Statistical Learning
Advancing Cross-domain Discriminability in Continual Learning of Vision-Language Models
During the CL of VLMs, we need not only to prevent the catastrophic forgetting on incrementally learned knowledge but also to preserve the zero-shot ability of VLMs.
A Similarity-preserving Network Trained on Transformed Images Recapitulates Salient Features of the Fly Motion Detection Circuit
Yanis Bahroun, Dmitri Chklovskii, Anirvan Sengupta
Neural Information Processing Systems http://nips.cc/
Beyond Log-concavity: Provable Guarantees for Sampling Multi-modal Distributions using Simulated Tempering Langevin Monte Carlo
Holden Lee, Andrej Risteski, Rong Ge
Neural Information Processing Systems http://nips.cc/
Sampling from Gaussian Process Posteriors using Stochastic Gradient Descent
In general, this has a cubic cost in dataset size and is sensitive to conditioning.
