Abstract -- Detection of execution anomalies is very important for the maintenance, development, and performance refinement of large scale distributed systems. Execution anomalies ...
The paper examines different possibilities to take advantage of the taxonomic organization of a thesaurus to improve the accuracy of classifying new words into its classes. The re...
Recent work has exploited boundedness of data in the unsupervised learning of new types of generative model. For nonnegative data it was recently shown that the maximum-entropy ge...
We give a new model of learning motivated by smoothed analysis (Spielman and Teng, 2001). In this model, we analyze two new algorithms, for PAC-learning DNFs and agnostically learn...
Adam Tauman Kalai, Alex Samorodnitsky, Shang-Hua T...
The supremacy of n-gram models in statistical language modelling has recently been challenged by parametric models that use distributed representations to counteract the difficult...