Statistical Learning
Inference via Interpolation: Contrastive Representations Provably Enable Planning and Inference
Given time series data, how can we answer questions like "what will happen in the future?" and "how did we get here?"
Learning Energy-Based Prior Model with Diffusion-Amortized MCMC Peiyu Y u
Latent space Energy-Based Models (EBMs), also known as energy-based priors, have drawn growing interests in the field of generative modeling due to its flexibility in the formulation and strong modeling power of the latent space. However, the common practice of learning latent space EBMs with non-convergent short-run MCMC for prior and posterior sampling is hindering the model from further progress; the degenerate MCMC sampling quality in practice often leads to degraded generation quality and instability in training, especially with highly multi-modal and/or high-dimensional target distributions. To remedy this sampling issue, in this paper we introduce a simple but effective diffusion-based amortization method for long-run MCMC sampling and develop a novel learning algorithm for the latent space EBM based on it. We provide theoretical evidence that the learned amortization of MCMC is a valid long-run MCMC sampler.
Precise Asymptotic Generalization for Multiclass Classification with Overparameterized Linear Models
We fully resolve the conjecture posed in Subramanian et al. (2022),
Sample and Computationally Efficient Robust Learning of Gaussian Single-Index Models
We study the task of learning SIMs in the agnostic (a.k.a.
Towards robust and generalizable representations of extracellular data using contrastive learning
In CEED, we assume that waveforms are already extracted from an extracellular recording. Each waveform is then passed through our stochastic view generation module, where different views are obtained by applying transformations.