Dynamic Perturbed Adaptive Method for Infinite Task-Conflicting Time Series

2025/05/17 by You Jiang, Xiaozhen Wang, You, Jiang +3
Computer Science · Neuroscience · #EEG and Brain-Computer Interfaces #FOS: Computer and information sciences #Machine Learning (cs.LG) #Machine Learning in Healthcare #Time Series Analysis and Forecasting

paper · pdf · doi:10.48550/arxiv.2505.11902

openalex publication_date 2025/05/17 · openalex created_date 2025/10/10 · openalex updated_date 2026/07/28

Abstract

We formulate time series tasks as input-output mappings under varying objectives, where the same input may yield different outputs. This challenges a model's generalization and adaptability. To study this, we construct a synthetic dataset with numerous conflicting subtasks to evaluate adaptation under frequent task shifts. Existing static models consistently fail in such settings. We propose a dynamic perturbed adaptive method based on a trunk-branch architecture, where the trunk evolves slowly to capture long-term structure, and branch modules are re-initialized and updated for each task. This enables continual test-time adaptation and cross-task transfer without relying on explicit task labels. Theoretically, we show that this architecture has strictly higher functional expressivity than static models and LoRA. We also establish exponential convergence of branch adaptation under the Polyak-Lojasiewicz condition. Experiments demonstrate that our method significantly outperforms competitive baselines in complex and conflicting task environments, exhibiting fast adaptation and progressive learning capabilities.

Citations

• TRACE: Time SeRies PArameter EffiCient FinE-tuning
• An Efficient Continual Learning Framework for Multivariate Time Series Prediction Tasks with Application to Vehicle State Estimation
• TimeFilter: Patch-Specific Spatial-Temporal Graph Filtration for Time Series Forecasting
• TimeRAF: Retrieval-Augmented Foundation model for Zero-shot Time Series Forecasting
• Kernel-U-Net: Multivariate Time Series Forecasting using Custom Kernels
• PaLM: Scaling Language Modeling with Pathways
• Learning Fast and Slow for Online Time Series Forecasting
• BitFit: Simple Parameter-efficient Fine-tuning for Transformer-based Masked Language-models
• LoRA: Low-Rank Adaptation of Large Language Models
• Informer: Beyond Efficient Transformer for Long Sequence Time-Series Forecasting
• Informer: Beyond Efficient Transformer for Long Sequence Time-Series Forecasting
• Language Models are Few-Shot Learners
• Parameter-Efficient Transfer Learning for NLP
• A Convergence Theory for Deep Learning via Over-Parameterization
• On First-Order Meta-Learning Algorithms
• An Overview of Multi-Task Learning in Deep Neural Networks
• Model-Agnostic Meta-Learning for Fast Adaptation of Deep Networks
• Linear Convergence of Gradient and Proximal-Gradient Methods Under the Polyak-Łojasiewicz Condition
• Optimization Methods for Large-Scale Machine Learning
• U-Net: Convolutional Networks for Biomedical Image Segmentation
• Stochastic Majorization-Minimization Algorithms for Large-Scale Optimization
• Prefix-Tuning: Optimizing Continuous Prompts for Generation
• Prefix-Tuning: Optimizing Continuous Prompts for Generation

Related

• Adaptive and Robust Multi-Task Learning
• Directed Structural Adaptation to Overcome Statistical Conflicts and Enable Continual Learning
• Persistent Test-time Adaptation in Recurring Testing Scenarios
• Adaptive Cascading Network for Continual Test-Time Adaptation
• A Model-based Approach to Assess Regular, Constant, and Progressive User Interface Adaptivity