Normalization techniques in training dnns: Methodology, analysis and application

L Huang, J Qin, Y Zhou, F Zhu, L Liu… - IEEE transactions on …, 2023 - ieeexplore.ieee.org
Normalization techniques are essential for accelerating the training and improving the
generalization of deep neural networks (DNNs), and have successfully been used in various …

Scalelong: Towards more stable training of diffusion model via scaling network long skip connection

Z Huang, P Zhou, S Yan, L Lin - Advances in Neural …, 2023 - proceedings.neurips.cc
In diffusion models, UNet is the most popular network backbone, since its long skip connects
(LSCs) to connect distant network blocks can aggregate long-distant information and …

Representative batch normalization with feature calibration

SH Gao, Q Han, D Li, MM Cheng… - Proceedings of the …, 2021 - openaccess.thecvf.com
Abstract Batch Normalization (BatchNorm) has become the default component in modern
neural networks to stabilize training. In BatchNorm, centering and scaling operations, along …

Understanding self-attention mechanism via dynamical system perspective

Z Huang, M Liang, J Qin, S Zhong… - Proceedings of the …, 2023 - openaccess.thecvf.com
The self-attention mechanism (SAM) is widely used in various fields of artificial intelligence
and has successfully boosted the performance of different models. However, current …

Crossnorm and selfnorm for generalization under distribution shifts

Z Tang, Y Gao, Y Zhu, Z Zhang, M Li… - Proceedings of the …, 2021 - openaccess.thecvf.com
Traditional normalization techniques (eg, Batch Normalization and Instance Normalization)
generally and simplistically assume that training and test data follow the same distribution …

A generic shared attention mechanism for various backbone neural networks

Z Huang, S Liang, M Liang - Neurocomputing, 2025 - Elsevier
The self-attention mechanism is crucial for enhancing various backbone neural networks'
performance. However, current methods add self-attention modules (SAMs) to each network …

Rethinking the pruning criteria for convolutional neural network

Z Huang, W Shao, X Wang, L Lin… - Advances in Neural …, 2021 - proceedings.neurips.cc
Channel pruning is a popular technique for compressing convolutional neural networks
(CNNs), where various pruning criteria have been proposed to remove the redundant filters …

Scale region recognition network for object counting in intelligent transportation system

X Guo, M Gao, W Zhai, Q Li… - IEEE Transactions on …, 2023 - ieeexplore.ieee.org
Self-driving technology and safety monitoring devices in intelligent transportation systems
require superb capacity for context awareness. Accurately inferring the counts of crowds and …

Re-thinking the effectiveness of batch normalization and beyond

H Peng, Y Yu, S Yu - IEEE Transactions on Pattern Analysis …, 2023 - ieeexplore.ieee.org
Batch normalization (BN) is used by default in many modern deep neural networks due to its
effectiveness in accelerating training convergence and boosting inference performance …

On fast simulation of dynamical system with neural vector enhanced numerical solver

Z Huang, S Liang, H Zhang, H Yang, L Lin - Scientific reports, 2023 - nature.com
The large-scale simulation of dynamical systems is critical in numerous scientific and
engineering disciplines. However, traditional numerical solvers are limited by the choice of …