Representation ↔ Generation
Learning methods for scientific discovery and real-world problems.
01 / REPRESENTATION LEARNING
Representation
How can models learn useful representations of complex structures?
- DEL Structural representations from physical response distributions.
- Vision Models / GraphAbstract Understanding graph structure through visual perception.
- Spectral Augmentation Revisited Understanding what makes contrastive representations effective.
- Protein Language Models Improving protein representations through negative sample mining.
Representation ← Generation MolEmb Adapting generative models into representation learners. Representation → Generation Latent JEPA Learning predictive representations for reasoning and generation.
02 / GENERATION & REASONING
Generation
How can representations and learning methods improve generation and reasoning?