DAC: Data-free Automatic Acceleration of Convolutional Networks

Abstract

Deploying a deep learning model on mobile/IoT devices is a challenging task. The difficulty lies in the trade-off between computation speed and accuracy. A complex deep learning model with high accuracy runs slowly on resource-limited devices, while a light-weight model that runs much faster loses accuracy. In this paper, we propose a novel decomposition method, namely DAC, that is capable of factorizing an ordinary convolutional layer into two layers with much fewer parameters. DAC computes the corresponding weights for the newly generated layers directly from the weights of the original convolutional layer. Thus, no training (or fine-tuning) or any data is needed. The experimental results show that DAC reduces a large number of floating-point operations (FLOPs) while maintaining high accuracy of a pre-trained model. If 2% accuracy drop is acceptable, DAC saves 53% FLOPs of VGG16 image classification model on ImageNet dataset, 29% FLOPS of SSD300 object detection model on PASCAL VOC2007 dataset, and 46% FLOPS of a multi-person pose estimation model on Microsoft COCO dataset. Compared to other existing decomposition methods, DAC achieves better performance.

Accepted by IEEE Winter Conference on Applications of Computer Vision (WACV 2019)

References in corpus (14)

• MobileNets: Efficient Convolutional Neural Networks for Mobile Vision Applications
• TensorFlow: Large-Scale Machine Learning on Heterogeneous Distributed Systems
• MobileNetV2: Inverted Residuals and Linear Bottlenecks
• Pruning Convolutional Neural Networks for Resource Efficient Inference
• Speeding up Convolutional Neural Networks with Low Rank Expansions
• Channel Pruning for Accelerating Very Deep Neural Networks
• Speeding-up Convolutional Neural Networks Using Fine-tuned CP-Decomposition
• Learning Sparse Neural Networks through $L_0$ Regularization
• Network Decoupling: From Regular to Depthwise Separable Convolutions
• Minimization of Transformed $L_1$ Penalty: Theory, Difference of Convex Function Algorithm, and Robust Application in Compressed Sensing
• Blended Coarse Gradient Descent for Full Quantization of Deep Neural Networks
• BinaryRelax: A Relaxation Approach For Training Deep Neural Networks With Quantized Weights
• Minimization of Transformed $L_1$ Penalty: Closed Form Representation and Iterative Thresholding Algorithms
• Transformed Schatten-1 Iterative Thresholding Algorithms for Low Rank Matrix Completion