Search

Your search keyword '"Huang, Hantao"' showing total 2 results
Start Over Author "Huang, Hantao" Topic artificial neural networks
2 results on '"Huang, Hantao"'

1. Robust neural network–based tracking control and stabilization of a wheeled mobile robot with input saturation.

Author

Huang, Hantao, Zhou, Jingye, Di, Qing, Zhou, Jiawei, and Li, Jiawang

Subjects
*ARTIFICIAL neural networks, *TRACKING control systems, *MOBILE robots, *LYAPUNOV functions, *SIMULATION methods & models
Abstract

Summary: This paper presents a robust neural network–based control scheme to deal with the problem of tracking and stabilization simultaneously for a wheeled mobile robot subject to parametric uncertainties, external disturbances, and input saturation. At first, a new error‐state transformation scheme is designed by introducing some auxiliary variables as an additional virtual control signals to reduce the adverse effect caused by the underactuation. These variables can change their structures for different desired trajectories to be tracked. Then, a robust control law is proposed combining with a kinematic controller and a dynamic controller, while a three‐layer neural network system is applied to approximate model uncertainties. Stability analysis via the Lyapunov theory shows that the proposed controller can make tracking errors converge to bounded neighborhoods of the origin. Finally, some simulation results are illustrated to verify the effectiveness of the proposed control strategy. [ABSTRACT FROM AUTHOR]

Published
2019
Full Text
View/download PDF

2. LTNN: A Layerwise Tensorized Compression of Multilayer Neural Network.

Author

Huang, Hantao and Yu, Hao

Subjects
*ARTIFICIAL neural networks, *DEEP learning, *DATA compression
Abstract

An efficient deep learning requires a memory-efficient construction of a neural network. This paper introduces a layerwise tensorized formulation of a multilayer neural network, called LTNN, such that the weight matrix can be significantly compressed during training. By reshaping the multilayer neural network weight matrix into a high-dimensional tensor with a low-rank approximation, significant network compression can be achieved with maintained accuracy. An according layerwise training is developed by a modified alternating least-squares method with backward propagation for fine-tuning only. LTNN can provide the state-of-the-art results on various benchmarks with significant compression. For MNIST benchmark, LTNN shows $64 \times $ compression rate without accuracy drop. For Imagenet12 benchmark, our proposed LTNN achieves $35.84 \times $ compression of the neural network with around 2% accuracy drop. We have also shown $1.615 \times $ faster on inference speed than the existing works due to the smaller tensor core ranks. [ABSTRACT FROM AUTHOR]

Published
2019
Full Text
View/download PDF

Catalog

Books, media, physical & digital resources
See catalog results