Adaptive Multi-Modal Phishing Detection (AMPD) Framework Using Attention-Based Dynamic Fusion

DOI

10.2991/978-94-6239-789-7_25How to use a DOI?

Keywords

Phishing detection; multi-modal feature fusion; attention mechanism; explainable AI; mobile security; deep learning; adaptive resource allocation; dynamic weighting; cross-modal SHAP; behavioral analysis

Abstract

Phishing attacks are a major issue in cybersecurity, and modern detection systems have a weakness of disjointed feature integration, which restricts their ability to handle advanced multi-vector campaigns. The current paper introduces a novel Adaptive Multi-Mode Phishing Detection (AMPD) system that combines URL lexical features, visual webpage features and behavioral interaction patterns in the form of an attention-based dynamic fusion system. Unlike the proposed conventional methods that use fixed feature weights, model uses a context-sensitive attention design that recalibrates modality input dynamically, depending on computation limitations and threat properties. The framework includes a lightweight threat-aware pre-filter of. resource-efficient mobile-optimized compressed visual hashing triage. convolutional neural networks and privacy-preserving behavioral proxies to interaction analysis. It is proven by experimental assessment of a multi-modal dataset of 12,000 phishing and 10,000 legitimate sites. that the AMPD framework gets 98.67% accuracy and 47ms mean inference time on mobile devices, which surpasses single-modality baselines, offering around 5.2 times faster inference and an accuracy improvement of 8.4%, while preserving explainability through cross-modal SHAP analysis. The Findings prove that attention-based dynamic fusion is viable to solve the accuracy-efficiency trade-off of multi-modal phishing detection, which offers an effective solution for real-time phishing detection.

Copyright

© 2026 The Author(s)

Open Access

Open Access This chapter is licensed under the terms of the Creative Commons Attribution-NonCommercial 4.0 International License (http://creativecommons.org/licenses/by-nc/4.0/), which permits any noncommercial use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license and indicate if changes were made.

Download article (PDF)

TY  - CONF
AU  - Sam Sharma
AU  - Saksham Singh
AU  - Paramjeet Singh
AU  - Shagun Kumar
AU  - Punit Mittal
AU  - Swati Sharma
PY  - 2026
DA  - 2026/10/08
TI  - Adaptive Multi-Modal Phishing Detection (AMPD) Framework Using Attention-Based Dynamic Fusion
BT  - Proceedings of theInternational Conference on Innovation and Practices in Artificial Intelligence (ICIPAI 2026)
PB  - Atlantis Press
SP  - 345
EP  - 379
SN  - 1951-6851
UR  - https://doi.org/10.2991/978-94-6239-789-7_25
DO  - 10.2991/978-94-6239-789-7_25
ID  - Sharma2026
ER  -