Title: Building an Algorithmic Toolkit for Data Science
Abstract: With data volume recognized as a primary factor behind advances in machine learning and data science, scaling statistical methods to take advantage of our largest datasets has become a central challenge.
He will discuss His efforts to design general purpose algorithmic tools that address this challenge. In particular, he will highlight his research on fast dimensionality reduction algorithms, which accelerate statistical methods by first compressing data to a small set of relevant examples or features. He will explain how he has deepened our theoretical understanding of dimensionality reduction, leading to provably accurate and practically effective algorithms for a wide range of core problems like linear regression, principal component analysis, and clustering.
He will also discuss the important challenge of understanding dimensionality reduction for nonlinear methods, which comprise a large fraction of today’s most powerful machine learning techniques. He will present recent progress and exciting open directions. Finally, he will explain how research on algorithmic tools can have impact beyond faster computation, improving our ability to explore, interpret, and ultimately learn from large datasets.
Bio: Christopher Musco is a PhD candidate in computer science at MIT, advised by Jonathan Kelner. His research focuses on scalable algorithms for core problems in data science and machine learning, with a particular interest in large-scale linear algebra and graph processing. His work is interdisciplinary, bringing together tools from theoretical computer science, scientific computing, and optimization to develop algorithms that push both the theoretical and practical state-of-the-art.
A reception for attendees will be held at 3:30 P.M. in CS 150. (The back of the presentation room.)