FPGA platform accelerates approximate multiplier evaluation for DNNs

Researchers have developed FAME, a new platform utilizing FPGAs to accelerate the evaluation of approximate multipliers for deep neural networks. This hardware-based approach significantly reduces the time needed to assess multiplier designs and their impact on DNN inference accuracy, outperforming traditional LUT-based emulation methods by up to 3.47x. Additionally, the platform incorporates a pattern-guided retraining technique that can recover up to 65.5% of accuracy loss compared to existing retraining methods. AI

IMPACT Accelerates hardware-level research for more efficient AI model inference.

RANK_REASON The cluster contains an academic paper detailing a new platform and methodology for hardware-based evaluation of approximate multipliers in deep neural networks. [lever_c_demoted from research: ic=1 ai=1.0]

Read on arXiv cs.LG →

• deep neural network
• Fame
• field-programmable gate array
• ImageNet
• MobileNetV2
• ResNet-18

• paper
• infra

AI-generated summary · Google Gemini · from 1 sources. How we write summaries →