MultiFake: A Multimodal Deep Learning Approach for Fake News Detection

View at publisher Summarise with VIRAT
Authors: S Hemkiran, Kavyasree Raj S., Harshavarthinie RJ, Devadharshini S.
Cited by: 0
References: 4

Summary

No summary yet. VIRAT can read this paper and write a plain-language summary of its aim, methods, key findings and limitations.

Summarise with VIRAT Free to try with your Science Catalogs account.

Abstract

The exponential rise of online content has made social media a primary source of news consumption, yet it has also accelerated the spread of misinformation and manipulated media. Detecting such fake news requires analyzing both textual and visual cues, as deceptive content often combines misleading text with fabricated or unrelated images. This work proposes MultiFake, a multimodal deep learning framework designed to identify fake news by integrating linguistic and visual information. The framework employs DistilBERT, a lightweight transformer-based language model, to extract high-level semantic representations from textual data, and ResNet-18, a convolutional neural network, to capture discriminative visual features from accompanying images. The proposed system is trained on a balanced dataset of multimodal news samples, ensuring diverse representation of real and fake content. The study also incorporates visualization modules, including confusion matrices and ROC curves, for interpretability and validation of the framework's behavior. This multimodal approach aims to bridge the gap between textual and visual semantics, offering a scalable and adaptable solution for misinformation detection across digital platforms.

Fields of Study & Subjects

Misinformation and Its Impacts Spam and Phishing Detection Digital Media Forensic Detection

Publication details

Journal / source
International Conference on Intelligent Systems for Communication, IoT and Security (ICISCoIS)
Pages
1–7
Publication date
8 January 2026
Language
EN