Skip to content
/ mixup Public
forked from unsky/mixup

mixup: Beyond Empirical Risk Minimization

Notifications You must be signed in to change notification settings

hdjsjyl/mixup

 
 

Repository files navigation

MixUp

This is an implement and Improvement  on mixup: Beyond Empirical Risk Minimization https://arxiv.org/abs/1710.09412

The improvement

  1. add backward
  2. add mix rate

Two scenes:

image

The detail design of MixUp layer:

image

The results:

The symbol of resnet50 is writen by mxnet https://github.com/apache/incubator-mxnet/tree/master/example/image-classification/symbols, there have many versions. And i havenot do any optimizion for it. All the results are based on this baseline.

        cifar10              alpha        mix_rate test Acc initial learning rate batch size
(ERM)resnet50 90epoch - - 0.87900390625 0.05 256
(ERM)resnet50 200epoch - - 0.89365234375 0.05 256
(ERM)resnet50 300epoch - - 0.8931640625 0.05 256
(mixup)resnet50 90epoch 0.2 0.7 0.8609375 0.7 256
(mixup)resnet50 200epoch 0.2 0.7 0.91611328125 0.7 256
(mixup)resnet50 300epoch 0.2 0.7 0.9224609375 0.7 256
mixup in feature maps(resnet50 head conv)90epoch     0.2     0.7         0.8544921875 0.7 256
mixup in feature maps(resnet50 head conv)200epoch     0.2     0.7         0.91796875 0.7 256
mixup in feature maps(resnet50 head conv)300epoch     0.2     0.7         0.91845703125 0.7 256

MixUp

image

Mixup in feature map (resnet50 head conv)

image

ERM

image

Usage

install mxnet0.12 The mixup is in:symbols/mixup.py you can use it in your codes like:

data ,label = mx.sym.Custom(data= data,label = label,alpha = 0.2,num_classes = num_classes,batch_size = batch_size,mix_rate =0.7,op_type = 'MixUp')

label is the vector like [4,8,...9]

download the dataset

http://data.mxnet.io/data/cifar10/cifar10_val.rec

http://data.mxnet.io/data/cifar10/cifar10_train.rec

train:

./train.sh

test:

./test.sh

Reference

@article{zhang2017mixup, title={mixup: Beyond Empirical Risk Minimization}, author={Zhang, Hongyi and Cisse, Moustapha and Dauphin, Yann N and Lopez-Paz, David}, journal={arXiv preprint arXiv:1710.09412}, year={2017} }

About

mixup: Beyond Empirical Risk Minimization

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages

  • Python 99.6%
  • Shell 0.4%