Summary of Mutual Information Preserving Neural Network Pruning, by Charles Westphal et al.
Mutual Information Preserving Neural Network Pruning
by Charles Westphal, Stephen Hailes, Mirco Musolesi
First submitted to arxiv on: 31 Oct 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 Mutual Information Preserving Pruning (MIPP), a novel technique for pruning neural networks (NNs) that leverages the mutual information (MI) between the pruning masks and the model’s training datasets. This approach, applicable before or after training, conserves MI shared between adjacent layers’ activations, ensuring re-trainability of pruned models. The authors demonstrate MIPP’s superiority over state-of-the-art methods in both pre- and post-training settings. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary Pruning helps make big neural networks smaller and more efficient. Scientists have been trying to figure out how to do this without losing the important information that makes the network work well. A new technique called Mutual Information Preserving Pruning (MIPP) is being introduced, which can be used before or after training a network. It works by preserving the connection between different parts of the network, so even when some parts are removed, the rest can still learn and remember things correctly. |
Keywords
* Artificial intelligence * Pruning
