Summary of Connect, Collapse, Corrupt: Learning Cross-modal Tasks with Uni-modal Data, by Yuhui Zhang et al.
Connect, Collapse, Corrupt: Learning Cross-Modal Tasks with Uni-Modal Data
by Yuhui Zhang, Elaine Sui, Serena Yeung-Levy
First submitted to arxiv on: 16 Jan 2024
Categories
- Main: Machine Learning (cs.LG)
- Secondary: Artificial Intelligence (cs.AI); Computation and Language (cs.CL); Computer Vision and Pattern Recognition (cs.CV)
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 paper introduces a novel approach to bridging the modality gap in multi-modal contrastive representation spaces, enabling more effective cross-modal learning from uni-modal data. By developing a three-step method, C^3, which includes connecting, collapsing, and corrupting modalities, the authors demonstrate significant improvements in zero-shot image/audio/video captioning and text-to-image generation tasks. This research builds upon recent works that utilize pre-trained multi-modal contrastive representation spaces for cross-modal applications. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary The paper helps us better understand how to mix different types of data together, like pictures and words or sounds and images. It creates a new way to make sure these different kinds of data can work well together. This is important because it lets computers do things like describe pictures without needing to see them before. |
Keywords
* Artificial intelligence * Image generation * Multi modal * Zero shot
