World Models for Cross-Machine CNC Transfer under Partial Sensor Overlap
cs.LG, cs.AI
Submitted: 2026-09-13
Updated: 2026-09-23
Comments: Code, configuration files, the twenty candidate specifications, and the audit scripts are available at : https://github.com/ostertagmatthieu- dev/saac- jepa ; Animated versions of the schematics are on the project page: https://ostertagmatthieu-dev.github.io/saac-jepa/
Project page: https://ostertagmatthieu-dev.github.io/saac-jepa
License: http://creativecommons.org/licenses/by/4.0/
The gist: Cross-machine deployment of industrial world models requires transfer across changes in dynamics, sensing interfaces, sampling regimes, and control units.
Terminology
Abstract
Cross-machine deployment of industrial world models requires transfer across changes in dynamics, sensing interfaces, sampling regimes, and control units. We study a schema-adaptive action-conditioned Joint-Embedding Predictive Architecture (SAAC-JEPA) for CNC dynamics, where the source machine has 17 canonical sensor channels and the target shares only 10. Evaluation uses group-disjoint source splits, source-only normalization, held-out self-supervised validation, unit audits, and a sealed target test after model locking. Across five seeds, JEPA pretraining gives no clean-source forecasting gain: scratch and pretrained-body models obtain RMSE=0.811 plus or minus0.022 and 0.813 plus or minus0.022. A source-only search over 20 candidates selects a schema-consistent action-conditioned JEPA after seven-seed stability checks. On the confirmatory target pass, the locked model reaches zero-shot RMSE=0.546, R 2=0.012, and NLL=0.52, outperforming persistence but not RevIN-equipped PatchTST and iTransformer baselines (0.503 and 0.498). A pre-declared paired ablation shows that RevIN in the same architecture improves RMSE to 0.495 plus or minus0.004 over three seeds, but degrades target calibration (NLL=20.6) on stationary context windows. A pre-lock adaptation sweep further reduces RMSE to 0.520 with limited target support. These results show that source-domain forecasting accuracy alone is insufficient to assess industrial predictive representations, and that cross-machine adaptation under partial sensor overlap is a distinct evaluation axis.
Sources
- Revisiting Feature Prediction for Learning Visual Representations from Video
- Learning Actionable World Models for Industrial Process Control
Related papers
- Polynomial-Augmented Neural Networks (PANNs) with Weak Orthogonality Constraints for Enhanced Function and PDE Approximation
- AIRL-S: Unifying Reinforcement Learning and Search-Based Test-Time Scaling via Adversarial Inverse Reinforcement Learning
- Transformers as Bayesian In-Context Experimenters: Smoothness-Adaptive Efficient ATE Estimation
- Convergence issues in Relational Concept Analysis based on AOC-posets
- Beliefs Beyond Posteriors: Local-Consistency Optimisation for Bayesian Neural Networks
- Understanding Diffusion Models via Ratio-Based Function Approximation with SignReLU Networks