Efficient Online Prediction for High-Dimensional Time Series via Joint Tensor Tucker Decomposition

2024-12-31 · Source: JMLR · Field: Technology & Digital — Artificial Intelligence & Machine Learning, Data Science & Analytics · Depth: Expert, quick

Summary

Zhenting Luan, Defeng Sun, Haoning Wang, and Liping Zhang introduce a novel online algorithm, Tensor Online Prediction Algorithm (TOPA), designed for real-time prediction in high-dimensional streaming time series. Published in 2025, TOPA addresses the challenge of continuously tracking evolving latent statistical patterns in data streams, crucial for applications like traffic congestion control and wireless channel resource allocation. The algorithm utilizes tensor factorization and updates its predictor with low complexity to adapt to time-evolving data. An automatically adaptive version, TOPA-AAW, is also presented to reduce the negative impact of outdated data. Simulation results indicate that TOPA-AAW achieves prediction accuracy comparable to conventional offline tensor prediction methods, but with significantly faster performance during extended online prediction tasks.

Key takeaway

For research scientists developing real-time control systems, TOPA-AAW offers a robust solution for online prediction of high-dimensional streaming time series. You should consider integrating this tensor factorization-based approach to achieve high prediction accuracy while significantly improving computational speed compared to traditional offline methods, especially for long-term deployments. This can enhance system responsiveness and resource optimization.

Key insights

TOPA-AAW offers efficient, accurate online prediction for high-dimensional streaming tensor time series via adaptive tensor factorization.

Principles

• Online updates adapt to evolving data.
• Tensor factorization handles high dimensionality.
• Adaptive weighting mitigates stale data impact.

Method

TOPA employs tensor factorization for streaming tensor time series prediction, updating the predictor online with low complexity. TOPA-AAW adds automatic adaptive weighting to reduce stale data influence.

In practice

• Apply TOPA-AAW for real-time traffic control.
• Use TOPA-AAW in wireless resource allocation.
• Implement for continuous high-dimensional data streams.

Topics

• Online Prediction
• Tensor Factorization
• High-Dimensional Time Series
• Streaming Data
• Adaptive Algorithms

Code references

• JmlrOrg/v26

Best for: Research Scientist, AI Researcher, AI Scientist, Machine Learning Engineer

Related on AIssential

• Efficient frequent directions algorithms for approximate decomposition of matrices and higher-order tensors — New frequent directions algorithms enhance low-rank matrix and tensor approximations for efficiency and accuracy.
• Self-Adaptive Scale Handling for Forecasting Time Series with Scale Heterogeneity — Self-Adaptive Scale-handling (AS) improves time series forecasting by adaptively managing scale heterogeneity without losing signal or amplifying errors.
• Robust Neural Tucker Factorization with Bias Correction and Adaptive Initialization — KaBiN improves neural Tucker factorization for HDI tensor completion via Kaiming initialization and bias correction.
• High-Dimensional Robust Change-Point Detection via Angular Kernel Statistics — A new angular kernel scan framework robustly detects high-dimensional change-points in low sample size, even with heavy-tailed data.
• Guaranteed Nonconvex Low-Rank Tensor Estimation via Scaled Gradient Descent — ScaledGD offers provably fast, condition-number-independent convergence for low-rank tensor estimation using t-SVD.
• A Typed Tensor Language for Federated Learning — The language formalizes federated learning computations, enabling fixed-dimensional shared state communication for efficiency.

Open in AIssential →

Editorial summary, takeaway, and curation by AIssential. Original article published by JMLR.