Samir El-Amrany
Doctoral researcher at the FSTM
Samir works on the project ‘DETECTION OF FAKE NEWS WITH MULTIMODAL DATA ANALYTICS‘ under the supervision of Pascal Bouvry
The proliferation of fake news poses a significant challenge to society, influencing economic decisions and political debates, while hindering responses to emergencies such as natural disasters and terrorist attacks. Therefore, it is uttermost important to develop effective fake news detection methods to initiate decisive counter-measures.
Our approach focuses on addressing several challenges in fake news detection, such as timely detection, capturing spreading patterns and mitigation measures.
The methodology consists of training deep learning models, such as large language models (LLMs), on large-scale multimodal datasets which require the utilisation of high-performance computing (HPC) infrastructure due to the computational complexity.
To create comprehensive representations of instances of fake news, we extract features from different modalities, such as word embeddings from textual content, feature vectors from pre-trained models for image analysis, and graph neural networks for capturing spreading patterns based on user interactions like retweets, likes, and shares.
Optimisation techniques such as Adaptive Momentum Estimation (Adam) and early stopping will be applied to learn discriminative patterns and improve the performance of the detection models.
By leveraging the propagation patterns, incorporating diverse modalities, employing optimisation techniques, and harnessing the power of HPC, this research contributes to advancing the detection techniques of fake news. The findings of this study hold significant potential for improving societal resilience against the spread of fake news and promoting media literacy.