void testMNISTLoading(){
MNISTDataset mnist;
mnist.loadData();
MLP mlp("result/MLPModel.dat");
TrainModel mlpModel(mlp);
printf("validate error : %.8lf%%\n", 100.0 * mlpModel.getValidError(&mnist, 20));
}
void mlpUp(Dataset * combinDataSet, int numHid, double lr){
MLP mlp;
Layer *firstLayer = new SigmoidLayer(combinDataSet->getFeatureNumber(), numHid);
Logistic *secondLayer = new Logistic(numHid,combinDataSet->getLabelNumber());
mlp.addLayer(firstLayer);
mlp.addLayer(secondLayer);
TrainModel mlpModel(&mlp);
mlpModel.train(combinDataSet, lr,10,1000);
}
void testWFICA(){
TrivialDataset data;
data.loadData("../data/minist1_lcn_mlp.bin", "../data/minist1_lcn_label.bin");
MLP mlp;
MLPLayer *firstLayer = new TanhLayer(data.getFeatureNumber(), 500);
Logistic *secondLayer = new Logistic(500, data.getLabelNumber());
mlp.addLayer(firstLayer);
mlp.addLayer(secondLayer);
TrainModel mlpModel(mlp);
mlpModel.train(&data, 0.01, 20, 1000);
}
void testMNIST(){
MNISTDataset mnist;
mnist.loadData();
MLP mlp;
SigmoidLayer *firstLayer = new SigmoidLayer(mnist.getFeatureNumber(), 500);
Logistic *secondLayer = new Logistic(500, mnist.getLabelNumber());
mlp.addLayer(firstLayer);
mlp.addLayer(secondLayer);
mlp.setModelFile("result/MLPModel.dat");
TrainModel mlpModel(mlp);
mlpModel.train(&mnist, 0.01, 20, 1);
}
