Colloquium

Digital Health is rapidly becoming a major theme in the Healthcare realm. Coupled with Personalized Medicine, It provides the means to track and monitor condition, and suggest treatments that are far more effective than ever before. It also poses major challenges in collecting and digesting big amounts of data, which in turn should be analyzed in order to detect, predict and suggested the best course of action. Thus, in recent years, there's a growing interest in employing advanced Machine Learning methods using data from smart medical devices and from sensors (e.g.

Computational and mathematical approaches are crucial to the study of evolution and therefore many aspects of biology, medicine, and anthropology.

Much of the current computer vision research is focused on labeling visual objects, and impressive results have been achieved for this task. However, human image understanding is much richer, involving understanding that is both below object recognition (e.g., localizing and labeling the object’s parts), as well as above the object level (e.g., categorizing the interactions between two or more ‘person’ objects). In particular, human understating involves structure: identifying objects and their parts, together with a rich set of semantic relations between them.

Cell imaging has entered the “big data” era. New technologies in microscopy and molecular biology produce high-content and multidimensional data sets encapsulating complex and dynamic patterns that are inaccessible by human visual observation. One example is the emergence of collective behavior in multi-cellular systems.

Graph matching is one of the most well-studied problems in combinatorial optimization, with applications ranging from scheduling and object recognition to numerical analysis and computational chemistry. Nevertheless, until recently very little was known about this problem in real-life dynamic networks, which aim to model the constantly changing physical world. In the first part of the talk we'll discuss our work on dynamic graph matching, and in the second part we'll highlight our work on a few related problems.

Caching is a fundamental technique in computer science. It creates the illusion of a faster memory by maintaining some data items in a memory that is faster or "closer" to the application. Such a technique works because in many real workloads the access pattern is far from uniform. Once the cache is full, in order to admit a new item to the cache one of the cached items should be evicted. During the last decades, the selection of which item to evict attracted plenty of research by industry and academia alike.

Machine learning has recently been revolutionized by the introduction of Deep Neural Networks. However, from a theoretical viewpoint these methods are still poorly understood. Indeed the key challenge in Machine Learning today is to derive rigorous results for optimization and generalization in deep learning. In this talk I will present several tractable approaches to training neural networks.

Massive systems and networks have become ubiquitous. While there is a remarkable amount of work on analyzing and designing algorithms for smaller systems, the vast majority of it simply does not scale: an algorithm that takes seconds to run on a system with thousands of nodes might take weeks on a system with billions. New ideas are required if we hope to have the same success with massive systems as we do with smaller ones.

I will begin by surveying my work within algorithms and uncertainty: the study of algorithms with uncertainty in their inputs. This study arises in different contexts such as average case analysis and understanding properties of complex networks. I will present results on bootstrap percolation in random graphs, the effects of adding a small number of random edges to connected graphs and the computational complexity of solving NP-hard problems over instances that are subjected to random deletions.

In this talk, I will relate to problems of incentive design in multiple domains. One such problem is that of team formation. Looking at this through the lens of game theory, I will suggest a complete information non-cooperative sequential team formation game in which players iteratively recommend teams, which are either accepted or rejected by their prospective members. In this game, there always exists a single subgame perfect equilibrium in which all team proposals are accepted.