TrueGPT: can you privately extract algorithms from ChatGPT in tabular classification?

Recently, it has been shown that Large Language Models (LLMs) achieve impressive zero-shot classification on tabular data, revealing an internal algorithm ALLM without explicit training data. We predict that ALLM will become a standard for tabular data classification, replacing resource-intensive...

Full description

Saved in:
Bibliographic Details
Main Author: Soegeng, Hans Farrell
Other Authors: Thomas Peyrin
Format: Final Year Project
Language:English
Published: Nanyang Technological University 2024
Subjects:
Online Access:https://hdl.handle.net/10356/175642
Tags: Add Tag
No Tags, Be the first to tag this record!
Institution: Nanyang Technological University
Language: English
id sg-ntu-dr.10356-175642
record_format dspace
spelling sg-ntu-dr.10356-1756422024-05-06T15:36:58Z TrueGPT: can you privately extract algorithms from ChatGPT in tabular classification? Soegeng, Hans Farrell Thomas Peyrin School of Physical and Mathematical Sciences Adrien Benamira thomas.peyrin@ntu.edu.sg, adrien.benamira@ntu.edu.sg Computer and Information Science Mathematical Sciences Machine learning Explainable AI Large language models Recently, it has been shown that Large Language Models (LLMs) achieve impressive zero-shot classification on tabular data, revealing an internal algorithm ALLM without explicit training data. We predict that ALLM will become a standard for tabular data classification, replacing resource-intensive custom ML models. However, LLM complexity hinders regulatory transparency. To address this, we introduce a method to approximate ALLM with human-interpretable binary feature rules Aerule. We utilize the TT-rules (Truth Table rules) model developed by Benamira et al., 2023 to extract the binary rules through the LLM inference of tabular datasets. Following the extraction and approximation processes, we set aside the LLM and exclusively rely on Aerule for inference. Our method is fully automatic. We validate the approach on 8 public tabular datasets, adding a user option to activate privacy-preserving feature to ensure owner data protection. Bachelor's degree 2024-05-02T04:38:18Z 2024-05-02T04:38:18Z 2024 Final Year Project (FYP) Soegeng, H. F. (2024). TrueGPT: can you privately extract algorithms from ChatGPT in tabular classification?. Final Year Project (FYP), Nanyang Technological University, Singapore. https://hdl.handle.net/10356/175642 https://hdl.handle.net/10356/175642 en application/pdf Nanyang Technological University
institution Nanyang Technological University
building NTU Library
continent Asia
country Singapore
Singapore
content_provider NTU Library
collection DR-NTU
language English
topic Computer and Information Science
Mathematical Sciences
Machine learning
Explainable AI
Large language models
spellingShingle Computer and Information Science
Mathematical Sciences
Machine learning
Explainable AI
Large language models
Soegeng, Hans Farrell
TrueGPT: can you privately extract algorithms from ChatGPT in tabular classification?
description Recently, it has been shown that Large Language Models (LLMs) achieve impressive zero-shot classification on tabular data, revealing an internal algorithm ALLM without explicit training data. We predict that ALLM will become a standard for tabular data classification, replacing resource-intensive custom ML models. However, LLM complexity hinders regulatory transparency. To address this, we introduce a method to approximate ALLM with human-interpretable binary feature rules Aerule. We utilize the TT-rules (Truth Table rules) model developed by Benamira et al., 2023 to extract the binary rules through the LLM inference of tabular datasets. Following the extraction and approximation processes, we set aside the LLM and exclusively rely on Aerule for inference. Our method is fully automatic. We validate the approach on 8 public tabular datasets, adding a user option to activate privacy-preserving feature to ensure owner data protection.
author2 Thomas Peyrin
author_facet Thomas Peyrin
Soegeng, Hans Farrell
format Final Year Project
author Soegeng, Hans Farrell
author_sort Soegeng, Hans Farrell
title TrueGPT: can you privately extract algorithms from ChatGPT in tabular classification?
title_short TrueGPT: can you privately extract algorithms from ChatGPT in tabular classification?
title_full TrueGPT: can you privately extract algorithms from ChatGPT in tabular classification?
title_fullStr TrueGPT: can you privately extract algorithms from ChatGPT in tabular classification?
title_full_unstemmed TrueGPT: can you privately extract algorithms from ChatGPT in tabular classification?
title_sort truegpt: can you privately extract algorithms from chatgpt in tabular classification?
publisher Nanyang Technological University
publishDate 2024
url https://hdl.handle.net/10356/175642
_version_ 1800916303729393664