BRo-JEPA: Learning Modular Transformations in Latent Space
Quick summary
arXiv:2606.01372v2 Announce Type: replace-cross Abstract: Can neural networks learn algebraic rules from visual inputs, or do they merely fit observed patterns? We study this question using MNIST (or EMNIST letters) as states and modular arithmetic operations as actions in a JEPA-style world model. Standard supervised and JEPA baselines with operation embeddings achieve high accuracy on seen operations but fail to extrapolate reliably to unseen operations. We propose BRo-JEPA, a world model with a block-rotation predictor that represents arithmetic operations as rotations, resulting in the cyc
Key takeaways
- arXiv:2606.01372v2 Announce Type: replace-cross Abstract: Can neural networks learn algebraic rules from visual inputs, or do they merely fit observed patterns?
- We study this question using MNIST (or EMNIST letters) as states and modular arithmetic operations as actions in a JEPA-style world model.
- Standard supervised and JEPA baselines with operation embeddings achieve high accuracy on seen operations but fail to extrapolate reliably to unseen operations.
Why it matters
“BRo-JEPA: Learning Modular Transformations in Latent Space” highlights the need for repeatable measurement rather than a single impressive demonstration. Independent validation across datasets and clearly stated limitations determine whether a result can guide product decisions.
Member comments