MATIC: A Lightweight Decision-Level Attention Model as a Strong Baseline for Multimodal Time-Series Classification
Big Data and Cognitive Computing · 17.11.2026 · 10 (11) · 393
DOI: 10.3390/bdcc10110393 ↗Kurz gefasst
Fachartikel in der Zeitschrift Big Data and Cognitive Computing, online erschienen am 17.11.2026. Verfasst von Rabin Nepal und 1 weiteren Autor:innen. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0).
Abstract (Originalsprache)
Multimodal time-series classification has been successfully applied to diverse domains, yet recent research has increasingly favored complex feature-level fusion architectures whose computational cost often limits practical deployment. In this work, we investigate whether such complexity is necessary through a comprehensive cross-domain empirical study of decision-level fusion. We introduce MATIC (Multimodal Attention-based Time-series Classifier), a lightweight framework in which channel-specific convolutional neural networks independently classify each signal stream and a self-attention module adaptively fuses the resulting class-probability distributions. This modular formulation decouples feature extraction from multimodal fusion, enabling efficient training, reduced model complexity, and straightforward scalability across heterogeneous modalities. We evaluate the proposed approach on six public benchmark datasets spanning emotion recognition, human activity recognition, and stock movement prediction. Across all benchmarks, MATIC achieves competitive performance while requiring substantially fewer trainable parameters (0.11–1.95 M) than many specialized multimodal architectures. Comparisons with conventional decision fusion strategies and architectural ablations show that adaptive attention outperforms static and MLP-based fusion in most settings, and robustness analyses characterize its sensitivity to noisy and missing inputs. These findings suggest that lightweight decision-level attention fusion constitutes a strong baseline for multimodal time-series classification and provides a simpler alternative to increasingly complex multimodal fusion architectures.
Zitieren
APA
Nepal, R., & Banerjee, B. (2026). MATIC: A Lightweight Decision-Level Attention Model as a Strong Baseline for Multimodal Time-Series Classification. Big Data and Cognitive Computing, 10(11), 393. https://doi.org/10.3390/bdcc10110393
Vancouver
Nepal R, Banerjee B. MATIC: A Lightweight Decision-Level Attention Model as a Strong Baseline for Multimodal Time-Series Classification. Big Data and Cognitive Computing. 2026;10(11):393. doi:10.3390/bdcc10110393
BibTeX
@article{Nepal2026,
author = {Nepal, Rabin and Banerjee, Bonny},
title = {MATIC: A Lightweight Decision-Level Attention Model as a Strong Baseline for Multimodal Time-Series Classification},
journal = {Big Data and Cognitive Computing},
year = {2026},
volume = {10},
number = {11},
pages = {393},
doi = {10.3390/bdcc10110393}
}
EUVIMED