Gradient is copied from Encog framework.KMeansCentriodsOutput is used to write the final k centers to file system.KMeansDataOutput is used to save tagged data into HDFS.KMeansMaster computes new k center points for next iteration.KMeansMasterParams is the master results for KMeans distributed guagua application.KMeansWorker re-computes each record tagged with new category.KMeansWorkerParams is the worker results for KMeans distributed guagua application.LinearRegressionMaster defines logic to update global linear regression model.LinearRegressionOutput is used to write the final model output to file system.LinearRegressionParams.parameters, while in other
iterations LinearRegressionParams.parameters is used to store gradients.LinearRegressionWorker defines logic to accumulate local linear regression gradients.LogisticRegressionMaster defines logic to update global logistic regression model.LogisticRegressionOutput is used to write the final model output to file system.LogisticRegressionParams.parameters, while in other
iterations LogisticRegressionParams.parameters is used to store gradients.LogisticRegressionWorker defines logic to accumulate local logistic regression gradients.BasicMLDataSet and BufferedMLDataSet together.MemoryDiskMLDataSet.fileName, MemoryDiskMLDataSet.inputCount and MemoryDiskMLDataSet.outputCountMemoryDiskMLDataSet.maxByteSize and MemoryDiskMLDataSet.fileNameNNMaster is used to accumulate all workers NN parameters.NNOutput is used to write the final model output to file system.NNWorker is used to compute NN model according to splits assigned.SumOutput is used to write the final model output to file system.SumSequenceFileWorker is used to accumulate the sum value for each line.SumWorker is used to accumulate the sum value for each line.Weight is used to update NN weights according to propagation option.Copyright © 2015. All Rights Reserved.