Summary of Joint-task Regularization For Partially Labeled Multi-task Learning, by Kento Nishi et al.
Joint-Task Regularization for Partially Labeled Multi-Task Learning
by Kento Nishi, Junsik Kim, Wanhua Li, Hanspeter Pfister
First submitted to arxiv on: 2 Apr 2024
Categories
- Main: Computer Vision and Pattern Recognition (cs.CV)
- Secondary: Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
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 Multi-task learning has become increasingly popular in machine learning, but its practicality is hindered by the need for large, labeled datasets. Most methods rely on fully labeled datasets with ground-truth labels for all target tasks, which can be prohibitively expensive to curate. We propose Joint-Task Regularization (JTR), an intuitive technique that leverages cross-task relations to regularize all tasks in a single joint-task latent space. This improves learning when data is not fully labeled for all tasks. Unlike existing approaches, JTR achieves linear complexity relative to the number of tasks. We extensively benchmark our method across partially labeled scenarios based on NYU-v2, Cityscapes, and Taskonomy. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary Joint-Task Regularization (JTR) helps machines learn from data that is not fully labeled for all tasks. This makes it more practical for real-world use. Right now, most machine learning methods require a lot of labeled data to work well. But gathering this kind of data can be very expensive and time-consuming. JTR solves this problem by using relationships between different tasks to help each task learn better. |
Keywords
* Artificial intelligence * Latent space * Machine learning * Multi task * Regularization
