Summary of Non-transferable Pruning, by Ruyi Ding et al.
Non-transferable Pruning
by Ruyi Ding, Lili Su, Aidong Adam Ding, Yunsi Fei
First submitted to arxiv on: 10 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 This paper proposes a novel approach to protecting intellectual property (IP) in pre-trained deep neural networks (DNNs). Specifically, the authors focus on preventing unauthorized transfer learning by developing a method called Non-Transferable Pruning (NTP). NTP leverages model pruning to control a DNN’s transferability to unauthorized data domains. The framework employs alternating direction method of multipliers (ADMM) for optimizing both model sparsity and an innovative non-transferable learning loss, augmented with Fisher space discriminative regularization to constrain the model’s generalizability. A novel metric, Area Under the Sample-wise Learning Curve (SLC-AUC), is proposed to measure model non-transferability. The authors demonstrate that NTP outperforms state-of-the-art non-transferable learning methods, achieving an average SLC-AUC of -0.54 across diverse pairs of source and target domains. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary This paper protects intellectual property in pre-trained deep neural networks by preventing unauthorized transfer learning. It develops a method called Non-Transferable Pruning (NTP) to control what a model can do with new data. The authors use a combination of techniques, including alternating direction method of multipliers and Fisher space regularization, to make sure the model doesn’t work well on new data it shouldn’t be using. They also created a special metric called SLC-AUC to measure how good this approach is at stopping unauthorized transfer learning. |
Keywords
» Artificial intelligence » Auc » Pruning » Regularization » Transfer learning » Transferability
