Probabilistic Hashing for Scalable, Sustainable, and Secure Machine Learning

Speaker Organization:  Computer Science, Rice University Organizer:  Prof. Vishwanathan, Computer Science Abstract: Large scale machine learning and data mining applicationsare constantly dealing with datasets at TB scale and the anticipationis that soon it will reach PB level. At this scale, simple data miningoperations such as search, learning, and clustering becomechallenging.In this talk, we will start with a basic introduction to probabilistichashing (or fingerprinting) and the classical LSH algorithm.  Then Iwill present some of my recent adventures with probabilistic hashing.In particular, I will show how the idea of probabilistic hashing canbe used to significantly reduce the computations in classical machinelearning algorithms such as SVMs and Deep Learning (using our recentsuccess with asymmetric hashing for inner products). I will highlightthe computational bottleneck, i.e. the hashing time, and will show anefficient variant of…


Link to Full Article: Probabilistic Hashing for Scalable, Sustainable, and Secure Machine Learning