Summary of Multi-modal Dialogue State Tracking For Playing Guesswhich Game, by Wei Pang et al.
Multi-Modal Dialogue State Tracking for Playing GuessWhich Game
by Wei Pang, Ruixue Duan, Jinfu Yang, Ning Li
First submitted to arxiv on: 15 Aug 2024
Categories
- Main: Artificial Intelligence (cs.AI)
- Secondary: None
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 GuessWhich game is a visual dialogue platform that involves interaction between two AI bots, the Questioner Bot (QBot) and the Answer Bot (ABot). The QBot’s goal is to guess an image by asking visually related questions to the ABot. To effectively model this visually related reasoning, researchers propose a novel approach that focuses on mentally representing the unknown image. This mental model allows the QBot to track the dialogue state, comprising representations of mental imagery and conversation entities. At each round, the QBot engages in visual reasoning, generating relevant questions and updating its internal representation. Experimental results on VisDial datasets demonstrate the effectiveness of this proposed model, achieving new state-of-the-art performance across all metrics and datasets. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary GuessWhich is a fun game where AI chatbots ask each other questions to guess what an image looks like. The question-asking bot (QBot) needs to figure out how to describe the image using only visual clues from the answer-bot (ABot). Researchers came up with a new way for the QBot to do this by creating a mental picture of the unknown image. This helps the QBot keep track of what’s being talked about and make good guesses. They tested their idea on some big datasets and it worked really well! |
