Summary of Unsolvable Problem Detection: Evaluating Trustworthiness Of Vision Language Models, by Atsuyuki Miyai et al.
Unsolvable Problem Detection: Evaluating Trustworthiness of Vision Language Models
by Atsuyuki Miyai, Jingkang Yang, Jingyang Zhang, Yifei Ming, Qing Yu, Go Irie, Yixuan Li, Hai Li, Ziwei Liu, Kiyoharu Aizawa
First submitted to arxiv on: 29 Mar 2024
Categories
- Main: Computer Vision and Pattern Recognition (cs.CV)
- Secondary: Artificial Intelligence (cs.AI); Computation and Language (cs.CL); 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 The paper introduces a novel challenge for Vision Language Models (VLMs) called Unsolvable Problem Detection (UPD), which evaluates their ability to withhold answers when faced with unsolvable problems in Visual Question Answering (VQA) tasks. UPD encompasses three settings: Absent Answer Detection (AAD), Incompatible Answer Set Detection (IASD), and Incompatible Visual Question Detection (IVQD). The study finds that most VLMs, including GPT-4V and LLaVA-Next-34B, struggle with these benchmarks to varying extents, highlighting opportunities for improvement. To address UPD, the authors explore both training-free and training-based solutions, offering insights into their effectiveness and limitations. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary This paper talks about a new problem called Unsolvable Problem Detection (UPD) that helps us understand how well computers can answer questions when they don’t have the answers. The problem is divided into three parts: AAD, IASD, and IVQD. Most computer models struggle with these problems, which means we need to improve them. To do this, scientists are looking at ways to make the computer models better without needing more training. |
Keywords
* Artificial intelligence * Gpt * Question answering
