The Taub Faculty of Computer Science Events and Talks

...
[ Full version ]

Online video content has grown enormously, such that it is a huge proportion of newly-created content. There are an estimated four billion different video links available to consumers for watching online. YouTube users alone are currently uploading about half a million videos a day. The leading mobile platforms iOS, Android and Windows are provided with ever larger phone screens for convenient video viewing, and even more so are the various tablets such as the iPad. Furtherm...
[ Full version ]

In this talk I give a short introduction to the process of synthesis, the automatic production of designs from their specifications. We are interested in reactive systems, systems that continuously interact with other programs, users, or their environment and specifications in linear temporal logic. Classical solutions to synthesis use either two player games or tree automata. We give a short introduction to the technique of using two player games for synthesis and how to ...
[ Full version ]

An important element of social choice theory are impossibility theorems, such as Arrow's theorem and Gibbard-Satterthwaite's theorem, which state that under certain natural constraints, social choice mechanisms are impossible to construct. In recent years, much work has been done in finding robust versions of these theorems, showing that impossibility remains even when the constraints are almost always satisfied. In this work we present a general spectral technique for tack...
[ Full version ]

The theory of distributed computing, which lies at the heart of understanding the power and limitations of distributed systems, underwent tremendous progress over the last few decades. Despite this progress, there seems to be a widening gap between the traditional models on top of which the theory of distributed computing is built and the real-world problems we wish to investigate through these models. In this talk we will discuss the different aspects of this widening gap and pre...
[ Full version ]

...
[ Full version ]

From an Engineering stand point it's difficult to argue against the view that software tools should meet, first and foremost, their specifications. This talk concentrates on reminding us a fundamental technique for achieving this goal, while arguing that some basic programming idioms, which most are well known (but not as well followed), can make this technique easier to handle. This part is an expansion of the terse version I've contributed to the book &q...
[ Full version ]

This is a 4 series of 4 talks about GPGPUS, intended for the practical engineer: 1. Motivation, AMD's architecture 2. OpenCL 3.Case studies, Dos and Don'ts 4.Tools and Profiling for Performance General Purpose GPU programming became a hot topic in the last few years, ranging from academic studies to being used by commercial software products. As an example, three out of the world's top10 supercomputers (June2011 list) contain GPUs in them. This s...
[ Full version ]

...
[ Full version ]

...
[ Full version ]

...
[ Full version ]

...
[ Full version ]

A compression algorithm for a computation problem is a polynomial-time algorithm that compresses instances of the given problem into equivalent instances. The performance of the compression is naturally measured with respect to its worst-case output size. While NP-hard problems cannot have compression algorithms with non-trivial performance guarantees in terms of the original input size (assuming NP is not in P), some NP-hard problems have surprisingly good compressions when...
[ Full version ]

Unlike wired network protocols, Wi-Fi does not separate the data channel from the control channel, since only a single sender and receiver can communicate at a given time slot. Flashback is a system that allows multiple transmitters to send 'flashes' of high power OFDM sub-carriers without affecting the normal data transmissions on the Wi-Fi channel. By taking advantage of SNR margins of Wi-Fi channel codes, the flashes do not impose any overhead on the regular data transmission. ...
[ Full version ]

Mobile devices face a growing demand to support computationally intensive applications like 3D graphics and computer vision. However, these devices are inherently limited by processor power density and device battery life. Dynamic remote execution addresses this problem, by enabling mobile devices to opportunistically offload computations to a remote server. We envision remote execution as a new type of cloud-based heterogeneous computing resource, or a "Cloudon-Chip", which would...
[ Full version ]

...
[ Full version ]

We all write code, as a matter of fact, for most of us is part of what we do on a daily basis be it our jobs, personal projects or studies. However, when all those lines of code that we throw here and there become part of an application and that application is developed by a significant number of developers, we should take proper care of our code. Many names have been given to proper management of code: Configuration Manager, Application Lifecycle Manager, DevOps, etc. ...
[ Full version ]

The decade of genomic revolution following the human genome's sequencing has produced significant medical advances, and yet again, revealed how complicated human biology is, and how much more remains to be understood. Biology is an extraordinary complicated puzzle; we may know some of its pieces but have no clue how they are assembled to orchestrate the symphony of life, which renders the comprehension and analysis of living systems a major challenge. Recent efforts to creat...
[ Full version ]

In seven years of rapid growth Facebook has become today's most popular social network, redefining digital identity and the basic set of communication channels for people worldwide. Many unique engineering challenges were handled along the way: building products extremely quickly while keeping the high quality of the service and the low cost of the hardware that supports it, allowing hori-zontal scaling of highly interconnected datar ,anking news items for divergent us...
[ Full version ]

...
[ Full version ]

Many parameter estimation problems involve a parameter dependent PDEs with multiple right hand sides. The computational cost and memory requirements of such problems increase linearly with the number of right hand sides. For many applications this is the main bottleneck of the computation. In this talk we show that problems with multiple right hand sides can be reformulated as stochastic optimization problems that are much cheaper to solve. We discuss the solution methodolog...
[ Full version ]

The task of automatically determining the correct sense of a polysemous word has remained a challenge to this day. It is crucial in many Natural Language Processing (NLP) applications such as speech recognition, information retrieval, machine translation and computational advertising. In our research, we introduce Concept-Based Disambiguation (CBD), a novel framework that utilizes recent semantic analysis techniques to represent both the context of the word and its senses in a hig...
[ Full version ]

Motivated by the classical problem of privacy amplification, Dodis and Wichs (STOC '09) introduced the notion of a non-malleable extractor, significantly strengthening the notion of a strong extractor. A non-malleable extractor is a function $\nmExt : \{0,1\}^n \times \{0,1\}^d \rightarrow \{0,1\}^m$ that takes two inputs: a weak source $W$ and a uniform (independent) seed $S$, and outputs a string $\nmExt(W,S)$ that is nearly uniform given the seed $S$ as well as the value ...
[ Full version ]

Three works will be described. In the first we present reactive secret sharing, that changes the secret according to unbounded sequence of common inputs, where no communication among the (dynamic set of) participants is allowed, we present a fully secure solution for simple functions but somewhat non secure solution for any function.. In the second work dynamic on-going multiparty computation, in which we consider the case of dynamic group of participants that should not know the ...
[ Full version ]

The decade of genomic revolution following the human genome's sequencing has produced significant medical advances, and yet again, revealed how complicated human biology is, and how much more remains to be understood. Biology is an extraordinary complicated puzzle; we may know some of its pieces but have no clue how they are assembled to orchestrate the symphony of life, which renders the comprehension and analysis of living systems a major challenge. Recent efforts to creat...
[ Full version ]

Restoration tasks, such as image denoising, are ill posed problems, often solved with image priors. As image priors are only approximate, this yields suboptimal restoration results. Given the numerous works on image priors and image denoising, it is thus important to understand the inherent limits posed by natural image statistics, and what potential gains we may expect from additional years of research efforts. Recent studies avoided image priors with a non-parametric app...
[ Full version ]

...
[ Full version ]

Humans have an uncanny ability to learn new concepts with very few examples. Cognitive theories have suggested that this is done by utilizing prior experience of related tasks. We propose to emulate this process in machines, by transforming new problems into old ones. These transformations are called metaphors. Obviously, the learner is not given a metaphor, but must acquire one through a learning process. We show that learning metaphors yield better results than existing transfer...
[ Full version ]

In this talk, we consider the problem of self-healing in reconfigurable networks (e.g. peer-to-peer and wireless mesh networks) that are under repeated attack by an omniscient adversary and propose fully distributed algorithms that 'heal' certain global and local properties while doing only local changes and using only local information. Our model assumes repeated attack by an omniscient adversary. We assume that, over a sequence of rounds, an adversary either inserts...
[ Full version ]

Lock-freedom is a progress guarantee that ensures overall program progress. Wait-freedom is a stronger progress guarantee that ensures the progress of each thread in the program. The latter property is valuable for systems that need to be responsive, such as real-time systems, operating systems, etc. While many practical lock-free algorithms are known in the literature, constructing wait-free algorithms is considered a complicated task that results in inefficient implementations. ...
[ Full version ]

...
[ Full version ]

Statistics of 'natural images' provides useful priors for solving under-constrained problems in Computer Vision. Such statistics is usually obtained from large collections of natural images. We claim that the substantial internal data redundancy within a single natural image (e.g., recurrence of small image patches), gives rise to powerful internal statistics, obtained directly from the image itself. While internal patch recurrence has been used in various applications, we p...
[ Full version ]

The linked-list data structure is fundamental and ubiquitous. Lock-free versions of the linked-list are well known. However, the existence of a practical wait-free linked-list has been open. A wait-free data-structure is a structure in which each thread is guaranteed to finish each method in a bounded number of steps, regardless of contention and of the scheduling in the system. We describe a practical wait-free linked-list based on the lock-free linked-list algorithm of Harris. W...
[ Full version ]

In this work we consider the problem of the reconstruction of a signal from the magnitude of its Fourier transform, also known as phase retrieval. The problem arises in many areas of astronomy, crystallography, optics, and coherent diffraction imaging (CDI). Our main goal is to develop an efficient reconstruction method based on continuous optimization techniques. Unlike current reconstruction methods, which are based on alternating projections, our approach leads to a much faster...
[ Full version ]

Linear classification is a fundamental problem of machine learning, in which positive and negative examples of a concept are represented in Euclidean space by their feature vectors, and we seek to find a hyperplane separating the two classes of vectors. In this talk we'll describe recent advances in efficient algorithms for linear classification and related machine learning problems. In particular we'll describe the first sublinear-time (information-optimal) algorithms...
[ Full version ]

In this talk, we present the Replicate and Bundle (RnB) scheme for relieving back-end processor and network bottlenecks in read-mostly key-value storage systems wherein each user request spawns a large number of back-end small-item requests. This is common in Web 2.0 and online social network systems. Adding processors is of little help because this increases the number of back-end requests per user request, thereby also increasing the overall processor and network load. Instead, ...
[ Full version ]

...
[ Full version ]

The brain has recognition capabilities that remain unmatched by computer algorithms. We hypothesize that using matched feedforward-feedback connections, recognition centers of the brain reconstruct an internal copy of inputs using knowledge the brain has previously accumulated, in accordance with a class models called "generative models". Subsequently, it minimizes the error between the internal copy and the input from the environment. We study how this strategy may enab...
[ Full version ]

This is a 4 series of 4 talks about GPGPUS, intended for the practical engineer: 1. Motivation, AMD's architecture 2. OpenCL 3.Case studies, Dos and Don'ts 4.Tools and Profiling for Performance General Purpose GPU programming became a hot topic in the last few years, ranging from academic studies to being used by commercial software products. As an example, three out of the world's top10 supercomputers (June2011 list) contain GPUs in them. This s...
[ Full version ]

We analyze the realistic, popular format of an ascending auction with anonymous item prices, when there are two items that are substitutes. This auction format entails increased opportunities for bidders to coordinate bids, as the bidding process is longer, and since bidders see the other bids and can respond to various signaling. This has happened in many real auctions, e.g., in the Netherlands 3G Telecom Auction and in the FCC auctions in the US. While on the face of it...
[ Full version ]

Coherency Sensitive Hashing (CSH) extends Locality Sensitivity Hashing (LSH) and PatchMatch to quickly find matching patches between two images. LSH relies on hashing, which maps similar patches to the same bin, in order to find matching patches. PatchMatch, on the other hand, relies on the observation that images are coherent, to propagate good matches to their neighbors, in the image plane. It uses random patch assignment to seed the initial matching. CSH relie...
[ Full version ]

This talk introduces theoretical foundations for system architectures and algorithms for creating truly robust autonomic systems -- systems that are able to recover automatically from unexpected failures. We consider various settings of system transparency. We consider black box and transparent box software packages. The general assumption is that a software package fails when it encounters an unexpected environment state -- a state the package was not programmed to cope with. Cre...
[ Full version ]

...
[ Full version ]

For abstract please see Hebrew page. ...
[ Full version ]

At least in the near term, micro-scale platforms like micro air vehicles and micro sensor nodes are unlikely to have power, volume, or mass budgets to support conventional imaging and post-capture processing for visual tasks like detection and tracking. These budgets are severe enough that even common computations, such as large matrix manipulations and convolutions, are difficult or impossible. To help overcome this, we are considering sensor designs that allow some compone...
[ Full version ]

...
[ Full version ]

Many processes in the physical, biological, engineering and social sciences involve information processing at a fundamental level and can be studied through computational models. This talk will describe the impact of such models on areas such as quantum computing, dtatistical physics, economics and game theory, mathematics and computational biology. ...
[ Full version ]

This talk highlights some of the tools made available by search engines to aid users in formulating information needs, digesting complex information spaces and completing tasks online. Much of the innovation and competition in the Web search industry focuses around such tooling, as search engines go well beyond the concept of returning "10 blue links" to users. Examples from multiple search engines will be shown. ...
[ Full version ]

In many practical situations heuristic algorithms reliably give satisfactory solutions to real-life instances of optimization problems, despite evidence from computational complexity theory that the problems are intractable in general.Our long-term goal is to contribute to an understanding of this seeming contradiction, and to put the construction of heuristic algorithms on a firmer footing. As a step in this direction we describe the evolution of a succesful ...
[ Full version ]

Linux is used extensively in systems research as a platform for the implementation of new ideas, exploiting its open-source nature. But Linux is also interesting as an object of study about software engineering. In particular, Linux defies common management theories, as it lacks any coherent plan or management structure, but still grows at an ever-increasing rate, while also gaining market share. We will review some previous studies of Linux development and add new observations re...
[ Full version ]

This is a 4 series of 4 talks about GPGPUS, intended for the practical engineer: 1. Motivation, AMD's architecture 2. OpenCL 3.Case studies, Dos and Don'ts 4.Tools and Profiling for Performance General Purpose GPU programming became a hot topic in the last few years, ranging from academic studies to being used by commercial software products. As an example, three out of the world's top10 supercomputers (June2011 list) contain GPUs in them. This s...
[ Full version ]

...
[ Full version ]

Cloud providers possessing large quantities of spare capacity must either incentivize clients to purchase it or suffer losses. Amazon is the first cloud provider to address this challenge, by allowing clients to bid on spare capacity and by granting resources to bidders while their bids exceed a periodically changing spot price. Amazon publicizes the spot price but does not disclose how it is determined. By analyzing the spot price histories of Amazon's EC2 cloud, we reverse engi...
[ Full version ]

We will discuss some models of interacting economic agents on social networks. The first part of the talk will include a short introduction to the topic, including: - Why assume agents are rational? What does it mean to be rational? - When is it computationally feasible to be rational? - When does interaction eventually lead to agreement, and when can disagreement persist indefinitely? - When does interaction lead to efficient aggregation of information, an...
[ Full version ]

There are many different notions of information-theoretic entropy and its computational analogues. The right notion and a toolbox of lemmas can make for beautifully simple proofs. Drawing on examples from information-theoretic key agreement, leakage-resilient cryptography, and deterministic encryption (no background in these topics is assumed), I will show how various extensions of entropy can lead to improved cryptographic constructions. ...
[ Full version ]

...
[ Full version ]

Locally decodable codes are codes that allow decoding of single message bits by reading a sublinear amount of the codeword. These codes have been the object of intense study, of particular interest is the trade-off between the rate of these codes and the query complexity of their local decoding algorithms. When limiting our attention to codes of constant rate, we know of only two such families of codes: Reed-Muller codes and multiplicity codes. While multiplicity codes achieve a h...
[ Full version ]

In this talk, we will describe a new, octree-based, FEM solver for performing geometry-aware signal processing on meshes. We show that by considering the restriction of functions defined in 3D to the surface, we can define a regular function space on the mesh that supports both multigrid solvers, and parallel and streaming computation. We will discuss applications of the solver to a number of traditional challenges, including texture stitching, parameterization, interactive ...
[ Full version ]

This talk will overview the results of my doctoral research in Property Testing of dense combinatorial structures. In Property Testing, we are concerned with the number of queries one has to make, or information one has to read, from an input combinatorial structure in order to make a rough distinctions between 'good' and 'significantly bad' inputs, where bad inputs are far from being good; specifically, most studies concerns such testing algorithms which read only a small frac...
[ Full version ]

Non uniform filtering is important for many image processing algorithms. However, for large kernel sizes it can become computationally expensive. In this talk we describe two efficient filtering techniques which are independent on kernel size. First we present Cosine Integral Images (CII) which represent a large set of spatial and range filters, based on their frequency decomposition. We apply CII for fast computation of the spatial Gaussian and Gabor kernels, whose complexity i...
[ Full version ]

Direct device assignment enhances the performance of guest virtual machines by allowing them to communicate with I/O devices without host involvement. But even with device assignment, guests are still unable to approach bare-metal performance, because the host intercepts all interrupts, including those interrupts generated by assigned devices to signal to guests the completion of their I/O requests. The host involvement induces multiple unwarranted guest/host context switches, whi...
[ Full version ]

Cooperative and opportunistic communication techniques in wireless networks promise significant performance benefits over traditional methods that do not exploit the broadcast nature of wireless transmissions. However, such techniques generally require coordination among the participating nodes, e.g. to discover available neighbors or negotiate the best course of action after every packet broadcast. The associated coordination overheads negate much of the cooperation benefits for ...
[ Full version ]

In this talk, Prof. Erhardt will discuss the ARTTS project (www.artts.eu) and summary its main results. He will also give an overview of the main activities he is conducted at gestigon, a company he created that deals with Gesture technologies (www.gestigon.com). Prof. Erhardt will also extends the discussion on different research topics that have been applied in this context to (i) the geometrical approach to feature extraction, (ii) the principle of sparse codi...
[ Full version ]

The Linux kernel is one of the largest scale free software projects. More than thousand developers contribute code to each kernel release. Becoming one of them is not an easy challenge. First you need to familiarize yourself with the technical side of kernel development, with its unique peculiarities. Then, you need to understand and carry out the long, and sometimes painful, process of patch submission. However, the reward for this pain is great. This lecture is ...
[ Full version ]

Cloud providers possessing large quantities of spare capacity must either incentivize clients to purchase it or suffer losses. Amazon is the first cloud provider to address this challenge, by allowing clients to bid on spare capacity and by granting resources to bidders while their bids exceed a periodically changing spot price. Amazon publicizes the spot price but does not disclose how it is determined. By analyzing the spot price histories of Amazon's EC2 cloud, we reverse eng...
[ Full version ]

Two of the hardest problems of developing data-processing applications are: (1) integrating data from heterogeneous sources, and (2) handling the inherent discrepancies between data models of the sources and models of programming languages, e.g., the object-relational impedance mismatch. The Semantic Web is a set of technologies (RDF, RDFS, OWL, SPARQL) that facilitate data integration. However, it does not solve the impedance mismatch problem. It merely exchanges object-rel...
[ Full version ]

...
[ Full version ]

Want to build your own Telco? you'll probable need mesh power. Avoid past mistakes by learning about the history of mesh networks, hear how the first wi-fi router was liberated and be sure to checkout what we're doing in project Arig ( http://arig.org.il), here in Israel! Be sure to attend the router emancipation party afterwords: bring your wi-fi router and wash away all it's sins by flashing it with a Foss OS such as OpenWRT. complete redemption guaranteed. ...
[ Full version ]

In binary-utility games, an agent can have only two possible utility values for final states, 1 (win) and 0 (lose). We define an unbiased rational agent as one that seeks to maximize its utility value, but is equally likely to choose between states with the same utility value. In particular, it will prefer winning over losing but will be indifferent as to which winning ( or losing state) is chosen. This induces a probability distribution over the game tree, from which an agen...
[ Full version ]

We propose and study a novel supervised approach to learning semantic relatedness from examples. Using an empirical risk minimization approach our algorithm computes a weighted measure of term co-occurrence with respect to a corpus of text documents, and utilizes the labeled examples to fit the model to the training sample. Our method is corpus independent and can essentially rely on any sufficiently large (unstructured) collection of coherent texts. We present the results of ...
[ Full version ]

0 A.D. is a FOSS game of ancient warfare, belonging to a genre of games called Real-Time Strategy (RTS). It is mostly implemented in C++, along with scripts in JavaScript, and runs on Windows, Linux and Mac OS. The game was last presented at Haifux in December 2009 and has developed tremendously since then, with advances in graphics, A* pathfinding, opponent AI and more. Aviv will demonstrate some of the new features of the game of particular interest to Comp Sci people and explai...
[ Full version ]

Model order reduction is essential for feasible analysis, design, realtime control of distributed systems. Recent uses also include accelerating detailed simulations and the extraction of actionable meaning from large scale data streams. Alas, a mature and mathematically rigorous theory is largely limited to the linear case, and even there, the mere computational complexity its tools entail, restrict its use to relatively moderate dimensions. Heuristics fill in the gap, wit...
[ Full version ]

We discuss a variational methodology, which involves locally modeling of data from noisy samples, combined with global model parameter regularization. We show that this methodology encompasses many previously proposed algorithms, from the celebrated moving least squares methods to the globally optimal over-parametrization methods recently published for smoothing and optic flow estimation. However, the unified look at the range of problems and methods previously c...
[ Full version ]

Since its first emergence in 2001, Wikipedia had grown drastically to become the fifth most viewed website over the Internet in 2012. With over 12,000,000 articles in more than 250 languages this is now the largest source of information ever existed. The Wikimedia foundation has been founded in 2003 as a non-profit organization to support Wikipedia as well as other online and offline free content projects. ...
[ Full version ]

...
[ Full version ]

Motivated by mass-spectrometry protein sequencing, we consider the simple problem of reconstructing a string from its substring compositions. Relating the question to the long-standing turnpike problem, polynomial factorization, and cyclotomic polynomials, we cleanly characterize the lengths of reconstructable strings and the structure of non-reconstructable ones. The talk is elementary and self contained and covers work with Jayadev Acharya, Hirakendu Das, Olgica Milenkovic...
[ Full version ]

Media streaming over TCP is becoming popular because TCP's congestion control provides remarkable stability to the Internet. However, TCP also introduces significant latency and throughput variability in the presence of mobility and frequent handovers. The effect of handovers can be mitigated if the packets received by the old base station during handover are forwarded to the new base station. When the bandwidth of the backbone is scarce, as in a multihop wireless network, packet ...
[ Full version ]

Abstract: Shared segments that are Identical-by-Descent (IBD) from a recent common ancestor of purported unrelateds provide unique source of information for population and medical genetics: IBD addresses recent, rare variation, and is therefore key for interpreting data on the seam between SNP-array and sequencing studies. We have developed a rapid algorithm to efficiently detect IBD segments, enabling such analysis in large cohorts. We formalize this analysis in the context...
[ Full version ]

Focusing on short term trend prediction in a financial context we consider the problem of selective prediction whereby the predictor can abstain from prediction in order to improve its performance. The main characteristic of selective predictors is the trade-off they exhibit between error and coverage rates. In the context of classification selective prediction is termed "classification with a reject option", and there the main idea for implementing rejection is Chow's ambig...
[ Full version ]

We construct an explicit disperser for affine sources over $\F_2^n$ with entropy $k=2^{\log^{0.9} n}=n^{o(1)}$. This is a polynomial time computable function $D:\F_2^n \to \set{0,1}$ such that for every affine space $V$ of $\F_2^n$ that has dimension at least $k$, $D(V)=\set{0,1}$. This improves the best previous construction of Ben-Sasson and Kopparty that achieved $k = \Omega(n^{4/5})$ and is the first pseudorandom object for affine sources with entropy less than $\sqrt{n}$. ...
[ Full version ]

We study the complexity of powering in GF(2^n) by constant depth arithmetic circuits over GF(2) (also known as AC0(parity)). Our study encompasses basic arithmetic operations such as computing cube-roots and cubic-residuosity of elements of GF(2^n). Our main result is that these operations require exponential size circuits. We also derive strong average-case versions of these results. For example, we show that no subexponential-size, constant-depth, arithmetic circ...
[ Full version ]

Video extrapolation is the task of extending a video beyond its original field of view. Extrapolating video in a manner that is consistent with the original video and visually pleasing is difficult. In this work we aim at very wide video extrapolation which increases the complexity of the task. Some video extrapolation methods simplify the task by using a rough color extrapolation. A recent approach focuses on artifact avoidance and run time reduction using foveated video ex...
[ Full version ]

We present highly-efficient synchronization techniques and experimentally show that these techniques outperform most state of the art lock-based and lock-free synchronization mechanisms. One of the techniques ensures, in addition, wait-freedom. We have used these techniques to implement common concurrent data structures, like stacks and queues. Our experiments show that these data structure implementations have much better performance than state of the art shared stack...
[ Full version ]

Symmetric positive definite matrices are widely used. Diffusion kernels, Curvature, Optimization and many more. However, Symmetric positive definite matrices are not a group and form an open unbounded manifold. Thus, not clear how to efficiently interpolate such data. In this lecture we give such a subdivision interpolation scheme based on the geodes in the Rieman metric of the Symmetric positive definite matrices. We prove convergences and smoothness and in addition spect...
[ Full version ]

Call a function f : {0,1}^n -> {0,1} odd-cycle-free if there are no x_1, ..., x_k in {0,1}^n with k an odd integer such that f(x_1) = ... = f(x_k) = 1 and x_1 + ... + x_k = 0. We show that one can distinguish odd-cycle-free functions from those eps-far from being odd-cycle-free by making poly(1/eps) queries to an evaluation oracle. To obtain this result, we use connections between Fourier analysis and spectral graph theory to show that one can reduce testing odd-cycle-freeness of ...
[ Full version ]

We present the first experimental realization of a new semi-classical quantum key distribution (QKD) protocol called classical Alice (with mirror) and its security analysis. The field of quantum information and computation is relatively new and rapidly emerging field of science, and secure key distribution is one of the most prominent practical applications in this area. Unlike other conventional key distribution schemes --- to which we refer as classical --- that rely on (not f...
[ Full version ]

Locally decodable codes are error-correcting codes that admit efficient decoding algorithms: They give a method to encode k bit messages into n bit codewords such that even after a constant fraction of the bits of the codeword get corrupted any bit of the original message can be recovered by only looking at q(k) bits of the corrupted codeword. The tradeoff between the rate of a code (i.e., the ratio k/n) and the locality/efficiency (the function q(k)) of its decoding a...
[ Full version ]

The projective space of order n over the finite field F is the set of all the subspaces of the vector space F^n. A code C in the projective space is defined as a subset of the projective space, i.e., the codewords in C are subspaces of F^n. If all the codewords in C have the same dimension, then C is called a constant dimension code. These codes gained renewed interest due to the work by Koetter and Kschischang (2008), where they presented an application of such codes to error cor...
[ Full version ]

Hierarchical spatial decompositions are a basic modeling tool in a variety of application domains including scientific visualization, finite element analysis and shape modeling and analysis. A popular class of such approaches is based on the regular simplex bisection operator, which bisects simplices (e.g. line segments, triangles, tetrahedra) along the midpoint of a predetermined edge. Regular simplex bisection produces adaptive simplicial meshes of high geometric quality, ...
[ Full version ]

General Purpose Input/Output (GPIO), Serial Peripheral Interface (SPI), and Inter-Integrated Circuit (I2C), are common methods for digital communication between electronic components. The Linux kernel, being a popular choice for embedded solutions, provides a general abstraction layer for each of those communication methods. Modern Linux kernels also include drivers for many hardware modules implementing GPIO, SPI, or I2C. The abstraction layers provide a generic way to communicat...
[ Full version ]

will describe some recent work towards scaling the process of visual recognition to handle thousands of objects classes on a limited computational budget. Formally, the system is designed such that the computational resources grow sub-linearly (poly-logarithmic) with the number of classes. This also implies that features used for recognition should be shared by several classes. Work done jointly with Shai Shalev-Schwartz and Yonatan Wexler...
[ Full version ]

We present a new approach to the development of efficient geometric algorithms for freeform curves and surfaces. Preprocessing the given curves and surfaces and representing them in a hierarchical data structure, we show that a variety of geometric algorithms can be greatly accelerated. We demonstrate the effectiveness of this approach by developing real-time algorithms for collision detection, minimum and Hausdorff distance computation, convex hull computation for freefo...
[ Full version ]

We present the PRIME tool which utilizes static specification mining techniques to extract useful specifications of library APIs from a large number of code fragments that use it, and then uses data mining techniques to aggregate the samples into use-cases and sort them according to popularity and complexity. Programming is becoming more and more about using frameworks and libraries, with most of them designed to support a wide range of usage scenarios. Typically, a programmer ...
[ Full version ]

The determinant and the permanent of a matrix, though deceivingly similar in their definitions, behave very differently with respect to how efficiently one can compute these quantities. The determinant of a matrix over a field can be easily computed via Gaussian elimination while computing the permanent, as shown by Valiant, is at least as hard as counting the number of satisfiable assignments to a Boolean formula. Given this, it is natural to ask "over which algebras, is the de...
[ Full version ]

Network Science is a newly emerging discipline with applications in a variety of domains, such as Communication Networks, Power Grid Networks, Transportation Networks, Social Networks and Biological Networks. Focusing on communication networks, we shall discuss what network science should be and what it should consist of. The talk will also feature some historical anecdotes, tracing back to ancient times. Further details would be a spoiler. ...
[ Full version ]

...
[ Full version ]

With the advent of the Internet it is now possible to collect hundreds of millions of images for computer vision. These images come with varying degrees of label information. "Clean" labels can be manually obtained on a small fraction, "noisy labels" may be extracted automatically from surrounding text, while for most images there are no labels at all. Semi-supervised learning is a principled framework for combining these different label sources. However, it scales polynomi...
[ Full version ]

...
[ Full version ]

The talk will include two parts. In the first part, I will present an overview of my research in geometry processing, including mesh parameterization, surface reconstruction, shape analysis and segmentation. In the second part, I will describe my recent work on geometry-driven image manipulation, including mesh-warping based image retargeting, photo composition optimization and parametric human reshaping. ...
[ Full version ]

The talk will include two parts. In the first part, I will present an overview of my research in geometry processing, including mesh parameterization, surface reconstruction, shape analysis and segmentation. In the second part, I will describe my recent work on geometry-driven image manipulation, including mesh-warping based image retargeting, photo composition optimization and parametric human reshaping. ...
[ Full version ]

List decoding of error-correcting codes is a generalization of unique decoding: while unique decoding relates to the case where the decoder outputs only one word (the correct codeword), list decoding allows to output a list of codewords, as long as the correct codeword is included in the list. Codes for burst error correction have been studied mainly for the purpose of unique decoding. Understanding list decoding of burst errors is desirable for reliable delivery of data over vari...
[ Full version ]

In dealing with the container bloat problem, we identify five memory compaction techniques, which can be used to reduce the footprint of the small objects that make these containers. Using these techniques, we describe two alternative methods for more efficient encoding of the JRE's ubiquitous HashMap data structure, and present a mathematical model in which the footprint of this can be analyzed. First of this is our fused-hashing encoding method, which reduces memory overhead b...
[ Full version ]

Molecular recognition plays a key role in processing information in biological systems which rely on the ability of bio-molecules to specifically recognize each other. However, the crowded biological environment contains a vast variety of molecules that are often structurally similar and may compete with the "right" target. Thus, the recognition process is prone to false binding, which introduces errors and may impair the proper information flow and must be taken into accoun...
[ Full version ]

We initiate a study of zero knowledge proof systems in the presence of side-channel attacks. Specifically, we consider a setting where a cheating verifier is allowed to obtain arbitrary bounded leakage on the entire state (including the witness and the random coins) of the prover during the entire protocol execution. We formalize a meaningful definition of leakage-resilient zero knowledge (LR-ZK) proof system, that intuitively guarantees that the protocol does not yield anything b...
[ Full version ]

Online social networks have become very popular in recent years and their number of users is already measured in many hundreds of millions. For various commercial and sociological purposes, an independent estimate of their sizes is important. In this work, algorithms for estimating the number of users in such networks are considered. The proposed schemes are also applicable for estimating the sizes of networks' sub-populations. The suggested algorithms interact with the...
[ Full version ]

...
[ Full version ]

General Purpose Input/Output (GPIO), Serial Peripheral Interface (SPI), and Inter-Integrated Circuit (I2C), are common methods for digital communication between electronic components. The Linux kernel, being a popular choice for embedded solutions, provides a general abstraction layer for each of those communication methods. Modern Linux kernels also include drivers for many hardware modules implementing GPIO, SPI, or I2C. The abstraction layers provide a generic way to communicat...
[ Full version ]

A Probabilistically Checkable Proof (PCP) is a proof that allows checking the validity of a statement by reading only a constant number of symbols of the proof. The PCP theorem (AS98, ALMSS98) asserts the existence of PCPs of polynomial length for any claim that can be stated as membership in an NP set. Surprisingly, all know constructions of PCPs use Locally Testable Codes (LTCs) as their combinatorial core. An LTC is an error-correcting code that has a randomized test...
[ Full version ]

A Probabilistically Checkable Proof (PCP) is a proof that allows checking the validity of a statement by reading only a constant number of symbols of the proof. The PCP theorem (AS98, ALMSS98) asserts the existence of PCPs of polynomial length for any claim that can be stated as membership in an NP set. Surprisingly, all know constructions of PCPs use Locally Testable Codes (LTCs) as their combinatorial core. An LTC is an error-correcting code that has a randomized test...
[ Full version ]

A Probabilistically Checkable Proof (PCP) is a proof that allows checking the validity of a statement by reading only a constant number of symbols of the proof. The PCP theorem (AS98, ALMSS98) asserts the existence of PCPs of polynomial length for any claim that can be stated as membership in an NP set. Surprisingly, all know constructions of PCPs use Locally Testable Codes (LTCs) as their combinatorial core. An LTC is an error-correcting code that has a randomized test...
[ Full version ]

A Probabilistically Checkable Proof (PCP) is a proof that allows checking the validity of a statement by reading only a constant number of symbols of the proof. The PCP theorem (AS98, ALMSS98) asserts the existence of PCPs of polynomial length for any claim that can be stated as membership in an NP set. Surprisingly, all know constructions of PCPs use Locally Testable Codes (LTCs) as their combinatorial core. An LTC is an error-correcting code that has a randomized test...
[ Full version ]

A Probabilistically Checkable Proof (PCP) is a proof that allows checking the validity of a statement by reading only a constant number of symbols of the proof. The PCP theorem (AS98, ALMSS98) asserts the existence of PCPs of polynomial length for any claim that can be stated as membership in an NP set. Surprisingly, all know constructions of PCPs use Locally Testable Codes (LTCs) as their combinatorial core. An LTC is an error-correcting code that has a randomized test...
[ Full version ]

In this thesis we study three well known combinatorial search problems in various settings: The coin weighing problem, the problem of reconstructing graphs using additive queries and the problem of reconstructing hypergraphs using additive queries. All of the three combinatorial search problems share a common structure. In each problem we have a set of objects called a \emph{universe} or an \emph{instance space}. From the instance space a unique object is selected, we call it the...
[ Full version ]

A k-spanner of a given graph is a subgraph that preserves all distances within factor k. This notion is useful in several contexts, from distributed computing to property testing. By examining spanners through flow-based linear programming relaxations, we design an O(n^{2/3})-approximation algorithm for the directed k-spanner problem that works for all k. This is the first sublinear approximation for arbitrary edge-lengths. We also design a different rounding scheme with a ...
[ Full version ]

...
[ Full version ]

Since its first emergence in 2001, Wikipedia had grown drastically to become the fifth most viewed website over the Internet in 2012. With over 12,000,000 articles in more than 250 languages this is now the largest source of information ever existed. The Wikimedia foundation has been founded in 2003 as a non-profit organization to support Wikipedia as well as other online and offline free content projects. In this talk I will present Wikipedia and other projects of the...
[ Full version ]

We construct a small set of explicit linear transformations mapping R^n to R^t, where t=O(log(\gamma^{-1}) \epsilon^{-2}), such that the L_2 norm of any vector in R^n is distorted by at most 1 \pm \epsilon in at least a fraction of 1-\gamma of the transformations in the set. Albeit the tradeoff between the size of the set and the success probability is sub-optimal compated with probablistic arguments, we nevertheless are able to apply our construction to a number of problems....
[ Full version ]

We present an approach for automatic verification of concurrent programs running under relaxed memory models. Verification under relaxed memory models is a hard problem. Given a finite state program and a safety specification, verifying that the program satisfies the specification under a sufficiently relaxed memory model is undecidable. For somewhat stronger memory models, the problem is decidable but has non-primitive recursive complexity. We use abstract interpretation to provi...
[ Full version ]

In the index coding problem, introduced by Birk and Kol (INFOCOM, 1998), the goal is to transmit n bits to n receivers (one bit to each), where the receivers reside at the nodes of a graph G and have prior access to the bits corresponding to their neighbors in the graph (side information). The objective is to find a code word of minimum length which will allow each receiver to learn their own bit given access to the code word and their side information. When the encoding is linea...
[ Full version ]

What if you could build the next Facebook or Twitter with just a few hundred lines of code ? and scale it to hundreds of millions of users across thousands of servers, right out of the box? Orleans is a cloud programming model and runtime from Microsoft Research, which can make this the new "norm". Orleans is a software framework for building 'client + cloud' applications. It encourages the use of simple concurrency patterns that are easy to understand and implement co...
[ Full version ]

In this work, we explore the use of the diffusion geometry framework for the fusion of geometric and photometric information in local and global shape descriptors. Our construction is based on the def- inition of a di ffusion process on the shape manifold embedded into a high-dimensional space where the embedding coordinates represent the photometric information. Experimental results show that such data fu- sion is useful in coping with di fferent challenges of shape analysis wher...
[ Full version ]

Data analysis of massive data sets is common today for web-search (e.g. Google), social networking (e.g. Facebook), financial applications, supermarkets, bioinformatics and many other fields. A coreset (or, core-set) for a given problem is a "compressed" representation of its input, in the sense that a solution for the problem with the (small) coreset as input would yield an approximate solution to the problem with the original (large) input. Using traditional techn...
[ Full version ]

Structured classification tasks such as sequence labeling and dependency parsing have seen much interest by the Natural Language Processing and the machine learning communities. Several online learning algorithms were adapted for structured tasks such as Perceptron, Passive-Aggressive and the recently introduced Confidence-Weighted learning . These online algorithms are easy to implement, fast to train and yield state-of-the-art performance. However, unlike probabilist...
[ Full version ]

Motivated by the Hadamard product of matrices we define the Hadamard product of noncommutative multivariate polynomials and study its arithmetic circuit and branching program complexity. We also give applications and connections to polynomial identity testing. One of our main results is a tight characterization of polynomial identity testing for noncommutative algebraic branching programs over the field of rationals(we show the problem is complete for logspace coun...
[ Full version ]

A decade of research has been devoted to addressing vulnerabilities in global Internet routing system. The result is a plethora of security proposals, each providing a different set of security guarantees. To inform decisions about which proposal should be deployed in the Internet, we present the first side-by-side quantitative comparison of the major security variants. We evaluate security variants on the basis of their ability to prevent one of the most fundamental forms of atta...
[ Full version ]

I will describe two recent results in unsupervised supervised learning (performing supervised learning tasks without labels). The first result concerns evaluating the accuracy of classifiers and regression models without labels. The second concerns training margin based classifiers such as SVM or logistic regression for high dimensional data without labels. ...
[ Full version ]

...
[ Full version ]

Maximally stable component detection is a very popular method for feature analysis in images, mainly due to its low computation cost and high repeatability. With the recent advance of feature-based methods in geometric shape analysis, there is significant interest in finding analogous approaches in the 3D world. In this paper, we formulate a diffusion-geometric framework for stable component detection in non-rigid 3D shapes, which can be used for geometric feature dete...
[ Full version ]

...
[ Full version ]

Random restrictions are commonly used for proving complexity lower bounds, eg. lower bounds on constant-depth circuits (Ajtai, Furst-Saxe-Sipser, Yao, Hastad) and lower bounds on formula size (Subbotovskaya, Impagliazzo-Nisan, Paterson-Zwick, Hastad). I will show how to use results on random restrictions to bound the running time of certain algorithms for Satisfiability which beat brute-force search. Specifically I will present an algorithm which runs in time 2^{n - \Om...
[ Full version ]

Reconstruction of genome-scale metabolic networks is considered as key step in quantifying the genotype-phenotype relationship. A major computational challenge involved in the reconstruction process is the identification of missing reactions in a metabolic network a process commonly referred to as gap-filling. Here, we present a novel gap-filling approach, MetabolIc Reconstruction via functionAl GEnomics (MIRAGE) that searches for missing reactions required to catalyze metaboli...
[ Full version ]

In the first part of the talk I will give a high-level introduction to Galois, a framework for designing and implementing parallel graph algorithms and related concepts. I will explain what are ordered/unordered graph algorithms and how optimistic parallelism is achieved using transactional boosting. In the second part of the talk I will describe a new shape analysis, which is used for analyzing graph algorithms written with Galois. The shape analysis infers properties...
[ Full version ]

Throughout the past decade there has been extensive research on algorithmic and data mining techniques for solving the problem of influence maximization in social networks: if one can convince a subset of individuals to influence their friends to adopt a new product or technology, which subset should be selected so that the spread of influence in the social network is maximized? Despite the progress in modeling and techniques, the incomplete information aspect of p...
[ Full version ]

...
[ Full version ]

A central problem in algebraic complexity theory and algorithms design is the problem of Polynomial Identity Testing (PIT): given an arithmetic circuit $C$ over a field $\F$, with input variables $x_1, x_2, ... , x_n$, determine whether $C$ computes the identically zero polynomial. Numerous applications and connections to other algorithmic and number theoretic problems further emphasize the significance of PIT. Among the examples are algorithms for finding perfect matchings in gr...
[ Full version ]

A central problem in algebraic complexity theory and algorithms design is the problem of Polynomial Identity Testing (PIT): given an arithmetic circuit $C$ over a field $\F$, with input variables $x_1, x_2, ... , x_n$, determine whether $C$ computes the identically zero polynomial. Numerous applications and connections to other algorithmic and number theoretic problems further emphasize the significance of PIT. Among the examples are algorithms for finding perfect matchings in g...
[ Full version ]

This talk will discuss some recent developments in search, emerging in various shapes and forms. We will highlight some challenges, and point to some search trends that play an increasing role in multiple domains. We will also discuss the power of data and the significant role of cloud technologies in facilitation of new opportunities. Some of the core technologies and global innovations developed in Google's R&D center in Israel will be highlighted. ...
[ Full version ]

...
[ Full version ]

Finding a match between partially available deformable shapes is a challenging problem with numerous applications. The problem is usually approached by computing local descriptors on a pair of shapes and then establishing a point-wise correspondence between the two. We introduce an alternative correspondence-less approach to matching fragments to an entire shape undergoing a non-rigid deformation. We use diffusion geometric descriptors and optimize over the integration domai...
[ Full version ]

Direct device assignment, where a guest virtual machine directlyinteracts with an I/O device without host intervention, is appealing,because it allows an unmodified (non-hypervisor-aware) guest toachieve near-native performance. But device assignment for unmodifiedguests suffers from two serious deficiencies: (1) it requires pinningof all the guest's pages, thereby disallowing memory overcommitment,and (2) it exposes the guest's memory to buggy device drivers. We solve...
[ Full version ]

Query auto completion is known to provide poor predictions of the user's query when her input prefix is very short (e.g., one or two characters). In this work we show that context, such as the user's recent queries, can be used to improve the prediction quality considerably even for such short prefixes. We propose a context-sensitive query auto completion algorithm, NearestCompletion, which outputs the completions of the user's input that are most similar to the context queries. T...
[ Full version ]

Generating random bits is a fundamental problem in cryptography. Coin-tossing protocols, which generate a random bit with uniform distribution, are used as a building box in many cryptographic protocols. Cleve [STOC 1986] has shown that if at least half of the parties can be malicious, then, in any $r$-round coin-tossing protocol, the malicious parties can cause a bias of $\Omega(1/r)$ in the bit that the honest parties output. However, for more than two decades the best known pro...
[ Full version ]

The talk will discuss deduplication, a form of compression in which duplicate copies of files are replaced by links to a single copy. Deduplication is known to reduce the space and bandwidth requirements of Cloud storage services by more than 90%, and is most effective when applied across multiple users. We study the privacy implications of cross-user deduplication. We demonstrate how deduplication can be used as a side channel which reveals information about the conte...
[ Full version ]

...
[ Full version ]

Facility location problems concern assigning requests to servers. The goal is to minimize the total cost which consists of the moving costs and the latency costs. The moving costs depend on the distances between matcheded request/server pairs, while the latency costs depend on the number of requests matched to each server. Facility location problems arise in many natural settings of resource sharing, such as server assignment in cloud computing, network routing and more. Universal...
[ Full version ]

The major spliceosome is a multi-component and highly dynamic complex that carries out the tightly regulated steps of splicing. It is composed of hundreds of proteins, including five small nuclear ribonucleoprotein complexes (snRNPs) that catalyze>99% of all pre-mRNA splicing events. In addition, there are alternative splicing regulators, such as the SR and hnRNP proteins, that either activate or repress a subset of the splicing events. Members of these protein families are ...
[ Full version ]

In this talk I will present a framework for synthesizing efficient synchronization in concurrent programs, a task known to be difficult and error-prone when done manually. The framework is based on abstract interpretation and can infer synchronization for infinite state programs. Given a program, a specification, and an abstraction, we infer synchronization that avoids all (abstract) interleavings that may violate the specification, but permits as many valid interleavings as possi...
[ Full version ]

Since its first emergence in 2001, Wikipedia had grown drastically to become the fifth most viewed website over the Internet in 2012. With over 12,000,000 articles in more than 250 languages this is now the largest source of information ever existed. The Wikimedia foundation has been founded in 2003 as a non-profit organization to support Wikipedia as well as other online and offline free content projects. In this talk I will present Wikipedia and other projects of the...
[ Full version ]

In this work we analyze judgement aggregation problems in which a group of agents independently votes on a set of complex propositions that has some interdependency constraint between them (e.g., transitivity when describing preferences). We generalize the previous results by studying approximate judgement aggregation. We relax the main two constraints assumed in the current literature, Consistency and Independence and consider mechanisms that only approximately satisfy these cons...
[ Full version ]

Brain-computer interfaces (BCI) are devices that can decode physiological signals from the brain and convert them into actions in a manner that reflects the brain’s intention. Their goal is to replace or restore lost function in paralyzed humans by routing movement-relate​d signals from the brain, around damaged parts of the nervous system, to external effectors. My research is aimed at developing a new generation of brain-computer interfaces at the single cell level with hu...
[ Full version ]

...
[ Full version ]

It is well-known that any Boolean function f:{-1,+1}^n \to {-1,+1} can be written uniquely as a polynomial f(x) = \sum_{S subset [n]} f_s \prod_{i in S} x_i. The collection of coefficients (f_S's) this expression are referred to (with good reason) as the Fourier spectrum of f. The Fourier spectrum has played a central role in modern computer science by converting combinatorial and algorithmic questions about f into algebraic or analytic questions about the spectrum. In this tal...
[ Full version ]

Genome-scale metabolic models enable to successfully predict a variety of metabolic phenotype in microorganisms. Still, the integration of metabolic networks with various 'omics' data towards the prediction of metabolic flux remains an open challenge. Here, we show that enzyme kinetic parameters are significantly correlated with measured fluxes in E. coli under various conditions, providing a higher correlation than that achieved by measured gene expression data. Based on the latt...
[ Full version ]

We study graph partitioning problems from a min-max perspective, in which an input graph on $n$ vertices should be partitioned into $k$ parts, and the objective is to minimize the maximum number of edges leaving a single part. The two main versions we consider are where the $k$ parts need to be of equal-size, and where they must separate a set of $k$ given terminals. We consider a common generalization of these two problems, and design for it an $O(\sqrt{\log n}\log^{3/2} k)...
[ Full version ]

As the size of High Performance Computing clusters grows, the increasing probability of interconnect hot spots degrades the latency and effective bandwidth the network provides. This paper presents a solution to this scalability problem for real life constant bisectional-bandwidth fat-tree topologies. It is shown that maximal bandwidth and cut-through latency can be achieved for MPI global collective traffic. To form such congestion-free configuration, MPI programs should utilize ...
[ Full version ]

...
[ Full version ]

A fast pattern matching scheme termed Matching by Tone Mapping (MTM) is introduced which allows matching under non-linear tone mappings. We exploit the recently introduced Slice Transform to implement a fast computational scheme requiring computational time similar to the fast implementation of Normalized Cross Correlation (NCC). In fact, the MTM measure can be viewed as a generalization of the NCC for non-linear mappings and actually reduces to NCC when mappings are restric...
[ Full version ]

A design matrix is a matrix whose attern of zeos/nonzeros satisfies a certain design-like condition. We will first prove that the rank of any design matrix is high. We shall discuss two applications of this rank lower bound: (1) Impossibility results for 2-query locally correctable codes over real/complex numbers, and (2) generalization of results in combinatorial geometry, for example, a robust analog of the Sylvester-Gallai theorem. Joint work with Boaz ...
[ Full version ]

Over the last days, millions of viewers witnessed computing history being made as IBM's Watson question answering system defeated Jeopardy! quiz show champions Brad Rutter and Ken Jennings. Watson is an application of advanced natural language processing, information retrieval, knowledge representation and reasoning, and machine learning technologies to the field of open domain question answering. Watson runs on a cluster of 90 IBM Power 750 servers in 10 racks with a total of 288...
[ Full version ]

We address the open problem of blindly separating single-path position varying image mixtures, without having prior information about the sources. We assume that the mixing system's spatial distortion and attenuation change with position. A staged method for estimating the mixing models is used in turn to recover source signals from such mixtures. Our method is based on a Staged Sprase Component Analysis (SSCA) of the mixtures. Our method consists of three stages: aligning ...
[ Full version ]

In this story, we'll follow the life story of alice - a file-systemized I/O request, and bob - a raw-device I/O request, from their birth, until they reach heaven (the disk or the USB camera or...). In addition, we will cover some system parameters that affect I/O requests, the buffer cache, disk I/O schedulers and tools used to track and count I/O (including - why is process-based I/O accounting so tricky). ...
[ Full version ]

In a recent paper, Warren, Schaefer, Hirani, and Desbrun proposed a simple method of interpolating a function defined on the boundary of a smooth convex domain, using an integral kernel with properties similar to those of barycentric coordinates on simplexes. When applied to vector-valued data, the interpolation can map one convex region into another, with various potential applications in computer graphics, such as curve and image deformation. In this pape...
[ Full version ]

We present an algorithm for implementing a secure oblivious RAM where the access pattern is perfectly hidden in the information theoretic sense, without assuming that the CPU has access to a random oracle. In addition we prove a lower bound on the amount of randomness needed for implementing an information theoretically secure oblivious RAM. Authors: Ivan Damgård, Sigurd Meldgaard, Jesper Buus Nielsen ...
[ Full version ]

...
[ Full version ]

The phenomenon of visual curve completion, where the visual system completes the missing part (e.g., due to occlusion) between two contour fragments, is a major problem in perceptual organization research. Previous computational approaches for the shape of the completed curve typically follow formal descriptions of desired, image-based perceptual properties (e.g, minimum total curvature, roundedness, etc...). Unfortunately, however, it is difficult to determine such desire...
[ Full version ]

...
[ Full version ]

The Classical Secretary Problem was introduced during the 60's of the 20-th century (nobody is sure exactly when). Since its introduction, many variants of the problem have been proposed and researched. In the classical secretary problem, and many of its variant, the input (which is a set of secretaries, or elements) arrives in a random order. We applied to the secretary problem a simple observation which states that the random order of the input can be generated by independently ...
[ Full version ]

Coordinating the proper ordering of events across remote sites is a central task of distributed applications. In asynchronous systems, such coordination depends in an essential way upon message chains, as captured by Lamport's happened-before relation. The relation provides a useful approximation of causality, in the sense that in asynchronous systems two event can only be causally related if they are Lamport related. The talk will consider coordination and causality in s...
[ Full version ]

...
[ Full version ]

...
[ Full version ]

In this story, we'll follow the life story of alice - a file-systemized I/O request, and bob - a raw-device I/O request, from their birth, until they reach heaven (the disk or the USB camera or...). In addition, we will cover some system parameters that affect I/O requests, the buffer cache, disk I/O schedulers and tools used to track and count I/O (including - why is process-based I/O accounting so tricky). ...
[ Full version ]

Pub-sub is a paradigm for asynchronous communication that is commonly used in a great variety of industrial applications, such as news tickers, delivery of financial data, Military applications, and many others. While client-server communication still remains the prevailing implementation paradigm for pub-sub, its limitations for large-scale applications are widely recognized. Major industrial players such as Google and Tibco are recognizing the potential of cooperative ov...
[ Full version ]

The so called Extended Frege system is one of the most natural propositional proof systems. Whereas we believe that there exist tautologies which require exponential Extended Frege proofs, the best known lower bound is linear. To find even a modestly superlinear lower bound is a challenging open problem. I will discuss a possible approach to this question, which is based on counting the number of commutativity axioms in an Extended Frege proof, which in turn can be phrased as an a...
[ Full version ]

For many years, the frequency of microprocessors has grown at an exponential pace (from 20 MHz in 1990 to 2 GHz in 2000). Since 2002 however, despite considerable effort, the frequency has been plateauing. Advances in technology and micro-architecture now translate into more parallelism. The consequences on the software industry are dramatic: most existing software has been designed with a sequential model in mind, and even parallel applications contain residual sequential section...
[ Full version ]

...
[ Full version ]

Variational and PDE-based methods have been extensively used for vision and image-processing tasks such as denoising, segmentation, inpainting, optical flow and more. An underlying assumption is the (piecewise) local correlation between pixels in typical images. Images also exhibit nonlocal correlations in repetitive structures and textures. A coherent mathematical framework for nonlocal regularization will be presented in this talk, which stems from graph theory. The notions of ...
[ Full version ]

We consider the problem of forecasting turning points in time series and develop an autoregressive prediction algorithm, that relies on a novel turning point indicator and support vector regression. The algorithm is analyzed and compared to existing methods in the context of financial forecasting....
[ Full version ]

The problem of finding the minimum s-t cut between a source s and a sink t in a graph is a well-studied problem in computer science. Finding the minimum s-t cut in a planar graph has applications in many fields - from traffic design to computer vision. The geometrical duality between a minimum s-t cut in an undirected planar graph and cycle on the plane that separates t from s was first used by Itai and Shiloach to solve the minimum cut problem, and more efficient algo...
[ Full version ]

IBM's newest technology, the Power Edge of Network (PowerEN) processor, merges network and server attributes to create a new class of wire-speed processor. PowerEN is a hybrid computer that employs: massive multithreading capabilities, integrated I/O and unique special-purpose accelerators for compression, cryptography, pattern matching, XML and Network processing. As a novel architecture, the PowerEN processor offers fertile ground for research. It can facilitate the ...
[ Full version ]

...
[ Full version ]

I will present two presentations about an EU FP7 IP project I work on, UniversAAL - http://universaal.org/. The goal of the project is to develop an open source platform for Ambient Assisted Living (AAL). Ambient Assisted Living is a kind of Smart Home environment for elderly people - for example, a house equipped with different sensors and in which different devices, sensors and home appliances are networked together and managed by software applications. In addition, th...
[ Full version ]

...
[ Full version ]

We suggest a new framework for block ciphers named Advanced Block Cipher, or shortly ABC. ABC has additional non-secret parameters that ensure that each call to the underlying block cipher uses a different pseudo-random permutation. It therefore ensures that attacks that require more than one block encrypted under the same secret permutation cannot apply. In particular, this framework protects against dictionary attacks, and differential and linear attacks, and eliminates weaknes...
[ Full version ]

As research on space-time reconstruction matures, we should ask ourselves what information we use to correctly leverage the temporal or motion component of the data. In my talk I will discuss several possible motion priors and their impact on the reconstruction. Specifically, I will address the observation that most changes in shape are gradual, in both intrinsic and Euclidean sense. I will discuss the impact of this observation on the interpretation of s...
[ Full version ]

When monocyte-derived human dendritic cells (DCs) are infected by Newcastle disease virus, the virus is known to be detected by RIG-I proteins, which induces interferon production. Interferon activates a host of genes, including the gene coding RIG-I. Single cell measurements is DCs show large cell to cell variation of 3-4 orders of magnitude at 6-10 hours after infection. In order analyze early times after infection, when reliable direct single cell data cannot be obtained,...
[ Full version ]

Live migration of virtual machines is an important component of the emerging cloud computing paradigm. While live migration provides extreme versatility of management, it comes at a price of degraded service performance during migration. The bulk of studies devoted to live migration of virtual machines focus on the duration of the copy phase as a primary metric of migration performance. While shorter down times are clearly desirable, the pre-copy phase imposes an overhead on ...
[ Full version ]

In this lecture I'll talk about the encryption methods that were common through history, How to use them, what their weaknesses are and how you can break them. During the lecture I will give examples. The audience is requested to bring writing equipment and paper. No previous background required. ...
[ Full version ]

...
[ Full version ]

We study a publish/subscribe service for mobile ad-hoc networks. Mobile Ad-Hoc Networks (MANETs) are formed by a collection of mobile nodes, each equipped with wireless communication capabilities, without relying on any fixed infrastructure or centralized administration. A publish/subscribe system is responsible for delivering data from a source to interested users. A user expresses interest in receiving certain data by submitting a predicate about corresponding data content. In ...
[ Full version ]

CIFS is the underlying protocol in Windows OS's for disk shares access such as copying files from and to remote stations in the LAN, exploring disk shares and sending jobs to a LAN printer. CIFS is also implemented for Linux/Unix OSs in SAMBA. The CIFS protocol is chatty and suffers from redundancy in messages even in simple and basic disk shares accesses. CIFS's ineffectiveness is intensified in satellite networks in which the round trip delay is large. This work ...
[ Full version ]

...
[ Full version ]

A motherboard, a CPU, and a memory stick. Add a fan and a power supply, and you have a little computer which boots from network and keeps all its data on a hosting computer's disk through NFS. The shopping list for new components goes as low as 600 NIS, which makes this an attractive solution in cases where a dedicated computer is useful, be it for mission-critical applications, cases where the hardware may be damaged or stolen, or because the OS is bound to crash often (kernel ha...
[ Full version ]

A Lattice Animal is a set of edge-connected cells in a given lattice. For example, the Tetris game is played with Lattice Animals with 4 cells in the two-dimensional orthogonal lattice. The enumeration of Lattice Animals is a long-time standing problem, arising in all of recreational mathematics, discrete geometry, and statistical physics. In this talk I will discuss a generalization of Redelmeier's algorithm to the enumeration of polyominoes that lie on any structural (repetitv...
[ Full version ]

...
[ Full version ]

A wireless, mobile, ad hoc network (MANET) is a network in which mobile nodes do not rely on the existence of fixed infrastructure mediation devices, but rather communicate with one another directly. Under certain scenarios, parties in a MANET may wish to remain unidentified, in order to forestall retaliation by an attacker. In this talk, we discuss mechanisms for anonymous routing in MANETs. First, we note threats to anonymity in MANETs, derive a general adversary model and use i...
[ Full version ]

In STOC 1999, Raz presented a (partial) function for which there is a quantum protocol communicating only $O(\log n)$ qubits, but for which any classical (randomized, bounded-error) protocol requires $\poly(n)$ bits of communication. That quantum protocol requires two rounds of communication. Ever since Raz's paper it was open whether the same exponential separation can be achieved with a quantum protocol that uses only one round of communication. In other words, can quantum one-w...
[ Full version ]

This research is concerned with novel Web-related data management scenarios. In particular, we consider database problems which arise from social networks. In the talk, we will overview the main results of this research. In the first part, we introduce the Query Network model, a basic model for social network automation using queries, with its evaluation algorithms and related experiments. We also discuss extensions for the model, and present a few theoretical results. In ...
[ Full version ]

A biological model of visual information representation is adopted. Images are represented accordingly in a multidimensional space that incorporates the well investigated dimensions of intensity, color and spatio-temporal frequency. The model is extended to incorporate additional, less investigated dimensions such as curvature, size and depth (for example - from binocular disparity). Along these and other dimensions, that are yet to be discovered, the human visual system (HVS) enh...
[ Full version ]

The topic of shape reconstruction is a major research branch of geometric processing. The two major problems which are commonly researched are reconstruction of two and three-dimensional objects from point sets, and reconstruction from cross-sections. This research focuses on the latter, which has received a substantial amount of attention in the last three decades, and which has specifically gained considerable momentum in the last few years. We propose two solutions for the ...
[ Full version ]

Standard image classification algorithms classify an image by its content. Sometimes, we don't care about the image content and want to classify images by style. For example, if we have a set of paintings and we want to divide them to groups by their painting style or painter, we don't care if a car or a horse was painted. Another usage for classifying by style is when we have a painting and we are not sure whether it was really painted by a specific painter, and we want to check ...
[ Full version ]

We consider the problem of approximately solving a system of homogeneous linear equations over reals, where each equation contains at most three variables. Since the all-zero assignment always satisfies all the equations exactly, we restrict the assignments to be ``non-trivial". Here is an informal statement of our result: it is NP-hard to distinguish whether there is a non-trivial assignment that satisfies 1-\delta fraction of the equations or every non-trivial assignment ...
[ Full version ]

A year ago I started working for a small multi-national software giant named Microsoft. What I found is a company with rather surprising and odd corporate culture and habbits. This lecture will try to convey What microsoft is like on the inside. Disclaimer: anything said in this lecture is my opinion alone and is not affiliated with microsoft in any way. ...
[ Full version ]

Scientists, engineers and physicians are now confronted with a fire hose of data. Immersive visualization environments provide these users with a novel way of interacting and reasoning with large datasets. They allow them to utilize the entirety of their visual bandwidth, effectively engulfing the user in the data and enabling collaborative interaction. We present a custom-built 5-wall Cave environment, called the Immersive Cabin (IC). It is driven by a GPU cluster for both comput...
[ Full version ]

...
[ Full version ]

A fundamental problem in dynamic graphs is the recovery of structural information in a network whose edges occasionally fail. In a failure event, some subset of edges D are deleted and we want to quickly understand the structure of the surviving network G \ D. In particular, I will discuss the problem of quickly recovering shortest paths and a minimum spanning tree in G \ D. Let G = (V,E) be a directed edge-weighted graph and let P be a shortest path from s to t in G. ...
[ Full version ]

...
[ Full version ]

This work focuses on characterizing scenery images. We semantically divide the objects in natural landscape scenes into background and foreground and show that the shapes of the regions associated with these two types are statistically different. We then focus on the background regions. We study statistical properties such as size and shape, location and relative location, the characteristics of the boundary curves and the correlation of the properties to the region's semantic ide...
[ Full version ]

Invariance principles or limit theorems have recently found several important applications in theoretical computer science. In this talk I'll present some recent results with the broad theme of constructing pseudorandom generators from invariance principles. The first set of results concerns the natural question of constructing pseudorandom generators (PRGs) for low-degree polynomial threshold functions (PTFs). We give a PRG with seed-length log n/eps^{O(d)}$ fooling degr...
[ Full version ]

As multi-core and multiprocessing architectures are becoming common, modern applications require concurrent data structures for their computations. Designing concurrent data structures and ensuring their correctness is a difficult task; significantly more challenging than doing so for their sequential counterparts. Transactional memory (TM), a programming model in which concurrent processes synchronize via in-memory transactions, is one prominent approach for alleviating the diff...
[ Full version ]

...
[ Full version ]

We show that the cell probe complexity of performing nearest neighbor (NNS) search on a metric space is tightly related to the expansion of the metric space: Given a metric space we look at the graph obtained by connecting every pair of points within a certain distance r. We then look at various notions of expansion in this graph relating them to the cell probe complexity of NNS for randomized and deterministic, exact and approximate algorithms. These relationships can be used to ...
[ Full version ]

...
[ Full version ]

...
[ Full version ]

The bottom-up “segmentation followed by recognition” strategy has for some time now given way to feature-based discriminative recognition with significant success. As the number of categories and exemplars per category increases, however, low-level features are no longer sufficiently discriminative, motivating the construction and use of more complex features. It is argued here that these complex features will necessarily be encoding shape and this in turn requires curves and ...
[ Full version ]

Preempt RT is a growing hard real time linux OS. In this session I will present Preempt RT basics and uniqueness, provide a Live demo, present a benchmark of a company that is migrating from vxworks to Preempt RT and discuss the non-techinical aspects of migrating to linux. ...
[ Full version ]

...
[ Full version ]