SemiMul: Floating-Point Free Implementations for Efficient and Accurate Neural Network Training

Keyphrases

• Neural Network Training 100%
• Floating-point 100%
• Test Accuracy 100%
• Point-free 100%
• Neural Network 66%
• Network Task 66%
• FP16 66%
• FP32 66%
• Power Consumption 33%
• Hardware Architecture 33%
• Deep Neural Network 33%
• CPU-GPU 33%
• Machine Learning Tasks 33%
• Floating Point Unit 33%
• Hardware Implementation 33%
• MNIST 33%
• Hardware Cost 33%
• Overfitting 33%
• Integer Addition 33%
• Floating-point Arithmetic 33%
• Floating-point Format 33%
• Multiply-accumulate Operation 33%
• Shallow Neural Network 33%
• CIFAR-100 33%
• CIFAR-10 33%
• Fashion-MNIST 33%
• Embedded Processor 33%

Computer Science

• Floating Point 100%
• Neural Network Training 100%
• Free Implementation 100%
• Neural Network 66%
• Hardware Architecture 33%
• Power Consumption 33%
• Machine Learning 33%
• Deep Neural Network 33%
• Hardware Implementation 33%
• Application Specific Integrated Circuit 33%
• Floating-Point Unit 33%
• Hardware Cost 33%
• Fundamental Component 33%
• Point Multiplication 33%
• Embedded Processor 33%