Research map / Continual learning / Online and replay-based
Continual Slow-and-Fast Adaptation of Latent Neural Dynamics (CoSFan): Meta-Learning What-How & When to Adapt
Read the paperCheck your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's abstract only. Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Proposes CoSFan, a continual slow-and-fast adaptation method for latent neural dynamics that meta-learns what, how, and when to adapt to shifting time-series dynamics while avoiding catastrophic forgetting.
Problem
Forecasting high-dimensional time-series whose underlying dynamics shift over time, where naive retraining causes catastrophic forgetting of previously learned systems.
Approach
Combines slow and fast adaptation in a latent neural dynamics model and meta-learns what to adapt, how to adapt, and when to adapt, so the model can handle distribution shifts in dynamics without forgetting earlier ones.
Closest papers by meaning
- Learning to Continually Learn with the Bayesian Principle (ICML 2024 · similarity 0.89)
- Meta-Dynamical State Space Models for Integrative Neural Data Analysis (ICLR 2025 · similarity 0.89)
- Learning Hybrid Dynamics Models with Simulator-Informed Latent States (AAAI 2024 · similarity 0.87)
- Neural Context Flows for Meta-Learning of Dynamical Systems (ICLR 2025 · similarity 0.87)
- Modeling Latent Neural Dynamics with Gaussian Process Switching Linear Dynamical Systems (NeurIPS 2024 · similarity 0.87)
- Learning dynamics in linear recurrent neural networks (ICML 2025 · similarity 0.86)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.