: [auto, binomial]."); + StreamTableEnvironment tEnv = + (StreamTableEnvironment) ((TableImpl) inputs[0]).getTableEnvironment(); + DataStream<LabeledPointWithWeight> trainData = + tEnv.toDataStream(inputs[0]) + .map( + dataPoint -> { + Double weight = + getWeightCol() == null + ? new Double(1.0) + : (Double) dataPoint.getField(getWeightCol()); + Double label = (Double) dataPoint.getField(getLabelCol()); + boolean isBinomial = + Double.compare(0., label) == 0 + || Double.compare(1., label) == 0; + if (!isBinomial) { + throw new RuntimeException( + "Multinomial classification is not supported yet. Supported options: [auto, binomial]."); + } + DenseVector features = + (DenseVector) dataPoint.getField(getFeaturesCol()); + return new LabeledPointWithWeight(features, label, weight); + }); + DataStream<double[]> initModelData = + trainData.transform( + "genInitModelData", + PrimitiveArrayTypeInfo.DOUBLE_PRIMITIVE_ARRAY_TYPE_INFO, + new GenInitModelData()); + + DataStream<LogisticRegressionModelData> modelData = train(trainData, initModelData); + LogisticRegressionModel model = + new LogisticRegressionModel().setModelData(tEnv.fromDataStream(modelData)); + ReadWriteUtils.updateExistingParams(model, paramMap); + return model; + } + + /** + * Generates initialized model data. Note that the parallelism of model data is same as the + * input train data, not one. + */ + private static class GenInitModelData extends AbstractStreamOperator<double[]> + implements OneInputStreamOperator<LabeledPointWithWeight, double[]>, BoundedOneInput { + + private int dim = 0; + + private ListState<Integer> dimState; + + @Override + public void endInput() { + output.collect(new StreamRecord<>(new double[dim])); + } + + @Override + public void processElement(StreamRecord<LabeledPointWithWeight> streamRecord) { + if (dim == 0) { + dim = streamRecord.getValue().features.size(); + } else { + if (dim != streamRecord.getValue().features.size()) { + throw new RuntimeException( + "The training data should all have same dimensions."); + } + } + } + + @Override + public void initializeState(StateInitializationContext context) throws Exception { + super.initializeState(context); + dimState = + context.getOperatorStateStore() + .getListState( + new ListStateDescriptor<>( + "dimState", BasicTypeInfo.INT_TYPE_INFO)); + dim = OperatorStateUtils.getUniqueElement(dimState, "dimState").orElse(0); + } + + @Override + public void snapshotState(StateSnapshotContext context) throws Exception { + dimState.clear(); + dimState.add(dim); + } + } + + /** + * Does machine learning training on the input data with the initialized model data. + * + * @param trainData The training data. + * @param initModelData The initialized model. + * @return The trained model data. + */ + private DataStream<LogisticRegressionModelData> train( + DataStream<LabeledPointWithWeight> trainData, DataStream<double[]> initModelData) { + LogisticGradient logisticGradient = new LogisticGradient(getReg()); + DataStreamList resultList = + Iterations.iterateBoundedStreamsUntilTermination( + DataStreamList.of(initModelData), + ReplayableDataStreamList.notReplay(trainData), + IterationConfig.newBuilder().build(), + new TrainIterationBody( + logisticGradient, + getGlobalBatchSize(), + getLearningRate(), + getMaxIter(), + getTol())); + return resultList.get(0); + } + + /** The iteration implementation for training process. */ + private static class TrainIterationBody implements IterationBody { + + private final LogisticGradient logisticGradient; + + private final int globalBatchSize; + + private final double learningRate; + + private final int maxIter; + + private final double tol; + + public TrainIterationBody( + LogisticGradient logisticGradient, + int globalBatchSize, + double learningRate, + int maxIter, + double tol) { + this.logisticGradient = logisticGradient; + this.globalBatchSize = globalBatchSize; + this.learningRate = learningRate; + this.maxIter = maxIter; + this.tol = tol; + } + + @Override + public IterationBodyResult process( + DataStreamList variableStreams, DataStreamList dataStreams) { + // The variable stream at the first iteration is the initialized model data. + // In the following iterations, it contains: the computed gradient, weightSum and + // lossSum. + DataStream<double[]> variableStream = variableStreams.get(0); + DataStream<LabeledPointWithWeight> trainData = dataStreams.get(0); + final OutputTag<LogisticRegressionModelData> modelDataOutputTag = + new OutputTag<LogisticRegressionModelData>("MODEL_OUTPUT") {}; + SingleOutputStreamOperator<double[]> gradientAndWeightAndLoss = + trainData + .connect(variableStream) + .transform( + "CacheDataAndDoTrain", + PrimitiveArrayTypeInfo.DOUBLE_PRIMITIVE_ARRAY_TYPE_INFO, + new CacheDataAndDoTrain( + logisticGradient, + globalBatchSize, + learningRate, + modelDataOutputTag)); + DataStreamList feedbackVariableStream = + IterationBody.forEachRound( + DataStreamList.of(gradientAndWeightAndLoss), + input -> { + DataStream<double[]> feedback = + DataStreamUtils.allReduceSum(input.get(0)); + return DataStreamList.of(feedback); + }); + DataStream<Integer> terminationCriteria = + feedbackVariableStream + .get(0) + .map( + reducedGradientAndWeightAndLoss -> { + double[] value = (double[]) reducedGradientAndWeightAndLoss; + return value[value.length - 1] / value[value.length - 2]; + }) + .flatMap(new TerminateOnMaxIterOrTol(maxIter, tol)); + return new IterationBodyResult( + DataStreamList.of(feedbackVariableStream.get(0)), + DataStreamList.of(gradientAndWeightAndLoss.getSideOutput(modelDataOutputTag)), + terminationCriteria); + } + } + + /** + * A stream operator that caches the training data in the first iteration and updates the model + * using gradients iteratively. The first input is the training data, and the second input is + * the initialized model data or feedback of gradient, weight and loss. + */ + private static class CacheDataAndDoTrain extends AbstractStreamOperator<double[]> + implements TwoInputStreamOperator<LabeledPointWithWeight, double[], double[]>, + IterationListener<double[]> { + + private final int globalBatchSize; + + private int localBatchSize; + + private final double learningRate; + + private final LogisticGradient logisticGradient; + + private DenseVector gradient; + + private DenseVector coefficient; + + private int coefficientDim; + + private ListState<DenseVector> coefficientState; + + private List<LabeledPointWithWeight> trainData; + + private ListState<LabeledPointWithWeight> trainDataState; + + private Random random = new Random(2021); + + private List<LabeledPointWithWeight> miniBatchData; + + /** The buffer for feedback record: {gradient, weightSum, loss}. */ + private double[] feedbackBuffer; + + private ListState<double[]> feedbackBufferState; + + private final OutputTag<LogisticRegressionModelData> modelDataOutputTag; + + public CacheDataAndDoTrain( + LogisticGradient logisticGradient, + int globalBatchSize, + double learningRate, + OutputTag<LogisticRegressionModelData> modelDataOutputTag) { + this.logisticGradient = logisticGradient; + this.globalBatchSize = globalBatchSize; + this.learningRate = learningRate; + this.modelDataOutputTag = modelDataOutputTag; + } + + @Override + public void open() { + int numTasks = getRuntimeContext().getNumberOfParallelSubtasks(); + int taskId = getRuntimeContext().getIndexOfThisSubtask(); + localBatchSize = globalBatchSize / numTasks; + if (globalBatchSize % numTasks > taskId) { + localBatchSize++; + } + this.miniBatchData = new ArrayList<>(localBatchSize); + } + + private List<LabeledPointWithWeight> getMiniBatchData( + List<LabeledPointWithWeight> fullBatchData, int batchSize) { + miniBatchData.clear(); + for (int i = 0; i < batchSize; i++) { + miniBatchData.add(fullBatchData.get(random.nextInt(fullBatchData.size()))); + } + return miniBatchData; + } + + private void updateModel() { + System.arraycopy(feedbackBuffer, 0, gradient.values, 0, gradient.size()); + double weightSum = feedbackBuffer[coefficientDim]; + BLAS.axpy(-learningRate / weightSum, gradient, coefficient); + } + + @Override + public void onEpochWatermarkIncremented( + int epochWatermark, Context context, Collector<double[]> collector) { + // TODO: let this method throws exception. + if (epochWatermark == 0) { + coefficient = new DenseVector(feedbackBuffer); + coefficientDim = coefficient.size(); + feedbackBuffer = new double[coefficientDim + 2]; + gradient = new DenseVector(coefficientDim); + } else { + updateModel(); + } + Arrays.fill(gradient.values, 0); + try { + if (trainData == null) { + trainData = IteratorUtils.toList(trainDataState.get().iterator()); + } + } catch (Exception e) { + throw new RuntimeException(e); + } + miniBatchData = getMiniBatchData(trainData, localBatchSize); + Tuple2<Double, Double> weightSumAndLossSum = + logisticGradient.computeLoss(miniBatchData, coefficient); + logisticGradient.computeGradient(miniBatchData, coefficient, gradient); + System.arraycopy(gradient.values, 0, feedbackBuffer, 0, gradient.size()); + feedbackBuffer[coefficientDim] = weightSumAndLossSum.f0; + feedbackBuffer[coefficientDim + 1] = weightSumAndLossSum.f1; + collector.collect(feedbackBuffer); + } + + @Override + public void onIterationTerminated(Context context, Collector collector) { Review comment: nit: Collector<double[]> -- This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. To unsubscribe, e-mail: [email protected] For queries about this service, please contact Infrastructure at: [email protected]
