SparseConvNet/cifar10indexLearning.cpp at master · SpiderMonkey1975/SparseConvNet · GitHub

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
#include "SparseConvNet.h"
#include "SpatiallySparseDatasetCIFAR10.h"

int epoch=0;
int pciBusID=-1; //PCI bus ID, -1 for default GPU
int batchSize=300;

Picture* OpenCVPicture::distort(RNG& rng, batchType type) {
  OpenCVPicture* pic=new OpenCVPicture(*this);
  if (type==TRAINBATCH)
    pic->jiggle(rng,4);
  return pic;
}

class CNN : public SparseConvNet {
public:
  CNN(int dimension, int nInputFeatures, int nClasses, int pciBusID) : SparseConvNet(dimension, nInputFeatures, nClasses, pciBusID) {
    addLeNetLayerMP(32,3,1,2,2,VLEAKYRELU);
    addLeNetLayerMP(32,2,1,2,2,VLEAKYRELU);
    addLeNetLayerMP(32,2,1,2,2,VLEAKYRELU);
    addLeNetLayerMP(32,2,1,1,1,VLEAKYRELU);
    addLeNetLayerMP(32*2,1,1,1,1,NOSIGMOID,0.5);
    addIndexLearnerLayer();
  }
};


int main() {
  std::string baseName="weights/Cifar10IndexLearner";

  SpatiallySparseDataset trainSet=Cifar10TrainSet();   //For training
  SpatiallySparseDataset trainSet2=Cifar10TrainSet();trainSet2.type=TESTBATCH; //For calculating the top hidden layer for the training set (without dropout or data augmentation)
  SpatiallySparseDataset testSet=Cifar10TestSet(); //For calculating the top hidden layer for the test set

  trainSet.summary();
  trainSet2.summary();
  testSet.summary();
  CNN cnn(2,trainSet.nFeatures,trainSet.pictures.size(),pciBusID);
  if (epoch>0) {
    cnn.loadWeights(baseName,epoch);
  }
  for (epoch++;epoch<=100000;epoch++) {
    std::cout <<"epoch: " << epoch << std::endl;
    cnn.processIndexLearnerDataset(trainSet, batchSize,0.0003);
    if (epoch%20==0) {
      cnn.saveWeights(baseName,epoch);
      cnn.processIndexLearnerDataset(trainSet2, batchSize);
      cnn.processIndexLearnerDataset(testSet, batchSize);
    }
  }
  // // // {//After pretraining, convert to a regular neural network
  // // //   cnn.nOutputFeatures=dynamic_cast<IndexLearnerLayer*>(cnn.cnn.back())->nFeaturesIn;
  // // //   delete cnn.cnn.back();
  // // //   cnn.cnn.pop_back();
  // // //   cnn.nClasses=trainSet.nClasses;
  // // //   cnn.addLearntLayer(1024);cout <<endl;
  // // //   cnn.addLearntLayer(1024);cout <<endl;
  // // //   cnn.addSoftmaxLayer();cout <<endl;
  // // // }
  // // // for (;;epoch++) {
  // // //   cout <<"epoch: " << epoch << flush;
  // // //   cnn.processDataset(trainSet, batchSize,0.0003);
  // // //   if (epoch%10==0) {
  // // //     cnn.saveWeights(baseName,epoch);
  // // //     cnn.processDatasetRepeatTest(testSet, batchSize, 1);
  // // //   }
  // // // }
  //
}