Interpretability in Deep Time Series Models Demands Semantic Alignment

2026-02-02

AI Summary

This work addresses the disconnect between high-performing deep time series models and human semantic understanding of temporal phenomena, a gap that undermines trustworthy deployment due to their black-box nature. We formally introduce the notion of “semantic alignment,” arguing that model predictions should be expressed through user-interpretable variables, mediated by mechanisms that satisfy spatiotemporal constraints, and remain consistent over time. To this end, we propose an interpretable modeling framework that jointly enforces semantic alignment, respects user-specified spatiotemporal dependencies, and ensures dynamic consistency—thereby overcoming the limitations of existing static explanation methods. Our approach lays a theoretical foundation for building deep time series systems that are not only accurate but also comprehensible to human users.

Technology Category

• Knowledge Representation and Reasoning: Geometric, Spatial, and Temporal Reasoning
• Computer Vision: Interpretability, Explainability, and Transparency
• Machine Learning: Time-Series/Data Streams

Application Category

• Semantics and Knowledge: Data modeling to support human-machine intelligence, including LLMs agents, intelligent system behavior, explanations, and user-friendly interactions
• User Modeling, Personalization and Recommendation: Explainable and interpretable methods for personalization
• Graph Algorithms and Modeling for the Web: Foundation models and LLMs for Web-related graphs

Abstract

Deep time series models continue to improve predictive performance, yet their deployment remains limited by their black-box nature. In response, existing interpretability approaches in the field keep focusing on explaining the internal model computations, without addressing whether they align or not with how a human would reason about the studied phenomenon. Instead, we state interpretability in deep time series models should pursue semantic alignment: predictions should be expressed in terms of variables that are meaningful to the end user, mediated by spatial and temporal mechanisms that admit user-dependent constraints. In this paper, we formalize this requirement and require that, once established, semantic alignment must be preserved under temporal evolution: a constraint with no analog in static settings. Provided with this definition, we outline a blueprint for semantically aligned deep time series models, identify properties that support trust, and discuss implications for model design.

Problem

• interpretability
• deep time series models
• semantic alignment
• human reasoning
• temporal evolution

Innovation

• semantic alignment
• interpretability
• deep time series models
• temporal evolution
• user-dependent constraints