mlsanityData to 14 Oct 2026

Dynamic Perturbed Adaptive Method for Infinite Task-Conflicting Time Series

arXiv 2505.11902 ↗First posted 17 May 2025cs.LGcited by 0

Not on the map. Why?

Abstract (arXiv)

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.

Key facts. No citing papers in the map yet (data through 14 Oct 2026). Builds on 25 of its 32 references that are in the map. First posted on arXiv 17 May 2025.

Builds on

25 of 32 references.

References · 25

  1. 2020Language Models are Few-Shot Learnerscited by 32,419
  2. 2015U-Net: Convolutional Networks for Biomedical Image Segmentationcited by 19,713
  3. 2021LoRA: Low-Rank Adaptation of Large Language Modelscited by 14,380
  4. 2022PaLM: Scaling Language Modeling with Pathwayscited by 5,326
  5. 2017Model-Agnostic Meta-Learning for Fast Adaptation of Deep Networkscited by 4,897
  6. 2019Parameter-Efficient Transfer Learning for NLPcited by 4,236
  7. 2021Prefix-Tuning: Optimizing Continuous Prompts for Generationcited by 3,915
  8. 2020Informer: Beyond Efficient Transformer for Long Sequence Time-Series Forecastingcited by 2,169
  9. 2018Continual Lifelong Learning with Neural Networks: A Reviewcited by 1,330
  10. 2021BitFit: Simple Parameter-efficient Fine-tuning for Transformer-based Masked Language-modelscited by 1,245
  11. 2016Optimization Methods for Large-Scale Machine Learningcited by 1,150
  12. 2018On First-Order Meta-Learning Algorithmscited by 1,018
13 more
  1. 2017An Overview of Multi-Task Learning in Deep Neural Networkscited by 803
  2. 2018A Convergence Theory for Deep Learning via Over-Parameterizationcited by 787
  3. 2016Linear Convergence of Gradient and Proximal-Gradient Methods Under the Polyak-Lojasiewicz Conditioncited by 539
  4. 2022Transformers in Time Series: A Surveycited by 497
  5. 2019Fast and Flexible Multi-Task Classification Using Conditional Neural Adaptive Processescited by 131
  6. 2025TimeFilter: Patch-Specific Spatial-Temporal Graph Filtration for Time Series Forecastingcited by 64
  7. 2022Learning Fast and Slow for Online Time Series Forecastingcited by 45
  8. 2013Stochastic Majorization-Minimization Algorithms for Large-Scale Optimizationcited by 28
  9. 2024TimeRAF: Retrieval-Augmented Foundation model for Zero-shot Time Series Forecastingcited by 17
  10. 2024Channel-Aware Low-Rank Adaptation in Time Series Forecastingcited by 12
  11. 2025TRACE: Time SeRies PArameter EffiCient FinE-tuningcited by 6
  12. 2024Kernel-U-Net: Multivariate Time Series Forecasting using Custom Kernelscited by 4
  13. 2025An Efficient Continual Learning Framework for Multivariate Time Series Prediction Tasks with Application to Vehicle State Estimationcited by 1

Built on by

Not cited yet.