Summary of (pass) Visual Prompt Locates Good Structure Sparsity Through a Recurrent Hypernetwork, by Tianjin Huang et al.
(PASS) Visual Prompt Locates Good Structure Sparsity through a Recurrent HyperNetwork
by Tianjin Huang, Fang Meng, Li Shen, Fan Liu, Yulong Pei, Mykola Pechenizkiy, Shiwei Liu, Tianlong Chen
First submitted to arxiv on: 24 Jul 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 framework, called PASS, is an innovative algorithm that leverages visual prompts to capture channel importance and derive high-quality structural sparsity for large-scale neural networks. This approach is inspired by the success of prompting-based techniques in generalizing large language models across diverse downstream tasks. By using a tailored hyper-network to process both visual prompts and network weight statistics, PASS outputs layer-wise channel sparsity while considering intrinsic channel dependencies between layers. Experimental results demonstrate the superiority of PASS on multiple network architectures and six datasets, achieving better accuracy with reduced FLOPs or improved speedup compared to baselines. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary Large-scale neural networks are very good at doing tasks like recognizing pictures and understanding language. However, they use a lot of computer power to do these tasks. One way to make them more efficient is to remove some of the parts that aren’t as important. This paper shows how to use visual prompts (like images or videos) to figure out which parts are most important and then remove the less important ones. The new algorithm, called PASS, is better than previous methods at finding the right parts to remove while still getting good results. |
Keywords
» Artificial intelligence » Prompting
