Abstract:
"The world of financial markets changes rapidly, which makes it hard to identify stock market
frauds. Even though traditional rule-based systems work, they don't always have the advanced
detection tools needed to identify complex frauds. These schemes n hurt the fairness of the
market, , also cost investors and companies considerable amounts of funds, which means that
financial world require more effective methods to identify it.
In order to address this problem, the author is proposed to create an novel method for finding
anomalies which employs a selective learning transformer-based architecture The model uses
selective learning in an attempt identify the complex patterns of market influence that other
methods miss a majority of the time. The novel approach has the potential to be far more
reliable than rule-based systems, also marks an important leap forward in using deep learning
to find frauds in finance.
The proposed model has achieved an overall better accuracy compared to other models, such
as LSTM, in the market manipulation domain. Additionally, this research has received better
feedback from evaluators. In conclusion, this project has managed to contribute significantly
to the financial domain."
