Paper @ NIPS: Cyclades: Asynchronous Machine Learning
Paper @ OPT Workshop: Analysis of Async Stochastic Opt
Our paper on "Perturbed Iterate Analysis for Asynchronous Stochastic Optimization" has been accepted at OPT workshop at NIPS for presentation on Dec 11, 2015.
Paper @ NIPS: Parallel Correlation Clustering
Our paper on "Parallel Correlation Clustering on Big Graphs" has been accepted at NIPS 2015, Montreal, Quebec, Canada for poster on Dec 8, 2015. The paper and code are available via the publications page.
Paper @ DISCML Workshop: Parallel Correlation Clustering
Our paper on "Scaling up Correlation Clustering through Parallelism and Concurrency Control" has been accepted at DISCML workshop at NIPS for presentation on Dec 13, 2014.
Presentation @ BayLearn 2014
We presented our work on Parallel Double Greedy Submodular Maximization at BayLearn. This is our second presentation at BayLearn discussing the application of concurrency control techniques to parallelizing machine learning algorithms.
Paper @ NIPS: Parallel Submodular Maximization
Our paper on "Parallel Double Greedy Submodular Maximization" has been accepted at NIPS 2014, Montreal, Quebec, Canada for poster on Dec 9, 2014. We will be making the paper and code available on this website soon.
Big Learning Workshop
We're organizing the workshop on Big Learning: Advances in Algorithms and Data Management to be held at NIPS, Lake Tahoe, NV, on December 9 or 10, 2013. This year, the workshop aims to bring together the Large-scale Machine Learning and Database Systems communities to facilitate the cross-pollination of ideas.
Presentation @ BayLearn 2013
We presented our work on Optimistic Concurrency Control for Distributed Unsupervised Learning at BayLearn. In this work we explore the application of optimistic concurrency control (OCC) to the design of scalable, provably correct machine learning algorithms, and demonstrate its application to DP-means, a novel clustering algorithm.