קולוקוויום וסמינרים

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


Computer Science events calendar in HTTP ICS format for of Google calendars, and for Outlook.

Academic Calendar at Technion site.

קולוקוויום וסמינרים בקרוב

  • Modelling Collaborative Discovery

    דובר:
    עדן סייג, הרצאה סמינריונית למגיסטר
    תאריך:
    יום שני, 20.11.2017, 15:30
    מקום:
    טאוב 601
    מנחה:
    Prof. E. Ben-Sasson

    The goal of this work is to develop collaborative mechanisms which help people gain understanding of complex phenomena. We start by presenting an online, collaborative system for the study of child development. Moving from practice to theory, we proceed by introducing an abstract mathematical model for user retention, facilitating the design of efficient crowd collaboration systems. The first part of the work is dedicated to the Baby CROINC (CROwd INtelligence Curation) system, which is an online early-childhood development tracker designed to be both personalized and objective. To meet these goals, we rely on Crowd Curated Intelligence (CCI), a process in which experts curate personalized inputs to connect with the crowd's aggregate data, providing parents with objective and personalized feedback on their children's development. We describe Baby CROINC's design, with a focus on CCI, and assess the extent to which it meets its design goals of objectivity and personalization. In the second part of the work, we present the Collaborative Discovery model of guru-follower dynamics, which explains why "smarter" gurus tend to retain a larger following in the face of competition and limited follower attention. We define a natural class of retentive scoring rules to model the way followers evaluate gurus they interact with, and show that these rules are tightly connected to the classical notion of truth-eliciting proper scoring rules studied in Decision Theory. We then move our attention from the dynamics of interaction between gurus and followers to the study of the intrinsic properties of distributions that deem them appropriate for instruction by a guru. Finally, we take a modest first step towards relating retention models to other established computational complexity measures, namely, dual distance and query complexity, when the phenomena in question can be modeled by a uniform distribution over a linear space.

  • ערב מיקרוסופט במדעי המחשב

    Microsoft Evening at CS

    תאריך:
    יום שני, 20.11.2017, 17:30
    מקום:
    לובי בניין טאוב למדעי המחשב

    חברת מיקרוסופט תקיים אירוע של הרצאות, שתייה וכיבודים ביום שני בערב, 20 בנובמבר 2017, בלובי בניין טאוב.

    באירוע מפגשים עם בכירים, חוקרים ומהנדסים אשר יציעו לכם אפשרויות תעסוקה ואף ינחו אתכם כיצד להשיגן, וכן הרצאה מטעם מנכ"ל מרכז המחקר והפיתוח, מר יורם יעקובי, בנושא המקצועות המבוקשים העתידיים וההיערכות בהתאם.

    פרטים נוספים בכרזה המצורפת.

    נא להירשם מראש.
     

  • Pixel Club: Co-occurrence Filter

    דובר:
    שי אבידן (אונ' תל-אביב)
    תאריך:
    יום שלישי, 21.11.2017, 11:30
    מקום:
    חדר 337, בניין טאוב למדעי המחשב

    Co-occurrence Filter (CoF) is a boundary preserving filter. It is based on the Bilateral Filter (BF) but instead of using a Gaussian on the range values to preserve edges it relies on a co-occurrence matrix. Pixel values that co-occur frequently in the image (i.e., inside textured regions) will have a high weight in the co-occurrence matrix. This, in turn, means that such pixel pairs will be averaged and hence smoothed, regardless of their intensity differences. On the other hand, pixel values that rarely co-occur (i.e., across texture boundaries) will have a low weight in the co-occurrence matrix. As a result, they will not be averaged and the boundary between them will be preserved. The CoF therefore extends the BF to deal with boundaries, not just edges. It learns co-occurrences directly from the image. We can achieve various filtering results by directing it to learn the co-occurrence matrix from a part of the image, or a different image. We give the definition of the filter, discuss how to use it with color images and show several use cases.

  • CSpecial Talk: Learning to Understand Source Code with Machine Learning

    דובר:
    מינטוס אלאמאניס (מיקרוסופט מחקר)
    תאריך:
    יום שלישי, 21.11.2017, 12:30
    מקום:
    טאוב 3

    Deep Neural Networks are succeeding at a range of natural language tasks such as machine translation and text summarization. Recently, the interdisciplinary field of "big code" promises a new set of learnable statistical static analyses. While machine learning tasks on source code have been considered recently, most work in this area does not attempt to capitalize on the unique opportunities offered by its known syntax and structure. In this talk, I discuss how graph neural networks that use code's syntactic and semantic structured information can detect variable misuses in code without any external information (e.g. unit tests).

    Bio: Miltos Allamanis (https://miltos.allamanis.com) is a researcher at Microsoft Research, Cambridge. He is interested in applications of machine learning and natural language processing to software engineering and programming languages to create smart software engineering tools for developers. Previously, he was a PhD student at the University of Edinburgh advised by Dr. Charles Sutton.

  • Information Theory of Deep Learning

    דובר:
    Naftali Tishby - COLLOQUIUM LECTURE
    תאריך:
    יום שלישי, 21.11.2017, 14:30
    מקום:
    חדר 337 טאוב.
    השתייכות:
    Hebrew University
    מארח:
    Yuval Filmus

    I will present a novel comprehensive theory of large scale learning with Deep Neural Networks, based on the correspondence between Deep Learning and the Information Bottlneck framework. The theory is based on the following components: (1) rethinking Learning theory. I will prove a new generalization bound, the input-compression bound, which shows that compression of the input variable is far more important for generalization than the dimension of the hypothesis class, an ill defined notion for deep learning. (2) I will than prove that for large scale Deep Neural Networks the mutual information on the input and the output variables, for the last hidden layer, provide a complete characterization of the sample complexity and accuracy of the network. This put the information Bottlneck bound as the optimal trade-off between sample complexity and accuracy with ANY learning algorithm. (3) I will then show how stochastic gradient descent, as used in Deep Learning, actually achieves this optimal bound. In that sense, Deep Learning is a method for solving the Information Bottlneck problem for large scale supervised learning problems. The theory gives concrete predictions for the structure of the layers of Deep Neural Networks, and design principles for such Networks, which turns out to depend solely on the joint distribution of the input and output and the sample size.

    Based partly on joint works with Ravid Shwartz-Ziv and Noga Zaslavsky.

    Short Bio: ==========
    Dr. Naftali Tishby is a professor of Computer Science, and the incumbent of the Ruth and Stan Flinkman Chair for Brain Research at the Edmond and Lily Safra Center for Brain Science (ELSC) at the Hebrew University of Jerusalem. He is one of the leaders of machine learning research and computational neuroscience in Israel and his numerous ex-students serve at key academic and industrial research positions all over the world. Prof. Tishby was the founding chair of the new computer-engineering program, and a director of the Leibnitz research center in computer science, at the Hebrew university. Tishby received his PhD in theoretical physics from the Hebrew university in 1985 and was a research staff member at MIT and Bell Labs from 1985 and 1991. Prof. Tishby was also a visiting professor at Princeton NECI, University of Pennsylvania, UCSB, and IBM research. His current research is at the interface between computer science, statistical physics, and computational neuroscience. He pioneered various applications of statistical physics and information theory in computational learning theory. More recently, he has been working on the foundations of biological information processing and the connections between dynamics and information. He has introduced with his colleagues new theoretical frameworks for optimal adaptation and efficient information representation in biology, such as the Information Bottleneck method and the Minimum Information principle for neural coding.

  • Bayesian Viewpoint-Dependent Robust Classification under Uncertainty

    דובר:
    יורי פלדמן, הרצאה סמינריונית למגיסטר
    תאריך:
    יום שלישי, 21.11.2017, 15:30
    מקום:
    טאוב 601
    מנחה:
    Asst. Prof. Vadim Indelman

    Object classification and more generally - semantic perception, are an important aspect in situational awareness in autonomous systems. Recent advances in visual information processing have enabled the use of rich semantic information in critical systems, spurring demand for robust, uncertainty-aware semantic perception. The integration of semantic information with noisy spatial (pose, world geometry) information results in mixed - continuous and discrete state belief, often leading to mixture models that are intractable in the general case. As a result, current methods generally ignore state uncertainty when dealing with semantic information, leading to errors where this uncertainty is manifested. Commonly, the problem is simplified even further by assuming spatial independence among measurements and only dealing with most-probable-class measurements, often discarding richer semantic information and making these methods more prone to noise. This seminar presents an approach for incorporating semantic (object class) information in Bayesian state estimation for robust visual classification of a scene object by a mobile robot operating in a previously unknown, partially observable environment, overcoming limitations of current methods. We make use of rich semantic measurements provided by a Bayesian Neural Network classifier with a measure of uncertainty. Fusion of classifier outputs takes into account viewpoint dependency and spatial correlation among observations, as well as pose uncertainty when these observations are taken. Our experiments confirm an improvement in robustness over state-of-the-art.

  • When the Network of a Smart City Is Not So Smart

    דובר:
    עלי טבאג'א, הרצאה סמינריונית למגיסטר
    תאריך:
    יום רביעי, 22.11.2017, 14:00
    מקום:
    טאוב 601
    מנחה:
    Prof. R. Cohen

    RPL is an IEFT standard for building an outdoor wireless mesh, whose main application is smart cities and sensor networks. Outdoor wireless mesh networks are known to be vulnerable to numerous attacks that may be launched on any layer. Such attacks are usually divided into those conducted on the Phy/MAC layer and those conducted on the Network layer. It is easy to conduct a Phy/MAC layer attack, and much more difficult to conduct a Network layer attack. However, the impact of a Phy/MAC layer attack is usually very limited, whereas a Network layer attack may sabotage the whole network. In our work, we present a novel attack that is as easy to conduct as a Phy/MAC layer attack yet is as effective as a Network layer attack. This attack takes advantage of the RPL failure recovery mechanism and the RPL nodes' limited awareness of topological changes.

  • CGGC Seminar: Computational Design for the Next Manufacturing Revolution

    דובר:
    אדריאנה שולץ (MIT)
    תאריך:
    יום שלישי, 28.11.2017, 12:30
    מקום:
    טאוב 401

    Over the next few decades, we are going to transition to a new economy where highly complex, customizable products are manufactured on demand by flexible robotic systems. This change is already underway in a number of fields. 3D printers are revolutionizing production of metal parts in aerospace, automotive, and medical industries. Whole garment knitting machines allow automated production of complex apparel and shoes. Manufacturing electronics on flexible substrates opens the door to a whole new range of products for consumer electronics and medical diagnostics. Collaborative robots, such as Baxter from Rethink Robotics, allow flexible and automated assembly of complex objects. Overall, these new machines enable batch-one manufacturing of products that have unprecedented complexity.

    In my talk, I argue that the field of computational design is essential for the next revolution in manufacturing. This new field has to embrace the following key concepts. First, new design methods have to become more intelligent by utilizing large data repositories and associated machine learning methods. Second, design tools need to transition from declarative to functional, automatically translating functional specifications of an object to manufacturing instructions. Third, workflows need to support concurrent design of shape, materials, control, and software in order to simplify the process and fully utilize the design space. I will showcase how these three concepts are applied by developing new systems for designing robots, drones, and furniture. I will conclude my talk by discussing open problems and challenges for this new emerging research field.

  • A Query Engine for Probabilistic Preferences

    דובר:
    עוזי כהן, הרצאה סמינריונית למגיסטר
    תאריך:
    יום רביעי, 29.11.2017, 11:30
    מקום:
    טאוב 3
    מנחה:
    Prof. B. Kimelfeld

    Models of uncertain preferences, such as Mallows, have been extensively studied due to their plethora of application domains. In a recent work, a conceptual and theoretical framework has been proposed for supporting uncertain preferences as first-class citizens in a relational database. The resulting database is probabilistic, and, consequently, query evaluation entails inference of marginal probabilities of query answers. In this paper, we embark on the challenge of a practical realization of this framework. We first describe an implementation of a query engine that supports querying probabilistic preferences alongside relational data. Our system accommodates preference distributions in the general form of the Repeated Insertion Model (RIM), which generalizes Mallows and other models. We then devise a novel inference algorithm for conjunctive queries over RIM, and show that it significantly outperforms the state of the art in terms of both asymptotic and empirical execution cost. We also develop performance optimizations that are based on sharing computation among different inference tasks in the workload. Finally, we conduct an extensive experimental evaluation and demonstrate that clear performance benefits can be realized by a query engine with built-in probabilistic inference, as compared to a stand-alone implementation with a black-box inference solver.

  • Faster and Simpler Distributed CONGEST-Algorithms for Testing and Correcting Graph Properties

    דובר:
    Guy Even - COLLOQUIUM LECTURE
    תאריך:
    יום שלישי, 19.12.2017, 14:30
    מקום:
    חדר 337 טאוב.
    השתייכות:
    Dept. of Electrical Engineering-Systems, Tel-Aviv University
    מארח:
    Yuval Filmus

    We consider the following problem introduced by [Censor-Hillel et al., DISC 2016]. Design a distributed algorithm (called an $\epsilon$-tester) that tests whether the network over which the algorithm is running satisfies a given property (e.g., acyclic, bipartite) or is $\epsilon$-far from satisfying the property. If the network satisfies the property, then all processors must accept. If the network is $\epsilon$-far from satisfying the property, then (with probability at least $2/3$) at least one processor must reject. Being $\epsilon$-far from a property means that at least $\epsilon\cdot |E|$ edges need to be deleted or inserted to satisfy the property. Suppose we have an $\epsilon$-tester that runs in $O(Diameter)$ rounds. We show how to transform this tester to an $\epsilon$-tester that runs in $O((\log |V|)/\epsilon))$ rounds. Since cycle-freeness and bipartiteness are easily tested in $O(Diameter)$ rounds, we obtain $\epsilon$-testers for these properties with a logarithmic number of rounds. Moreover, for cycle-freeness, we obtain a \emph{corrector} of the graph that locally corrects the graph so that the corrected graph is acyclic. The corrector deletes at most $\epsilon \cdot |E| + distance(G,P)$ edges and requires $O((\log |V|)/\epsilon))$ rounds. Joint work with Reut Levy and Moti Medina. Short Bio: ========== ============================================= Refreshments will be served from 14:15 Lecture starts at 14:30

  • On the Expressive Power of ConvNets and RNNs as a Function of their Architecture

    דובר:
    Amnon Shashua - COLLOQUIUM LECTURE
    תאריך:
    יום שלישי, 2.1.2018, 14:30
    מקום:
    חדר 337 טאוב.
    השתייכות:
    Hebrew University; CEO & CTO, Mobileye; Senior Vice President, Intel Corporation
    מארח:
    Yuval Filmus
  • Bridging the Gap between End-users and Knowledge Sources: Discovery, Selection and Utilization

    דובר:
    Yael Amsterdamer - COLLOQUIUM LECTURE
    תאריך:
    יום שלישי, 9.1.2018, 14:30
    מקום:
    חדר 337 טאוב.
    השתייכות:
    Department of Computer Science, Bar Ilan University
    מארח:
    Yuval Filmus

    TBA

  • Personalization is a Two-Way Street

    דובר:
    Ronny Lempel - COLLOQUIUM LECTURE
    תאריך:
    יום שלישי, 16.1.2018, 14:30
    מקום:
    חדר 337 טאוב.
    השתייכות:
    Outbrain
    מארח:
    Yuval Filmus

    Recommender systems are first and foremost about matching users with items the systems believe will delight them. The "main street" of personalization is thus about modeling users and items, and matching per user the items predicted to best satisfy the user. This holds for both collaborative filtering and content-based methods. In content discovery engines, difficulties arise from the fact that the content users natively consume on publisher sites does not necessarily match the sponsored content that drives the monetization and sustains those engines. The first part of this talk addresses this gap by sharing lessons learned and by discussing how the gap may be bridged at scale with proper techniques. The second part of the talk focuses on personalization of audiences on behalf of content marketing campaigns. From the marketers' side, optimizing audiences was traditionally done by refining targeting criteria, basically limiting the set of users to be exposed to their campaigns. Marketers then began sharing conversion data with systems, and the systems began optimizing campaign conversions by serving the campaign to users likely to transact with the marketer. Today, a hybrid approach of lookalike modeling combines marketers' targeting criteria with recommendation systems' models to personalize audiences for campaigns, with marketer ROI as the target. This talk was given as a keynote in ACM RecSys 2017. Short bio: ========== Ronny Lempel joined Outbrain in May 2014 as VP of Outbrain's Recommendations Group, where he oversees the computation, delivery and auction mechanisms of the company's recommendations. Prior to joining Outbrain, Ronny spent 6.5 years as a Senior Director at Yahoo Labs. Ronny joined Yahoo in October 2007 to open and establish its Research Lab in Haifa, Israel. During his tenure at Yahoo, Ronny led R&D activities in diverse areas, including Web Search, Web Page Optimization, Recommender Systems and Ad Targeting. In January 2013 Ronny was appointed Yahoo Labs' Chief Data Scientist in addition to his managerial duties. Prior to joining Yahoo, Ronny spent 4.5 years at IBM Research, where his duties included research and development in the area of enterprise search systems. During his tenure at IBM, Ronny managed the Information Retrieval Group at IBM's Haifa Research Lab for two years. Ronny received his PhD, which focused on search engine technology, from the Faculty of Computer Science at Technion, Israel Institute of Technology in early 2003. Ronny has authored over 40 research papers in leading conferences and journals, and holds 18 granted US patents. He regularly serves on program and organization committees of Web-focused conferences, and has taught advanced courses on Search Engine Technologies and Big Data Technologies at Technion. ========================================= Refreshments will be served from 14:15 Lecture starts at 14:30