Various kinds of deep residual networks for CIFAR.

Requirements

• Python 3.5+
• TensorFlow 0.10+
• numpy

Models & Commands

original redisual network

--model resnet 

He, Kaiming, et al. "Deep residual learning for image recognition." arXiv preprint arXiv:1512.03385 (2015).

--model resnet --residual_type 0

Bottleneck design

pre-activation residual network

--model resnet --unit_type 1

He, Kaiming, et al. "Identity mappings in deep residual networks." arXiv preprint arXiv:1603.05027 (2016).

wide residual network

--model [any model] --wide_factor k

Zagoruyko, Sergey, and Nikos Komodakis. "Wide Residual Networks." arXiv preprint arXiv:1605.07146 (2016).

RoR

--model ror 

Zhang, Ke, et al. "Residual Networks of Residual Networks: Multilevel Residual Networks." arXiv preprint arXiv:1608.02908 (2016).

DenseNet

--model densenet

Huang, Gao, Zhuang Liu, and Kilian Q. Weinberger. "Densely Connected Convolutional Networks." arXiv preprint arXiv:1608.06993 (2016).

residual network with shared weights

--model resrnn

Liao, Qianli, and Tomaso Poggio. "Bridging the Gaps Between Residual Learning, Recurrent Neural Networks and Visual Cortex." arXiv preprint arXiv:1604.03640 (2016).

Running Examples

python -m train --data_dir ~/work/vrn/data/ --outputs_dir ~/work/vrn/outputs/ --dataset cifar-10 --model resnet
python -m train --data_dir ~/work/vrn/data/ --outputs_dir ~/work/vrn/outputs/ --dataset cifar-10 --model ror
python -m train --data_dir ~/work/vrn/data/ --outputs_dir ~/work/vrn/outputs/ --dataset cifar-10 --model resnet --print_step 3 --summary_step 40 --eval_step 3
python -m train --dataset cifar-10 --model resnet --print_step 3 --summary_step 40 --eval_step 400

License

MIT