Nonlinear Tensor Ring Network

TL;DR

This paper introduces a nonlinear tensor ring network (NTRN) that compresses neural network layers using tensor ring decomposition and nonlinear activation to reduce parameters while maintaining accuracy in image classification tasks.

Contribution

The paper proposes a novel nonlinear tensor ring network that combines tensor decomposition with nonlinear activation to improve compression and accuracy in neural networks.

Findings

Effective compression of LeNet-5 and VGG-11 models.

Superior performance on MNIST, Fashion MNIST, and Cifar-10 datasets.

Reduced model size with maintained accuracy.

Abstract

The state-of-the-art deep neural networks (DNNs) have been widely applied for various real-world applications, and achieved significant performance for cognitive problems. However, the increment of DNNs' width and depth in architecture results in a huge amount of parameters to challenge the storage and memory cost, limiting to the usage of DNNs on resource-constrained platforms, such as portable devices. By converting redundant models into compact ones, compression technique appears to be a practical solution to reducing the storage and memory consumption. In this paper, we develop a nonlinear tensor ring network (NTRN) in which both fullyconnected and convolutional layers are compressed via tensor ring decomposition. Furthermore, to mitigate the accuracy loss caused by compression, a nonlinear activation function is embedded into the tensor contraction and convolution operations inside the compressed layer. Experimental results demonstrate the effectiveness and superiority of the proposed NTRN for image classification using two basic neural networks, LeNet-5 and VGG-11 on three datasets, viz. MNIST, Fashion MNIST and Cifar-10.