![[2602.21467] Geometric Priors for Generalizable World Models via Vector Symbolic Architecture](https://arxiv.org/static/browse/0.3.4/images/arxiv-logo-fb.png){kind=logo}
[2602.21467] Geometric Priors for Generalizable World Models via Vector Symbolic Architecture
Summary
This article presents a novel approach to world modeling in AI using Vector Symbolic Architecture (VSA) to enhance generalization and interpretability in neural networks.
Why It Matters
Understanding how neural systems learn representations is crucial for advancing AI and neuroscience. This research introduces a structured method that improves sample efficiency and generalization in world models, which is vital for real-world applications in planning and reasoning.
Key Takeaways
- Introduces a generalizable world model based on Vector Symbolic Architecture.
- Achieves 87.5% zero-shot accuracy on unseen state-action pairs.
- Demonstrates 53.6% higher accuracy on long-term predictions compared to traditional models.
- Shows 4x higher robustness to noise than MLP baselines.
- Highlights the importance of structured representations for data efficiency and interpretability.
Computer Science > Machine Learning arXiv:2602.21467 (cs) [Submitted on 25 Feb 2026] Title:Geometric Priors for Generalizable World Models via Vector Symbolic Architecture Authors:William Youngwoo Chung, Calvin Yeung, Hansen Jin Lillemark, Zhuowen Zou, Xiangjian Liu, Mohsen Imani View a PDF of the paper titled Geometric Priors for Generalizable World Models via Vector Symbolic Architecture, by William Youngwoo Chung and 5 other authors View PDF HTML (experimental) Abstract:A key challenge in artificial intelligence and neuroscience is understanding how neural systems learn representations that capture the underlying dynamics of the world. Most world models represent the transition function with unstructured neural networks, limiting interpretability, sample efficiency, and generalization to unseen states or action compositions. We address these issues with a generalizable world model grounded in Vector Symbolic Architecture (VSA) principles as geometric priors. Our approach utilizes learnable Fourier Holographic Reduced Representation (FHRR) encoders to map states and actions into a high dimensional complex vector space with learned group structure and models transitions with element-wise complex multiplication. We formalize the framework's group theoretic foundation and show how training such structured representations to be approximately invariant enables strong multi-step composition directly in latent space and generalization performances over various experiments. On a...
