1.exporation of data
2.demo of java class to classify data
Building neural network
- Specify the learning rate,input/output,batchsize,epochs and hidden nodes.
- Load the training data
- Build a network (seed,algorithm,iteration,learning rate,first layer as the dense layer(including input and output,activation function,ect),second layer(lose function) as the output layer)
- once it build up configuration, make sure the format of the output is JSON(configuration.toJson())
- Running model
- remember to init model Before run it
- Evaluate the model
- You can get the features and the data of predicted
Import keras neural network model into dl4j
- Read the model(.importsequentialmodel(modelpath))
- Create a recordreader and initialize it by reading the file where the data is.
- Specify a label index,number of classes and batchsize,then build a iterator of dataset and build dataset.
Now we have a model and a record reader to read data,so the next step is to evaluate the output.