Summary of Defining and Extracting Generalizable Interaction Primitives From Dnns, by Lu Chen et al.
Defining and Extracting generalizable interaction primitives from DNNs
by Lu Chen, Siyu Lou, Benhao Huang, Quanshi Zhang
First submitted to arxiv on: 29 Jan 2024
Categories
- Main: Machine Learning (cs.LG)
- Secondary: Artificial Intelligence (cs.AI); 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 The paper by Ren et al. (2024) tackles the challenge of explainable AI by deriving a series of theorems to prove that deep neural network (DNN) inference scores can be explained as a small set of interactions between input variables. While this approach is promising, it lacks generalization power and cannot yet be considered faithful primitive patterns encoded by the DNN. To address this limitation, the authors develop a new method to extract shared interactions from different DNNs trained for the same task. Experimental results show that these extracted interactions can better reflect common knowledge shared among multiple DNNs. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary The paper is trying to make deep neural networks (DNNs) more understandable. Right now, we don’t really know how they work or what patterns they’re recognizing. The authors came up with some math-based ways to show that the answers a DNN gives are related to simple interactions between different input pieces. This helps us understand the DNN better, but it’s still not perfect. To make it even better, the authors developed a new method to find common patterns in different DNNs that are doing the same job. It seems like this can help us get closer to truly understanding what these powerful computers are learning. |
Keywords
* Artificial intelligence * Generalization * Inference * Neural network
