Authors: Xingpeng Sun, Haoming Meng, Souradip Chakraborty, Amrit Singh Bedi, Aniket Bera
Published on: February 05, 2024
Impact Score: 8.3
Arxiv code: Arxiv:2402.03494
Summary
- What is new: ‘Beyond Text’ integrates audio features into LLMs for human-robot interaction, enhancing decision-making.
- Why this is important: Text-based LLMs fail to capture nuances in human-robot interactions, leading to trust issues.
- What the research proposes: An approach that includes audio transcription and paralinguistic features along with text for better interaction.
- Results: ‘Beyond Text’ outperforms existing LLMs with a 70.26% winning rate and shows greater resistance to adversarial attacks.
Technical Details
Technological frameworks used: nan
Models used: Large Language Models (LLMs) with audio feature integration
Data used: Audio transcriptions, paralinguistic features
Potential Impact
Social robot navigation, AI system developers, human-robot interaction platforms
Want to implement this idea in a business?
We have generated a startup concept here: SentiBotics.
Leave a Reply