Colloquium

Training accurate visual classifiers from large data sets critically depend on learning the right representation for the problem. I will discuss a representation learning framework based on an iterative interaction of two components: a feature generator suggesting candidate features, and a feature selector choosing among them.

Information diffusion is the process in which nuggets of information spread in a network (typically a social network). This is a complex process that depends on the network topology, the social structures and the information itself (content/language).

Classic error correcting codes are designed to encode messages sent over a noisy channel from one party to another. They are optimized to correct a large number of errors, while still having efficient encoding and decoding algorithms. However, most modern communication is interactive, where two or more parties are actively sending messages based on the information they received. Classic error correcting codes fail to achieve optimal parameters for interactive communication, and in some cases fail to achieve any error correction at all.

The notion of Nash equilibrium (NE) is fundamental to game theory. While a mixed Nash equilibrium is guaranteed to exist in any game, there is no known polynomial-time algorithm for finding one. The tractability of the problem has received much attention in the past decade, in large part due to its theoretical and philosophical significance.

Data-Centers have been emerged in the past few years. Organizations such Google, Microsoft, and Amazon offer different services which are served by their data-centers. In order to provide sufficient and versatile service, commercial data-centers are always researching and developing for better scaling, performance, security, virtualization, management, and topologies.

An Oblivious RAM (ORAM), introduced by Goldreich and Ostrovsky (JACM 1996), is a (probabilistic) RAM that hides its access pattern, i.e. for every input the observed locations accessed are similarly distributed. Great progress has been made in recent years in minimizing the overhead of ORAM constructions, with
the goal of obtaining the smallest overhead possible.

The deployment of Formal Methods for Software Engineering, paved the way for a remarkably powerful (Mathematical) analysis at, each stage of the SW Development Process. Indeed, the Mathematics provides us with an adequate level of rigour, not only to precisely Specify & Design SW Systems, but also to meticulously reason about their Properties. This provides a very high level of confidence in the Correctness, Safety and Reliability of the final SW Product.

Statistics in ranked lists plays an important role in analyzing molecular biology measurement data. For example—differential expression analysis yields ranked lists of genes. These ranked lists are analyzed to infer rules and properties. Statistical enrichment methods are of specific interest in this context allowing for the identification of gene sets or sequence elements that are over-represented in the top of measurement derived ranked lists.

Media streaming is among the most popular services provided over the Internet. The lack of a central authority that controls the users, motivates the analysis of Media on Demand (MoD) services using game theoretic concepts. In this work we define and study the corresponding resource-allocation game, where users correspond to self-interested players who choose a MoD server with the objective of minimizing their individual cost. A server provides both storage and broadcasting needs.