cs.LGNov 16, 2026

Warm Starting Neural Dynamics Models From Few Jacobian Samples

Authors: Maxime Rochkoulets, Joppe De Jonghe, Mariya Ishteva

Organizations: Independent Researcher · NUMA, KU Leuven, Belgium

Abstract

Model-based reinforcement learning (MBRL) leverages a model of the environment, used for planning and generating samples without interacting with the real world. Motivated by evidence that informed initialization significantly improves the efficiency of MBRL, we propose TDec-Init, an algorithm to warm-start neural dynamics models using transition Jacobians. Using a decoupling algorithm based on the decomposition of a tensor of stacked Jacobian samples, we recover weights and flexible nonlinearities, and use them to initialize a feedforward neural network. In the low-data regime, we show on several benchmark environments that our approach outperforms random initialization and value training, and either matches or outperforms Sobolev training, while requiring fewer training steps between episodes.

Figures & tables

Explore similar work

CardsList
  1. Integrating Lagrangian Neural Networks into the Dyna Framework for Reinforcement Learning

    Mar 9, 2026Shreya Das, Kundan Kumar, Muhammad Iqbal +4Lagrangian Neural NetworksModel-Based RL

  2. IADD-TR: Intervention-Aware Dynamics Decoupling with Targeted Regularization for Model-Based Reinforcement Learning

    Aug 11, 2026Zefeng Liang, Jie Qiao, Ruichu Cai +2Reinforcement LearningRobust RL

  3. ProDVI: Programmatic Dynamics Priors for Value Network Initialization

    Aug 6, 2026Xinwei Liu, Junyuan Liang, Jianting Zhang +1Reinforcement LearningLLM-Guided RL