Summary of Less Is More: Summarizing Patch Tokens For Efficient Multi-label Class-incremental Learning, by Thomas De Min et al.
Less is more: Summarizing Patch Tokens for efficient Multi-Label Class-Incremental Learning
by Thomas De Min, Massimiliano Mancini, Stéphane Lathuilière, Subhankar Roy, Elisa Ricci
First submitted to arxiv on: 24 May 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 The proposed MULTI-LANE method enables learning disentangled task-specific representations in multi-label class incremental learning (MLCIL) by maintaining task-specific pathways, reducing patch token embeddings to summarized tokens, and applying prompt tuning. This approach eliminates the need for selecting prompts corresponding to different foreground objects belonging to multiple tasks. The method achieves a new state-of-the-art in MLCIL and is also competitive in the class-incremental learning (CIL) setting. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary A new way of teaching machines to learn has been discovered! It’s called MULTI-LANE, and it helps machines understand different things at once. Normally, when we teach a machine something, it’s like giving them a set of instructions. But what if the instructions change every time? That’s what happens in this type of learning, called multi-label class incremental learning (MLCIL). To solve this problem, scientists created a new way to reduce the amount of information needed for each task and then teach the machine using prompts. This helps the machine learn faster and more accurately. They tested it on some famous datasets and found that it worked better than before! Now, anyone can try this new method by looking at the code online. |
Keywords
» Artificial intelligence » Prompt » Token
