Custom Hardware Inference Accelerator for TensorFlow Lite for Microcontrollers

Keyphrases

• Microcontroller 100%
• Edge Devices 100%
• Custom Hardware 100%
• Inference Accelerator 100%
• TensorFlow Lite 100%
• Hardware Accelerator 80%
• Low Power 60%
• Neural Unit 60%
• Machine Learning Inference 60%
• IoT Devices 20%
• Neural Network 20%
• High Performance 20%
• Performance Improvement 20%
• Proposed Methodology 20%
• Low Area 20%
• Low-resource 20%
• Limited Resources 20%
• Computational Complexity 20%
• Software Implementation 20%
• Neural Network Architecture 20%
• Efficient Deployment 20%
• Neural Network Model 20%
• Computational Load 20%
• Resource-constrained 20%
• Hardware Resources 20%
• Efficient Mapping 20%
• Power Limit 20%
• Hardware Complexity 20%
• Computational Demand 20%
• Body Weight Support 20%
• Efficient Hardware 20%
• Software Framework 20%
• Area Footprint 20%
• CIFAR-10 20%
• Weight Compression 20%
• Software Resources 20%
• Quantized Neural Networks 20%
• Model Sparsity 20%
• Embedded Processor 20%
• MLPerf 20%
• Neural Network Accelerator 20%
• Constrained Edge 20%
• Real-time Machine Learning 20%
• Pruned Models 20%
• Software Deployment 20%

Computer Science

• Microcontroller 100%
• Custom Hardware 100%
• Machine Learning 60%
• Processing Unit 60%
• Hardware Accelerator 60%
• Neural Network 40%
• hardware acceleration 40%
• Internet of Things 20%
• Performance Improvement 20%
• Experimental Result 20%
• Computational Complexity 20%
• Software Implementation 20%
• Neural Network Architecture 20%
• Sparsity 20%
• Neural Network Model 20%
• Hardware Resource 20%
• Computational Load 20%
• Reference Model 20%
• Software Frameworks 20%
• Embedded Processor 20%
• Software Resource 20%
• Software Deployment 20%