Toward conversational interpretations of neural networks: data collection

Toward conversational interpretations of neural networks: data collection

Neural networks are powerful techniques for automated decision making. However, they are also blackboxes, which human experts find difficult to understand. Recent work performed at NTU and internationally suggests that conversation is an effective form of interpreting neural networks to layperson us...

Full description

Saved in:
Bibliographic Details
Main Author: Yeow, Ming Xuan
Other Authors: Li Boyang
Format: Final Year Project
Language:English
Published: Nanyang Technological University 2024
Subjects:
Computer and Information Science
Computer science
Machine learning
Explainable AI
LLM
Online Access:https://hdl.handle.net/10356/181279
Tags: Add Tag
No Tags, Be the first to tag this record!
Institution: Nanyang Technological University
Language: English
Description
Summary:Neural networks are powerful techniques for automated decision making. However, they are also blackboxes, which human experts find difficult to understand. Recent work performed at NTU and internationally suggests that conversation is an effective form of interpreting neural networks to layperson users. In this project, we aim to collected conversation data where layperson users interact with human experts, who explain the neural networks to them. We then finetune an LLM, in an attempt to combine conversational AI with XAI.

Similar Items