Summary of Approxdarts: Differentiable Neural Architecture Search with Approximate Multipliers, by Michal Pinos et al.
ApproxDARTS: Differentiable Neural Architecture Search with Approximate Multipliers
by Michal Pinos, Lukas Sekanina, Vojtech Mrazek
First submitted to arxiv on: 8 Apr 2024
Categories
- Main: Machine Learning (cs.LG)
- Secondary: None
GrooveSquid.com Paper Summaries
GrooveSquid.com’s goal is to make artificial intelligence research accessible by summarizing AI papers in simpler terms. Each summary below covers the same AI paper, written at different levels of difficulty. The medium difficulty and low difficulty versions are original summaries written by GrooveSquid.com, while the high difficulty version is the paper’s original abstract. Feel free to learn from the version that suits you best!
| Summary difficulty | Written by | Summary |
|---|---|---|
| High | Paper authors | High Difficulty Summary Read the original abstract here |
| Medium | GrooveSquid.com (original content) | Medium Difficulty Summary The paper introduces ApproxDARTS, a neural architecture search (NAS) method that enables DARTS to utilize approximate multipliers and reduce the power consumption of generated neural networks. The approach is applied to convolutional neural networks (CNNs) on the CIFAR-10 dataset, achieving significant energy savings with minimal accuracy loss. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary ApproxDARTS helps design efficient CNNs by leveraging approximate computing principles in DNN implementations. This innovative method performs NAS within 10 GPU hours and produces competitive results using approximate multipliers in convolutional layers. By reducing arithmetic operations during inference, ApproxDARTS achieves energy savings of up to 53.84% compared to native floating-point multipliers and 5.97% compared to exact fixed-point multipliers. |
Keywords
* Artificial intelligence * Inference
