RoboKA: KAN Informed Multimodal Learning for RoboCall Surveillance System

2026-04-30

AI Summary

This study addresses the limitations in robocall monitoring research caused by the lack of publicly available datasets and the inadequacy of existing methods against multimodal adversarial strategies. To bridge this gap, the authors introduce Robo-SAr, the first synthetic multidimensional adversarial dataset encompassing psychological manipulation, emotional speech, and voice cloning. They further propose RoboKA, a novel multimodal fusion framework based on Kolmogorov–Arnold Networks (KANs), which aligns acoustic and linguistic representations through cross-modal contrastive learning and uniquely models their structured nonlinear interactions using KANs. Experimental results demonstrate that RoboKA significantly outperforms current unimodal and multimodal baselines in both in-domain and out-of-domain settings, achieving state-of-the-art performance in recall and F1 score.

Abstract

Wide exploration on robocall surveillance research is hindered due to limited access to public datasets, due to privacy concerns. In this work, we first curate Robo-SAr, a synthetic robocall dataset designed for robocall surveillance research. Robo-SAr comprises of ~200 unwanted and ~1200 legitimate synthetic robocall samples across three realistic adversarial axes: psycholinguistics-manipulated transcripts, emotion-eliciting speech, and cloned voices. We further propose RoboKA, a Kolmogorov-Arnold Network (KAN)-based multimodal fusion framework designed to model structured nonlinear interactions between acoustic and linguistic cues that characterize diverse adversarial robocall strategies. RoboKA first leverages cross-modal contrastive learning to align latent modality representations and feeds the resulting embeddings to a KAN-projection head for final classification. We benchmark RoboKA against strong unimodal and multimodal baselines in both in-domain and out-of-domain setups, finding RoboKA to surpass all baselines in terms of recall and F1-score.

Problem

• robocall surveillance
• privacy concerns
• public datasets
• synthetic data
• multimodal learning

Innovation

• Kolmogorov-Arnold Network
• multimodal fusion
• robocall detection
• synthetic dataset
• cross-modal contrastive learning