PILLAR 2 · 6 PAPERS · NO SEARCH RUN YET
World models & representation learning
Non-generative world models that predict in latent space, and graph or hyperbolic representations suited to hierarchical enterprise structure.
Search string (see Methods for the per-database form)
("joint-embedding predictive" OR "JEPA" OR "world model" OR "latent dynamics" OR "latent prediction") AND ("graph" OR "relational" OR "hyperbolic" OR "Poincar*" OR "multi-agent") AND ("self-supervised" OR "action-conditioned" OR "predict*" OR "anomaly" OR "safety")Included papers
| Paper | Year | Venue | Contribution | Methods | Relevance | Tags |
|---|---|---|---|---|---|---|
| Graph-level representation learning with joint-embedding predictive architectures Skenderi G.; Li H.; Tang J.; Cristani M. |
2025 | TMLR | Graph-JEPA: masked sub-graph latent prediction with a hyperbolic coordinate target. | world-model self-supervised hyperbolic-geometry graph-learning | baseline RQ1 RQ4 seed | |
| Self-supervised learning from images with a joint-embedding predictive architecture Assran M.; Duval Q.; Misra I.; Bojanowski P.; Vincent P.; Rabbat M.; LeCun Y.; Ballas N. |
2023 | CVPR | I-JEPA: the first practical JEPA with an EMA target encoder and masked latent prediction. | world-model self-supervised | RQ1 seed | |
| A path towards autonomous machine intelligence LeCun Y. |
2022 | OpenReview | Proposes JEPA world models that predict in representation space rather than reconstructing inputs. | world-model energy-based | RQ1 seed | |
| Hyperbolic graph convolutional neural networks Chami I.; Ying R.; Ré C.; Leskovec J. |
2019 | NeurIPS | Message passing in hyperbolic space for hierarchical graphs; the encoder family behind the context encoder. | hyperbolic-geometry graph-learning | baseline RQ1 RQ4 seed | |
| Poincaré embeddings for learning hierarchical representations Nickel M.; Kiela D. |
2017 | NeurIPS | Embeds hierarchies in the Poincaré ball; the distance PreCog uses for its safe-manifold energy. | hyperbolic-geometry representation-learning | RQ1 RQ4 seed | |
| VICReg: variance-invariance-covariance regularization for self-supervised learning Bardes A.; Ponce J.; LeCun Y. |
2022 | ICLR | Variance and covariance regularisers that prevent representational collapse in joint-embedding training. | self-supervised regularisation | RQ1 seed |
GAPS THIS PILLAR LEAVES OPEN
- No included paper in this pillar addresses latency / jitter aware
- No included paper in this pillar addresses time-to-irreversibility
- No included paper in this pillar addresses edge deployment
- No included paper in this pillar addresses human handoff
- No included paper in this pillar addresses cross-domain evaluation