Lost in Latent Space: Examining failures of disentangled models at combinatorial generalisation
–Neural Information Processing Systems
Recent research has shown that generative models with highly disentangled representations fail to generalise to unseen combination of generative factor values.
Neural Information Processing Systems
Feb-8-2026, 14:04:04 GMT
- Country:
- North America > United States
- Colorado (0.04)
- Europe > United Kingdom
- England > Cambridgeshire > Cambridge (0.04)
- Asia > Japan
- Honshū > Tōhoku > Iwate Prefecture > Morioka (0.04)
- North America > United States
- Genre:
- Research Report > New Finding (0.46)
- Technology: