Summary of Limited but Consistent Gains in Adversarial Robustness by Co-training Object Recognition Models with Human Eeg, By Manshan Guo and Bhavin Choksi and Sari Sadiya and Alessandro T. Gifford and Martina G. Vilas and Radoslaw M. Cichy and Gemma Roig
Limited but consistent gains in adversarial robustness by co-training object recognition models with human EEG
by Manshan Guo, Bhavin Choksi, Sari Sadiya, Alessandro T. Gifford, Martina G. Vilas, Radoslaw M. Cichy, Gemma Roig
First submitted to arxiv on: 5 Sep 2024
Categories
- Main: Machine Learning (cs.LG)
- Secondary: Artificial Intelligence (cs.AI); Human-Computer Interaction (cs.HC)
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 paper proposes a novel approach to improving the robustness of artificial neural networks (ANNs) against adversarial attacks by aligning model representations to human brain activity. The researchers trained ResNet50-backbone models on a dual task of classification and EEG prediction, using a rich set of real-world images as stimuli. They found that the networks’ EEG prediction accuracy was significantly correlated with their gains in adversarial robustness, particularly around 100 ms post-stimulus onset. The study also explored the contribution of individual EEG channels to this effect, finding strongest contributions from parieto-occipital regions. This work demonstrates the potential of human EEG data for improving ANNs’ robustness and opens up avenues for future research. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary This paper helps artificial neural networks (ANNs) be better at dealing with tricky attacks by making them learn like humans do. Scientists trained special models that could both recognize pictures and predict what’s happening in people’s brains when they look at those pictures. They found that the more accurately these models predicted brain activity, the better they were at defending against sneaky attacks. This is an important step towards making ANNs more reliable and opens up new ways to study how humans think. |
Keywords
» Artificial intelligence » Classification
