Learning Low-rank Deep Neural Networks via Singular Vector Orthogonality Regularization and Singular Value Sparsification

by   Huanrui Yang, et al.

Modern deep neural networks (DNNs) often require high memory consumption and large computational loads. In order to deploy DNN algorithms efficiently on edge or mobile devices, a series of DNN compression algorithms have been explored, including factorization methods. Factorization methods approximate the weight matrix of a DNN layer with the multiplication of two or multiple low-rank matrices. However, it is hard to measure the ranks of DNN layers during the training process. Previous works mainly induce low-rank through implicit approximations or via costly singular value decomposition (SVD) process on every training step. The former approach usually induces a high accuracy loss while the latter has a low efficiency. In this work, we propose SVD training, the first method to explicitly achieve low-rank DNNs during training without applying SVD on every step. SVD training first decomposes each layer into the form of its full-rank SVD, then performs training directly on the decomposed weights. We add orthogonality regularization to the singular vectors, which ensure the valid form of SVD and avoid gradient vanishing/exploding. Low-rank is encouraged by applying sparsity-inducing regularizers on the singular values of each layer. Singular value pruning is applied at the end to explicitly reach a low-rank model. We empirically show that SVD training can significantly reduce the rank of DNN layers and achieve higher reduction on computation load under the same accuracy, comparing to not only previous factorization methods but also state-of-the-art filter pruning methods.


page 1

page 2

page 3

page 4


Maestro: Uncovering Low-Rank Structures via Trainable Decomposition

Deep Neural Networks (DNNs) have been a large driver and enabler for AI ...

Deep Learning Weight Pruning with RMT-SVD: Increasing Accuracy and Reducing Overfitting

In this work, we present some applications of random matrix theory for t...

Scalable Deep Neural Networks via Low-Rank Matrix Factorization

Compressing deep neural networks (DNNs) is important for real-world appl...

Compression-aware Continual Learning using Singular Value Decomposition

We propose a compression based continual task learning method that can d...

Deep Learning Approximation: Zero-Shot Neural Network Speedup

Neural networks offer high-accuracy solutions to a range of problems, bu...

Compact Speaker Embedding: lrx-vector

Deep neural networks (DNN) have recently been widely used in speaker rec...

Stabilizing Gradients for Deep Neural Networks via Efficient SVD Parameterization

Vanishing and exploding gradients are two of the main obstacles in train...

Please sign up or login with your details

Forgot password? Click here to reset