analiza aiqbench
DCIts offers a novel deep-learning architecture that balances forecasting accuracy with intrinsic interpretability by explicitly factorizing interactions into source selection and signed contribution components. This approach enables direct inspection of effective connectivity and lag-resolved patterns in complex multivariate time series without sacrificing predictive performance. The method is particularly relevant for domains requiring causal understanding of nonlinear dynamics, such as applied physics and complex systems analysis.
Interpretable deep convolutional model for nonlinear multivariate time series in complex systems
arXiv:2501.04339v2 Announce Type: replace-cross Abstract: We introduce the Deep Convolutional Interpreter for Time Series (DCIts), a deep-learning architecture for nonlinear multivariate time series that provides sample-specific, locally interpretable descriptions of the underlying interaction structure. Unlike standard black-box forecasters, DCIts learns a time- and lag-dependent transition tensor explicitly factorized into two components: a Focuser, which selects relevant source series and time lags via a sparse masking mechanism, and a Modeler, which assigns signed coefficients to these selected interactions. This decomposition yields a local lag-adjacency structure and signed source-lag contributions for every forecast instance, enabling direct inspection of effective connectivity; when higher-order branches are activated, the same framework yields order-resolved elementwise polynomial contributions. Architecturally, DCIts uses a diverse bank of convolutional filters to capture temporal and cross-variable dependencies, which are mapped through a bottleneck network to the transition tensor. On controlled benchmark datasets with a known interaction structure, we demonstrate that DCIts achieves competitive forecasting error relative to a strong interpretable baseline while recovering stable, signed, lag-resolved interaction patterns. The framework thus prioritizes intrinsic interpretability, using forecasting accuracy as a faithfulness constraint rather than the sole objective.
Abstract:We introduce the Deep Convolutional Interpreter for Time Series (DCIts), a deep-learning architecture for nonlinear multivariate time series that provides sample-specific, locally interpretable descriptions of the underlying interaction structure. Unlike standard black-box forecasters, DCIts learns a time- and lag-dependent transition tensor explicitly factorized into two components: a Focuser, which selects relevant source series and time lags via a sparse masking mechanism, and a Modeler, which assigns signed coefficients to these selected interactions. This decomposition yields a local lag-adjacency structure and signed source-lag contributions for every forecast instance, enabling direct inspection of effective connectivity; when higher-order branches are activated, the same framework yields order-resolved elementwise polynomial contributions. Architecturally, DCIts uses a diverse bank of convolutional filters to capture temporal and cross-variable dependencies, which are mapped through a bottleneck network to the transition tensor. On controlled benchmark datasets with a known interaction structure, we demonstrate that DCIts achieves competitive forecasting error relative to a strong interpretable baseline while recovering stable, signed, lag-resolved interaction patterns. The framework thus prioritizes intrinsic interpretability, using forecasting accuracy as a faithfulness constraint rather than the sole objective.
Machine Learning (stat.ML); Machine Learning (cs.LG); Applied Physics (physics.app-ph)
arXiv:2501.04339 [stat.ML]
(or arXiv:2501.04339v2 [stat.ML] for this version)
https://doi.org/10.48550/arXiv.2501.04339 Focus to learn more arXiv-issued DOI via DataCite
https://doi.org/10.1063/5.0325209 Focus to learn more DOI(s) linking to related resources
arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website.
Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.
Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs.
Text integral sub licența deschisă a publicației · 288 cuvinte
Subiecte conexe
Fuzzy-MoE: Interpretable Regime-Conditioned Expert Routing for Non-Stationary Multivariate Time Series Forecasting
arXiv · 24 aug. 2026
Frequency Selective Neural Networks as a Foundation Architecture for Time Series Learning
arXiv · 1 sept. 2026
A Data-Driven Framework for Unsupervised Monitoring of Transmission Systems Using End-of-Line Testing Data: A Case Study at Ford Motor Company
arXiv · 7 oct. 2026
Causally-Constrained Probabilistic Forecasting for Time-Series Anomaly Detection
arXiv · 15 oct. 2026