Click on any project card to view full details, screenshots, and code links.

Robustness Evaluation of Authorship Verification Models in Multi-Domain and LLM-based Scenarios (working title)

This project explores how reliably authorship verification (AV) models can detect the same author across different text domains, such as news articles and tweets. By constructing a cross-domain evaluation set from the CROSSNEWS dataset and experimenting with both classical and LLM-based AV methods, the study examines the robustness of stylistic similarity detection under domain and style shifts.

PythonPyTorchDeep LearningNLPLLM

EEG-based Harmful Brain Activity Classification using Deep Learning

EEG signals are inherently challenging to interpret, even for trained clinicians. This study investigates whether deep learning models can accurately detect harmful brain activity from EEG data, while also addressing the interpretability of model predictions.

PythonPyTorchMachine LearningCNNXAI