Summary of Active Learning For Finely-categorized Image-text Retrieval by Selecting Hard Negative Unpaired Samples, By Dae Ung Jo et al.
Active Learning for Finely-Categorized Image-Text Retrieval by Selecting Hard Negative Unpaired Samples
by Dae Ung Jo, Kyuewang Lee, JaeHo Chung, Jin Young Choi
First submitted to arxiv on: 25 May 2024
Categories
- Main: Computer Vision and Pattern Recognition (cs.CV)
- Secondary: 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 This paper proposes an active learning algorithm for image-text retrieval (ITR) that efficiently collects paired data. Unlike previous studies, which assume category labels are provided, this work decreases the importance of category labels since a retrieval model can be trained with only image-text pairs. The proposed algorithm selects unpaired images or texts that can be hard negative samples for existing texts or images using a novel scoring function. This approach is validated on Flickr30K and MS-COCO datasets. The paper aims to address the issue of collecting paired data at a low cost, which is crucial for training an ITR model. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary This research helps computers find images that match text descriptions. To do this, they need a lot of examples, but getting these examples can be very expensive. The authors of this paper came up with a new way to get these examples more efficiently. They used a special method to choose which unpaired images or texts are most helpful for training the computer model. This approach was tested on two big datasets and showed promising results. |
Keywords
» Artificial intelligence » Active learning
