Chaining Methods and their Applications to Computer Science (Conference)
Abstract
Statement of Work:Chaining has found several applications to computer science, especially in the areas of randomized algorithms and machine learning. This proposal is for partial support for holding the first conference on Chaining Methods and their Applications in Computer Science (CMACS 2016).Objective:To hold the first Conference on Chaining Methods and their Applications to Computer Science (CMACS 2016).Approach:Chaining is a specific approach, related to maximal inequalities and multi-scale problems, for bounding the maximum of a collection of (dependent) random variables. Chaining is often more efficient than performing a vanilla union bound. Chaining dates back to work of Kolmogorov and since then has been further developed by many others. Today, within computer science it has found a diverse array of applications to problems related to coding theory, random walks ongraphs, dimensionality reduction, compressed sensing, manifold learning, dictionary learning, large-scale linearalgebra, streaming algorithms, and more.The first CMACS conference will be held at Harvard University, Cambridge, MA, June 22~23, 2016, immediately after STOC 2016. STOC is a premier conference in theoretical computer science. The conference will consist of two days ofpresentations. They plan to have 12 speakers, 6 on each day giving one-hour talks. Topics to be covered include (a) Dictionary learning, (b) Dimensionality reduction with applications, (c) Compressed sensing, and (d) Random walks on graphs. The conference website is http://toc.seas.harvard.edu/cmacs .Overall Merit and ONR Mission/Relevance:This conference is related to the ONR~s Information Dominance focus area, and also to Autonomy focus area. Chaining methods have applications in many Naval problems including (a) mathematical optimization, such as subspace embedding to speed up interior methods and linear programming; and (b) Video surveillance for breaking images into low-rank components (background) and sparse components (moving objects). This conference will bring leading mathematicians and computer scientists with expertise in chaining methods to advance the theory behind chaining and examine application areas in computer science.
Document Details
- Document Type
- DoD Grant Award
- Publication Date
- Aug 12, 2016
- Source ID
- N000141612646
Entities
People
- Jelani Nelson
Organizations
- Office of Naval Research
- President and Fellows of Harvard College
- United States Navy