Lei Huang

NormalizationSurvey. This repo is for our paper: Normalization Techniques in Training DNNs: Methodology, Analysis and Application

OthogonalWN. This project is the Torch implementation of our accepted AAAI 2018 paper : orthogonal weight normalization method for solving orthogonality constraints over Steifel manifold in deep neural networks (arXiv:1709.06079 )

IterNorm-pytorch. This is the pytorch re-implementation of the IterNorm

XBNBlock. This project is the PyTorch implementation of our CVPR 2022 paper:

IterNorm. This project is the Torch implementation of our accepted CVPR 2019 paper, Iterative Normalization: Beyond Standardization towards Efficient Whitening

ONI. PyTorch and Torch implementation for our accepted CVPR 2020 paper (Oral): Controllable Orthogonalization in Training DNNs

CenteredWN. This project is the Torch implementation of our ICCV 2017 paper: Centered Weight Normalization in Accelerating Training of Deep Neural Networks

LRCount-CNN. This code is for our ICML 2020 paper "On the Number of Linear Regions of Convolutional Neural Networks."

StochasticityBW. PyTorch and Tensorflow implementation for our accepted CVPR 2020 paper (Oral): An Investigation into the Stochasticity of Batch Whitening

GroupWhitening. This project is the PyTorch implementation of our accepted CVPR 2021 paper, Group Whitening: Balancing Learning Efficiency and Representational Capacity

Active-GBSSL. This is a project of my paper related to active learning and graph based semi-supervised learning, which is published on ICME 2014. The project is written by matlab

NormProjection. This project is the Torch implementation of the paper: Projection Based Weight Normalization for Deep Neural Networks (arXiv:1710.02338)

LayerwiseCA. This project is for the paper: Exploring learning dynamics of DNNs via layerwise conditioning analysis

Ladder_deepSSL_NP. The reimplementation of Ladder networks with projection based weight normalization. We achieved test errors as 2.52%, 1.06%, and 0.91% on Permunate invariant MNIST dataset with 20, 50, and 100 labeled samples respectively, which is the state-of-the-art results.