Eric P. Xing

TL;DR: This work uses near-orthogonality to characterize “diversity” and impose angular constraints (ACs) on the components of LSMs to promote diversity and demonstrates that ACs improve generalization performance of LS Ms and outperform other diversitypromoting approaches.

TL;DR: A novel adversarial domain adaptation framework that is guided by Wasserstein distance, therefore with better stability and convergence than typical adversarial methods is built and demonstrated that WGAN guided domain adaptation obtains a state-of-the-art performance for the joint optic disc-and-cup segmentation in fundus images.

TL;DR: The Triple-Basis Estimator (3BE) as mentioned in this paper is the first nonparametric estimator that can scale to massive data-sets, and it has been shown to have promising performance in terms of prediction speed and reduction in error.

TL;DR: Haploi as mentioned in this paper is a hierarchical nonparametric Bayesian model to address the problem of how many clusters in a co-clustering scenario, in which one needs to solve multiple clustering problems simultaneously because of the presence of common centroidids shared by clusters.

TL;DR: A scalable Gaussian process model is presented for identifying and characterizing smooth multidimensional changepoints, and automatically learning changes in expressive covariance structure, and is demonstrated on numerical and real world data.