Enhancing Detection of Misleading Visualizations Using DeepSeek Models

Project Code :TCMAPY1802

Objective

This project focuses on fine-tuning and evaluating deep learning models like GPT-2, XLNet, and RoBERTa to classify real and fake tweets, emphasizing detection of deceptive visuals and AI-generated content. It integrates multimodal data—combining text and images—to improve fake tweet detection. The study compares model performances using accuracy, precision, recall, and F1-score to identify strengths and weaknesses. Finally, it aims to develop a user-friendly system that allows users to input tweets for real-time authenticity checks, making the research practical and helping to combat misinformation on social media platforms.

Abstract

The emergence of social media networks has increased the challenges in telling the real and fake information. False information and deceptive visualization can go viral and impact on the masses, and even lead to social damage. This paper presents a solution to improve the techniques of fake twitter and especially those ones that present a false image or a message. The aim is to calibrate and compare the results of different state-of-the-art deep learning nature in detecting fake news, among which are GPT-2, DeepSeek-R1, DeepSeek-V3, XLNet, and RoBERTa.

Keywords: GPT-2, DeepSeek-R1, DeepSeek-V3, XLNet, RoBERTa, LLM, Deep Learming.

NOTE: Without the concern of our team, please don't submit to the college. This Abstract varies based on student requirements.