Summary of How Robust Are Energy-based Models Trained with Equilibrium Propagation?, by Siddharth Mansingh et al.
How Robust Are Energy-Based Models Trained With Equilibrium Propagation?
by Siddharth Mansingh, Michal Kucer, Garrett Kenyon, Juston Moore, Michael Teti
First submitted to arxiv on: 21 Jan 2024
Categories
- Main: Machine Learning (cs.LG)
- Secondary: Computer Vision and Pattern Recognition (cs.CV)
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 This research paper explores the robustness of energy-based models (EBMs), a type of deep neural network designed for efficient implementation in neuromorphic hardware and physical systems. Unlike traditional DNNs, EBMs incorporate feedback connections that make them naturally robust to natural corruptions and adversarial attacks. The study uses the CIFAR-10 and CIFAR-100 datasets to evaluate the robustness of EBMs against different types of attacks, including gradient-based (white-box) and query-based (black-box) attacks, as well as natural perturbations. The results show that EBMs are more robust than transformers and comparable to adversarially-trained DNNs without sacrificing clean accuracy or requiring additional training techniques. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary This paper looks at a new kind of artificial intelligence called energy-based models (EBMs). These models are special because they have feedback connections, which makes them naturally good at dealing with noisy or confusing information. The researchers tested EBMs on two big datasets to see how well they could handle different types of attacks and noise. They found that EBMs are better than other AI systems at handling these challenges without sacrificing their ability to learn from clean data. |
Keywords
* Artificial intelligence * Neural network
