Eric P. Xing

TL;DR: In this paper, the authors investigate the relationship between the frequency spectrum of image data and the generalization behavior of convolutional neural networks (CNN) and find that high frequency components are almost imperceptible to a human.

TL;DR: The created PathVQA dataset represents the first one addressing the pathology VQA problem and the results demonstrate the effectiveness of the proposed learning-by-ignoring method and cross-modal self-supervised learning methods.

TL;DR: This paper explores salient questions about user interests, conversations and friendships in the Facebook social network, using a novel latent space model that integrates several data types, allowing for the interplay between user friendships, interests, and higher-order network-wide social trends on Facebook.

TL;DR: A new principled framework for estimating the structure of undirected graphical models from such multivariate (or multi-attribute) nodal data is proposed and a method that efficiently maximizes this objective is developed.

TL;DR: The evaluation shows that for a number of ML applications, Orion can parallelize a serial program while preserving critical dependences and thus achieve a significantly faster convergence rate than data-parallel programs and a matching convergence rate and comparable computation throughput to state-of-the-art manual parallelizations including model-par parallel programs.