קולוקוויום וסמינרים

כדי להצטרף לרשימת תפוצה של קולוקוויום מדעי המחשב, אנא בקר בדף מנויים של הרשימה.

קולוקוויום וסמינרים בקרוב

  • The Cryptographic Lens

    דובר:
    Shafi Goldwasser - Colloquium Lecture - POSTPONED!
    תאריך:
    יום שלישי, 21.10.2014, 14:30
    מקום:
    חדר 337-8 טאוב.
    קישור:
    http://www.cs.technion.ac.il/~colloq/20141021_14_30_Goldwasser.html
  • Computational Approaches in Metabolic Flux Analysis (MFA)

    דובר:
    נעמה טפר, הרצאה סמינריונית לדוקטורט
    תאריך:
    יום רביעי, 22.10.2014, 12:30
    מקום:
    טאוב 601
    מנחה:
    Prof. T. Shlomi

    Cellular metabolism represents fundamental biochemical activities that enable cells to break down food nutrients, generate energy, and produce molecular building blocks required for cell replication. Metabolic processes in living cells involve thousands of enzymes, whose joint activity can be represented via metabolic networks. In these networks, nodes represent small molecules called metabolites, and edges represent biochemical reactions that transform substrate metabolites to products. A major challenge in Systems Biology and Bioinformatics is to develop methods for inferring the rate through reactions in a metabolic network (also referred to as metabolic flux) - essentially, assigning values to edges in the network. A common approach that addresses this challenge is isotope tracing. It works by feeding cells with nutrients (i.e. metabolites) that are labeled with (heavy) stable isotopes, measuring the incorporation of these isotopes within various metabolites in the network though time, and employing computational methods to interpret these metabolite labeling patterns to infer metabolic flux. Intuitively, if metabolites in the network are envisioned as a set of water pools and edges as rivers connecting the pools, isotope tracing is analogous to pouring colored water to one pool and inferring river water flows by tracking the coloring of various water pools. Considering a key observation that metabolic flux through all reactions in the network uniquely determine the labeling pattern of all metabolites, these computational methods typically search for metabolic fluxes that would give rise to metabolite labeling that optimally match experimental measurements. A major limitation of these methods is that computing metabolit labeling given a candidate flux vector is a computationally intensive task. Here, we describe a new computational method called tandemers that enables rapid simulation of metabolite labeling patterns given candidate fluxes through reactions in the network. The method is shown to provide a two-order of magnitude improvement of running time compared to state-of-the-art methods in computing special types of metabolite labeling patterns measured via a technology called tandem-mass spectrometry.The talk will provide an overview of metabolic network analysis and isotope tracing, and will assume no prior biological background.

  • Theory Seminar: From Average Case Complexity to Improper Learning Complexity

    דובר:
    עמית דניאלי (האונ' העברית בירושלים)
    תאריך:
    יום רביעי, 22.10.2014, 12:30
    מקום:
    טאוב 201

    It is presently still unknown how to show hardness of learning problems. There are huge gaps between our upper and lower bounds in the area. The main obstacle is that standard NP-reductions do not yield hardness of learning. All known lower bounds rely on (unproved) cryptographic assumptions.

    We introduce a new technique to this area, using reductions from problems that are hard on average. We put forward a natural generalization of Feige's assumption about the complexity of refuting random $K$-SAT instances. Under this assumption we show:

    1. Learning DNFs is hard. 2. Learning an intersection of super logarithmic number of halfspaces is hard.

    In addition, the same assumption implies the hardness of virtually all learning problems that were previously shown hard under cryptographic assumptions.

    Joint work with Nati Linial and Shai Shelev-Shwartz

  • ceClub: System Approach to Distributed Balanced Graph

    דובר:
    גבי קליאוט (מיקרוסופט)
    תאריך:
    יום חמישי, 23.10.2014, 11:30
    מקום:
    טאוב - יפורסם

    Balanced Graph Partitioning is a hard problem. Doing it at large scale on graphs of millions of nodes and edges is even harder. Doing it in a distributed way makes the problem even more challenging. And finally, dong it with linear or even sub linear time and space complexity may sound like pushing the limits too far. In this talk I will present our practical approach to this hard problem motivated by the systems we build and describe two algorithms that solve it in two different settings. The first operates on a static graph where nodes arrive one by one in a streaming fashion. The second operates on a dynamic, constantly changing graph with random access to nodes and edges. I will also describe how those algorithms were used in two different large distributed systems that we built: Horton - distributed graph database and Orleans – distributed actor based middleware.

    Bio:
    Gabriel Kliot obtained his PhD in Computer Science from the Technion in 2009, working on Distributed Systems and Networking. Since then he has been with Microsoft Research realizing his dream of bringing distributed computing to the masses.

  • Robust Probabilistic Inference

    דובר:
    Yishay Mansour - Colloquium Lecture
    תאריך:
    יום שלישי, 4.11.2014, 14:30
    מקום:
    חדר 337-8 טאוב.
    קישור:
    http://www.cs.technion.ac.il/~colloq/20141104_14_30_Mansour.html
  • When Machines Dominate Humans: The Challenges of Mining and Consuming Machine-generated Web Mail

    דובר:
    Yoelle Maarek - Colloquium Lecture
    תאריך:
    יום שלישי, 11.11.2014, 14:30
    מקום:
    חדר 337-8 טאוב.
    קישור:
    http://www.cs.technion.ac.il/~colloq/20141111_14_30_Maarek.html
  • The Cryptographic Lens

    דובר:
    Shafi Goldwasser - Colloquium Lecture
    תאריך:
    יום שלישי, 9.12.2014, 14:30
    מקום:
    חדר 337-8 טאוב.
    קישור:
    http://www.cs.technion.ac.il/~colloq/20141209_14_30_Goldwasser.html