Summary of Modeling Output-level Task Relatedness in Multi-task Learning with Feedback Mechanism, by Xiangming Xi et al.
Modeling Output-Level Task Relatedness in Multi-Task Learning with Feedback Mechanism
by Xiangming Xi, Feng Gao, Jun Xu, Fangtai Guo, Tianlei Jin
First submitted to arxiv on: 1 Apr 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 In this paper, researchers explore multi-task learning (MTL) by focusing on output-level task relatedness, where tasks share information through correlated outputs. They introduce a feedback mechanism into MTL models, transforming static models into dynamic ones. This is achieved by incorporating a posteriori information that considers the mutual influences between tasks’ outputs. To ensure training convergence, they propose a convergence loss that measures the trend of each task’s output during iterations. Additionally, they suggest a Gumbel gating mechanism to determine optimal feedback signal projections. The effectiveness of this approach is evaluated through experiments conducted on various baseline models in spoken language understanding. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary This paper explores how machines can learn many tasks at once by sharing information between them. Instead of just looking at what features are important, it looks at how the answers might be related too. It creates a special feedback loop that lets one task help another task make better guesses. This helps the machine learn more efficiently and accurately. The authors test their approach on several models and show that it works well for spoken language understanding tasks. |
Keywords
* Artificial intelligence * Language understanding * Multi task
