Summary of Automated and Holistic Co-design Of Neural Networks and Asics For Enabling In-pixel Intelligence, by Shubha R. Kharel et al.
Automated and Holistic Co-design of Neural Networks and ASICs for Enabling In-Pixel Intelligence
by Shubha R. Kharel, Prashansa Mukim, Piotr Maj, Grzegorz W. Deptuch, Shinjae Yoo, Yihui Ren, Soumyajit Mandal
First submitted to arxiv on: 18 Jul 2024
Categories
- Main: Machine Learning (cs.LG)
- Secondary: Artificial Intelligence (cs.AI); Hardware Architecture (cs.AR)
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 A recent study tackles the challenge of designing optimal AI systems for extreme edge applications, such as radiation detection. To operate under stringent hardware constraints like micron-level dimensions, sub-milliwatt power, and nanosecond-scale speed, while providing clear accuracy advantages over traditional architectures, researchers employ a novel approach that combines multi-objective Bayesian optimization with neural network search and ASIC synthesis. This integrated loop provides reliable feedback on the collective impact of all cross-domain design choices, enabling the discovery of Pareto-optimal design choices for effective and efficient neural networks. The study showcases its effectiveness by finding several such designs that perform real-time feature extraction from input pulses within the individual pixels of a readout ASIC. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary A team of researchers is working on designing better AI systems for special tasks like detecting radiation. They want these AI systems to work well even when they’re very small and use very little power, which makes it hard to design them. The scientists used a new way of searching through lots of options to find the best ones that work well and are also efficient. This approach helps find the perfect combination of choices for designing good AI networks. They tested this method by finding some examples of great designs that can quickly analyze data from small pixels in special devices. |
Keywords
* Artificial intelligence * Feature extraction * Neural network * Optimization
