Summary of Machine Unlearning For Document Classification, by Lei Kang et al.
Machine Unlearning for Document Classification
by Lei Kang, Mohamed Ali Souibgui, Fei Yang, Lluis Gomez, Ernest Valveny, Dimosthenis Karatzas
First submitted to arxiv on: 29 Apr 2024
Categories
- Main: Computer Vision and Pattern Recognition (cs.CV)
- Secondary: Artificial Intelligence (cs.AI)
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 Document understanding models have achieved remarkable performance by leveraging large collections of user documents. However, this raises concerns about user privacy and trust in AI services. To address these issues, legislation advocating the “right to be forgotten” has been proposed, allowing users to request removal of private information from computer systems and neural network models. Our research explores machine unlearning for document classification problems, a novel approach designed to make AI models forget specific data classes. We investigate this concept in a realistic scenario where a remote server houses a well-trained model and possesses only a small portion of training data, aiming to develop efficient forgetting manipulation methods. This work represents a pioneering step towards addressing privacy concerns in document analysis applications using machine unlearning. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary AI models are getting really good at understanding documents, but this raises big concerns about people’s privacy. Governments are proposing laws that allow people to ask AI systems to forget certain information about them. This is called the “right to be forgotten”. We’re trying to make AI models better at forgetting specific things they’ve learned from documents. We’re doing this by simulating a real-life scenario where a server has a powerful AI model and only a small part of its training data. Our goal is to create ways for AI systems to quickly forget information that’s no longer needed, which will help protect people’s privacy. |
Keywords
» Artificial intelligence » Classification » Neural network
