Issue |
ITM Web Conf.
Volume 70, 2025
2024 2nd International Conference on Data Science, Advanced Algorithm and Intelligent Computing (DAI 2024)
|
|
---|---|---|
Article Number | 02003 | |
Number of page(s) | 7 | |
Section | Machine Learning in Healthcare and Finance | |
DOI | https://doi.org/10.1051/itmconf/20257002003 | |
Published online | 23 January 2025 |
Robustness of Big Language Modeling in Finance
Economic and Management, Tiangong University, 300380, Tianjin, China
Corresponding author: 1812010523@stu.hrbust.edu.com
With the gradual entry of artificial intelligence into all aspects of people’s lives, people begin to use big language models to solve problems in various fields. In the financial field, people use financial big prediction models to solve problems such as stock prediction, risk assessment, etc., but the big language models can be incorrect due to model hallucination and adversarial attacks. Therefore, investigating the robustness of large language models in finance is the main topic of this article, and searches the literature using the keywords “large language model”, “adversarial attack”, “model illusion”, etc. in recent years. We searched the literature in recent years. The existing literature explains the causes of adversarial attacks and model illusion, and methods that enhance the robustness of large language models are come up. It is shown that an attacker can trigger the model illusion of a large language model through an adversarial attack to reduce the reliability of the large language model. There is a lack of specific datasets of big language models in the financial domain to get a solution to improve the big language models in the financial domain in a better way. Future research should be specific in the financial domain for further adversarial training and robustness optimization of big language models in the financial domain.
© The Authors, published by EDP Sciences, 2025
This is an Open Access article distributed under the terms of the Creative Commons Attribution License 4.0, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Current usage metrics show cumulative count of Article Views (full-text article views including HTML views, PDF and ePub downloads, according to the available data) and Abstracts Views on Vision4Press platform.
Data correspond to usage on the plateform after 2015. The current usage metrics is available 48-96 hours after online publication and is updated daily on week days.
Initial download of the metrics may take a while.