Paper Review

[2020.Q1] Deep Learning Paper List

Louis.T.Kim 2020. 1. 21. 13:50

[Up Comming]

[Reviewed]

[Less Concerning]

____________________________________________________________________________________________________________________________________________________________

 

 

[Compression, FPGA, Weight Sharing]

  1. DEEP COMPRESSION: COMPRESSING DEEP NEURAL NETWORKS WITH PRUNING, TRAINED QUANTIZATION AND HUFFMAN CODING 
  2. Going Deeper with Embedded FPGA Platform for Convolutional Neural Network
  3. Compressing Neural Networks with the Hashing Trick 

 

[Pruning]

 

  1. Learning both Weights and Connections for Efficient Neural Networks (NIPS 2015)
  2. NISP- Pruning Networks using Neuron Importance Score Propagation
  3. Pruning Filters for Efficient ConvNets
  4. Pruning Convolutional Neural Networks for Resource Efficient Inference 
  5. DSD: Dense-Sparse-Dense Training for Deep Neural Networks 
  6. Exploiting Sparsity in RNN 

 

[Quantizaiton]

 

  1. BinaryConnect - Training Deep Neural Networks with Binary Weights during Propagations 
  2. Binarized Neural Networks (arXiv 16)
  3. A Fully connected layer elimination for a binarizes convolutional neural network on an FPGA. (IEEE FPL 17)
  4. Accelerating Binarized Neural Networks : Comparision of FPGA, CPU, GPU, and ASIC

 

[VISION]

  • ImageNet Challenge NETWORKs
    1. VGGNet : Very DEEP Convolutional Neural Networks for OLarge-Scale Image Recognition (ICLR 15)
    2. GoogleNET : Going Deeper with Convolutions (CVPR 15)
    3. ResNet : Deep Residual Learning for Image Recognition (CVPR 16)
    4. DenseNet : Densely Connected Convolutional Networks (CVPR 17)
  • Semantic Segmentation 
    1. Fully Convolutional network for Semantic Segmentation (CVPR 15)
    2. DeepLab : Semantic Image Segmentation with Deep Convolutional Nets, Atrous Convolution, and Fully Connected CRFs (TPAMI 18)
    3. DeconvNet :  Learning Deconvolution Network for Semantic Segmentation(ICCV 15)
    4. U-NET : Convolutional networks for biomedical Image Segmentation (MICCAI 15)
  • Instance Segmentation
    1. Mask R-CNN (ICCV 17)
    2. Faster R-CNN (2016)

[GAN] 

  • SAGAN : Self-Attention Generative Adversarial Networks (2019)
  • WGAN : Wasserstein GAN (2017)

[Neuromorphic]

  • Reservoir Computing using dynamic memristors for temporal information processing (Nature 17)

[Attention]

  • Attention is all you need (NIPS 17)
  • Accelerating Neural Network Attention Mechanism with Approximation (HPCA 20) (ASIC with Attention)