DSC: Dense-sparse convolution for vectorized inference of convolutional neural networks

Keyphrases

• Convolutional Neural Network 100%
• Vectorized 100%
• Sparse Convolution 100%
• Pruning 66%
• Word Length 66%
• Sparsity 33%
• Safety-critical 33%
• Design Optimization 33%
• Real-world Application 33%
• Programming Techniques 33%
• Hardware Resources 33%
• Optimization Programming 33%
• Low-latency Inference 33%
• Number of Cores 33%
• Hardware Accelerator 33%
• Proposed Methodology 33%

Computer Science

• Convolutional Neural Network 100%
• Word Length 66%
• Open Source 33%
• Deep Neural Network 33%
• Sparsity 33%
• World Application 33%
• Programming Technique 33%
• Hardware Accelerator 33%
• Hardware Resource 33%
• Compute Unit 33%
• Design Optimization 33%
• Computer Hardware 33%