Description

MultilayerPerceptronClassifier is a neural network based multi-class classifier. Valina neural network with all dense layers are used, the output layer is a softmax layer. Number of inputs has to be equal to the size of feature vectors. Number of outputs has to be equal to the total number of labels.

Parameters

Name Description Type Required？ Default Value
layers Size of each neural network layers. int[] ✓
blockSize Size for stacking training samples, the default value is 64. Integer 64
initialWeights Initial weights. DenseVector null
vectorCol Name of a vector column String null
featureCols Names of the feature columns used for training in the input table String[] null
labelCol Name of the label column in the input table String ✓
maxIter Maximum iterations, The default value is 100 Integer 100
epsilon Convergence tolerance for iterative algorithms (>= 0), The default value is 1.0e-06 Double 1.0E-6
l1 the L1-regularized parameter. Double 0.0
l2 the L2-regularized parameter. Double 0.0

Script Example

Code

  1.     mlpc = MultilayerPerceptronClassifier() \
  2.         .setVectorCol("bitmap") \
  3.         .setLabelCol("label") \
  4.         .setLayers([628, 100, 100]) \
  5.         .setMaxIter(100) \
  6.         .setPredictionCol("pred_label") \
  7.         .setPredictionDetailCol("pred_detail")
  9.     # mlpc.fit(batch_data.bo_mnist)
  10.     pipeline = Pipeline().add(mlpc)
  11.     model = pipeline.fit(batch_data.bo_mnist)
  12.     model.save('/tmp/mlpc.csv')
  13.     BatchOperator.execute()

Results

  1. -1,"{""schema"":[""model_id BIGINT,model_info VARCHAR,label_value BIGINT""],""param"":[""{\""vectorCol\"":\""\\\""bitmap\\\""\"",\""maxIter\"":\""100\"",\""layers\"":\""[628,100,100]\"",\""labelCol\"":\""\\\""label\\\""\"",\""predictionCol\"":\""\\\""pred_label\\\""\"",\""predictionDetailCol\"":\""\\\""pred_detail\\\""\""}""],""clazz"":[""com.alibaba.alink.pipeline.classification.MultilayerPerceptronClassificationModel""]}"
  2. 0,"0^{""vectorCol"":""\""bitmap\"""",""isVectorInput"":""true"",""layers"":""[628,100,100]"",""featureCols"":null}^"
  3. ......