CASC: Causal Adversarial Subspace Clustering for Multivariate Spatiotemporal Data

2026-07-23

AI Summary

Existing deep subspace clustering methods struggle to model the causal dependencies, local spatial interactions, and long-range temporal dynamics inherent in multivariate spatiotemporal data, often relying on the unrealistic assumption of static subspace structures. To address these limitations, this work proposes the Causal-aware Subspace Clustering (CASC) framework, which integrates causal modeling into deep subspace clustering for the first time. CASC employs a U-Net–style adversarial clustering architecture coupled with FAConvLSTM and a graph attention Transformer-based self-expressive network to jointly capture complex spatiotemporal dependencies. It further introduces a causal subspace preservation loss and a dynamic temporal subspace evolution loss to uncover latent causal mechanisms under non-stationary conditions. Evaluated on tasks including sea ice monitoring, disease spread tracking, and neurodegenerative progression analysis, CASC substantially outperforms state-of-the-art methods, revealing evolutionarily coherent spatiotemporal clusters with meaningful causal interpretations.

Abstract

Deep subspace clustering plays a critical role in applications involving multivariate spatiotemporal data, such as sea ice monitoring, disease spread analysis, and tracking neuro-degeneration over time. Despite recent advances, existing methods primarily rely on geometric self-expressiveness, assume static subspace structures, and often fail to capture causal dependencies, local spatial interactions, and long-range temporal dynamics inherent in complex spatiotemporal systems. To address these limitations, we propose a novel Causal Adversarial Subspace Clustering (CASC) framework for discovering evolving latent regimes in high-dimensional spatiotemporal data. CASC integrates a U-Net-inspired deep adversarial clustering architecture with stacked FAConvLSTM layers to preserve spatial and temporal structure while learning robust latent representations. A graph attention transformer-based self-expressive network is introduced to jointly model local spatial relationships, global dependencies, and long-range temporal interactions. Furthermore, we propose two new learning objectives: (1) a Causal Subspace Preservation Loss that aligns self-expression coefficients with latent causal relationships, encouraging clusters to reflect underlying causal processes rather than simple feature similarity, and (2) a Dynamic Temporal Subspace Evolution Loss that captures evolving subspace structures and temporal regime transitions in nonstationary environments. Together, these components transform deep subspace clustering from a correlation-driven paradigm into a causal-temporal regime discovery framework.

Problem

• causal dependencies
• spatiotemporal data
• subspace clustering
• temporal dynamics
• spatial interactions

Innovation

• Causal Adversarial Subspace Clustering
• Spatiotemporal Data
• Graph Attention Transformer
• FAConvLSTM
• Causal Subspace Preservation