The Taub Faculty of Computer Science Events and Talks

After a software product is shipped, it typically goes into a maintenance phase whereby related software updates are made available form time to time. Such updates should in principle have a positive effect (e.g. fixing bugs), but in reality the users often favor stability over the possible improvements brought by updates, worrying about the possibility of updates somehow adversely affecting their systems. However, leaving security vulnerabilities fixes unapplied might lead to hig...
[ Full version ]

...
[ Full version ]

Expander graphs have been intensively studied in the last four decades. In recent years a high dimensional theory of expanders has emerged. In this talk I will introduce the notion of high dimensional expanders and some of the motivations for studying them. As opposed to (1-dimensional) expanders, where a random bounded degree graph is an expander, a probabilistic construction of a bounded degree high dimensional expander is not known. A major open problem, formulated by...
[ Full version ]

Multi-write codes hold great promise to reduce write-amplification in flash-based storage devices. In this talk we propose two novel mapping architectures that show clear advantage over known schemes using multi-write codes, and over schemes not using such codes. To evaluate the performance gain we use industry-accepted benchmark traces, as well as synthetically-generated workloads with time locality. The results show write-amplification savings of double-digit percentages, for as...
[ Full version ]

In reactive systems, responses react to interesting occurrences (events). Event detectors allow detecting complex events by gathering information and hierarchically composing lower-level event detectors. In this work, we introduce a CEGAR (Counterexample Guided Abstraction Refinement)-based compositional verification technique for verifying complex event detectors and response guarantees and finding the necessary assumptions of the response specification about lower-level event de...
[ Full version ]

...
[ Full version ]

Cryo-electron microscopy (cryo-EM) is a microscopy technique used to discover the 3D structure of molecules from very noisy images. We discuss how algebra can describe two aspects of cryo-EM datasets. First, we'll describe common lines datasets. Common lines are lines of intersection between cryo-EM images in 3D. They are a crucial ingredient in some 2D->3D reconstruction algorithms, and they can be characterized by polynomial equalities and inequalities. ...
[ Full version ]

...
[ Full version ]

...
[ Full version ]

...
[ Full version ]

We prove that finding an epsilon-approximate Nash equilibrium is PPAD-complete for constant epsilon and a particularly simple class of games: polymatrix, degree 3 graphical games, in which each player has only two actions. As corollaries, we also prove similar inapproximability results for Bayesian Nash equilibrium in a two-player incomplete information game with a constant number of actions, for market equilibrium in a non-monotone market, for the generalized circuit problem ...
[ Full version ]

The Bitcoin cryptocurrency records its transactions in a public log called the blockchain. Its security rests critically on the distributed protocol that maintains the blockchain, run by participants called miners. Conventional wisdom had asserted that the protocol was incentive-compatible and secure against colluding minority groups, i.e., it incentivized miners to follow the protocol as prescribed. I will show that the Bitcoin protocol is not incentive-compatible...
[ Full version ]

We will discuss how binary files are born and created,loaded and executed, about the creation of object files,files linkage and the differences between Linux format (ELF) and Windows format (PE) of such files. These will be combined with both information security and attack methods and techniques: what a vicious compiler can do, how viruses and other malwares create problems and more. ...
[ Full version ]

Crowdsourcing is an increasingly powerful method for combining amateurs' efforts to recreate an expert's abilities. However, across domains from design to engineering to art, few goals are truly the effort of just one person — even one expert. If we can now crowdsource simple tasks such as image labeling, how might we coordinate many peoples' abilities toward far more complex and interdependent goals? In this talk, I present computational systems for gathering and guiding crowds...
[ Full version ]

...
[ Full version ]

...
[ Full version ]

Generation of precise contact motions for freeform geometry is essential for many applications such as mechanical and manufacturing engineering and motion planning in robotics. Yet contemporary approaches construct the contact motions of geometric objects only for simple geometric objects, typically those bounded by line segments and circular arcs. In this talk, we present an efficient algorithm for constructing precise contact motions (within machine precision) to the case of gen...
[ Full version ]

Locally testable codes (LTCs) are error-correcting codes that admit very efficient codeword tests. An LTC is said to be strong if it has a proximity-oblivious tester; that is, a tester that makes only a constant number of queries and reject non-codewords with probability that depends solely on their distance from the code. Locally decodable codes (LDCs) are complimentary to LTCs. While the latter allow for highly efficient rejection of strings that are far from being codewords, LD...
[ Full version ]

We discuss a new type of wireless devices in the domain between RFIDs and sensor networks - Energy Harvesting Active Networked Tags (EnHANTs - http://enhants.ee.columbia.edu). Future EnHANTs will be small, flexible, and self-powered devices that can be attached to objects that are traditionally not networked (e.g., books, toys, clothing), thereby providing the infrastructure for various Internet-of-Things tracking applications. We describe the paradigm shifts associated with the u...
[ Full version ]

...
[ Full version ]

...
[ Full version ]

Spectral CT is a novel, rapidly emerging imaging technique which offers important new functional and specific information. The possibility to acquire CT scans with different x-ray spectra improves the capability to differentiate and characterize materials and tissues. This adds a new dimension to CT imaging, delivering anatomical information and the ability to characterize structures based on their material makeup. The additional level of spectral information in th...
[ Full version ]

We give a new and improved proof that the shrinkage exponent of De Morgan formulae is 2. Namely, we show that for any Boolean function $f: \{0,1\}^n \to \{0,1\}$, setting each variable out of $x_1, ..., x_n$ with probability $1-p$ to a randomly chosen constant, reduces the expected formula size of the function by a factor of $O(p^2)$. This result is tight and improves the work of Hastad [SIAM J. C., 1998] by removing logarithmic factors. As a consequence of our resu...
[ Full version ]

GraphLab started as research project at Carnegie Mellon University were our main goal was implementing machine learning methods in large scale, (and writing papers about it!). Despite thousands of users of our open source software (and many papers written), the experience of setting up the system and using a distributed system was quite frustrating, where our target audience was mainly double PhDs with machine learning and graph theory & distributed systems background. ...
[ Full version ]

...
[ Full version ]

We employ a geometric framework to extend the concept of data normalization to the domain of functions that lie on manifolds. We pose normalization in this context as an embedding of all examples into manifolds nearly isometric to one another. Using novel geometric tools, we propose an implementation for the case of discretized functions on triangulated two-dimensional meshes. We apply the proposed geometric normalization technique to the task of automatic Action Uni...
[ Full version ]

Haifux Club: Topics in advanced Python: * (Advanced) decorators * Descriptors * Multiple inheritance * The magic of "type" * Metaclasses * Slots vs dictionaries * Generators in depth * Async programming...
[ Full version ]

Medial representation of object shape (skeleton and radial function) is a powerful and widely used tool for image shape analysis. Originally, the concept of skeleton was denoted for continuous objects: the skeleton of a closed region in Euclidean plane is a locus of centers of maximum empty circles in this region. And radial function is defined in every skeleton point and is equal to the radius of inscribed circle centered in this point. However, this concept became the most popul...
[ Full version ]

Fluid simulation plays a key role in various domains of science including computer graphics. While most existing work addresses fluids on bounded Euclidean domains, we consider the problem of simulating the behavior of an incompressible fluid on a curved surface represented as an unstructured triangle mesh. Unlike the commonly used Eulerian description of the fluid using its time-varying velocity field, we propose to model fluids using their vorticity, i.e., by a (time varying) sc...
[ Full version ]

We present new job scheduling algorithms and pricing schemes for computing systems. The scheduling mechanisms we design provide guaranteed Service Level Agreements (SLAs) to users (e.g., meeting job deadlines), while obtaining desired properties driven by both system-aware goals and economic considerations. In our framework, users submit jobs along with a value function that specifies the user value (i.e., willingness to pay) as a function of the job completion time. In addition,...
[ Full version ]

The notion of a q-representative set for a family of subsets, originally arising in the Two-Families Theorem of Bollobás, has recently proven to be very useful in the design of parameterized and exact algorithms. In this talk I will explain this notion. Then, to illustrate its usefulness, I will show how it was used by Fomin, Lokshtanov and Saurabh to design a fast algorithm for finding long simple paths in a directed graph. Finally, I will describe a re...
[ Full version ]

Many of our computer networks, not the least of which the Internet, are built upon hop-by-hop destination-based routing. Here, network devices are equipped with a unique address and routers use giant lookup tables to forward packets towards the intended destination based on the address encoded in the header. At the moment, it is not clear whether we will be able to scale the hop-by-hop routing paradigm into the future economically, as the memory requirement for storing th...
[ Full version ]

In cardiac ultrasound, clutter is an artifact that obscures parts of the heart and may cause inaccurate diagnosis. In particular, a cluttered ultrasound signal is seen as a superposition of tissue, clutter and noise components. In this work, we apply a method called Morphological Component Analysis (MCA) for sparse signal separation with the objective of reducing such clutter artifacts. The MCA approach assumes that the signals corresponding to the clutter and the tissue have each...
[ Full version ]

Contemporary storage systems use redundancy - typically either three- way replication or erasure coding - to reduce the risk of permanent data loss due to simultaneous disk failures. Replication greatly reduces usable disk space, thus increasing costs. Erasure coding adds complexity, is not commonly used for mutable data in a distributed setting, and requires high network bandwidth to recover from a failed device. We propose to alleviate these problems with RAID-P, a storage sy...
[ Full version ]

The IOMMU allows the OS to encapsulate I/O devices in their own virtual memory spaces, thus restricting their DMAs to specific memory pages. The OS uses the IOMMU to protect itself against buggy drivers and malicious/errant devices. But the added protection comes at a cost, degrading the throughput of I/O-intensive workloads by up to an order of magnitude. This cost has motivated system designers to trade off some safety for performance, e.g., by leaving stale information in the I...
[ Full version ]

As the amount of data being generated continues to grow at a staggering rate, streaming algorithms are increasingly becoming more important as a practical tool to analyze and make sense of all the information. In practice, such applications generate vast amounts of data in a very short period of time, and hence it is infeasible to store everything. This presents a pressing question: when is it possible to summarize data while still providing approximate solutions with good theor...
[ Full version ]

Network Functions Virtualization (NFV) uses IT-based virtualization technologies to create classes of virtualized network functions, or VNFs, to serve as building blocks for complex communication services. NFV, with its ability to leverage virtualization techniques to run on standard servers and even the Cloud, features the ability to quickly develop and deploy new products and services, as well as a reliable, flexible and elastic platform. The use of NFV can accelerat...
[ Full version ]

This talk will introduce the Network Functions Virtualization (NFV) concept. NFV uses IT-based virtualization technologies to create classes of virtualized network functions, or VNFs, to serve as building blocks for complex communication services. NFV, with its ability to leverage virtualization techniques to run on standard servers and even the Cloud, features the ability to quickly develop and deploy new products and services, as well as a reliable, flexible and elastic platform...
[ Full version ]

...
[ Full version ]

The talk describes the Shape-from-Polarimetry (SFP) method, a new passive optical technique that measures the instantaneous slope field of a water surface under natural lighting conditions, from a single camera. The SFP method relates the two-dimensional slope of the water surface within a pixel field-of-view to the change in polarization of skylight reflecting from the surface. For the first time the SFP technique makes it possible to recover the explicit topography of the ocean...
[ Full version ]

Motivated by Internet targeted advertising, we address several ad allocation problems. While prior work has established that these problems admit no randomized online algorithm with competitive ratio better than $1-\frac{1}{e}~63.2%$ [KVV90,MSVV2005], simple heuristics have been observed to perform much better in practice than suggested by these bounds. We explain this phenomenon by studying a generalization of the bounded-degree inputs considered by Buchbinder et al. [BJN2007], g...
[ Full version ]

Cloud technology is a dynamic and innovative field, which brings with it many changes in the way we perform large-scale computing, supply and consume data, and manage our data online. Recently, Cloud technology has begun to show it's impact on the manner in which Service Providers (SPs) manage their networks. Today's leading SPs have recThe NFV Infrastructure (NFVI) is composed mainly of standard IT servers and Network elements, such as routers and switches. However, some Virtual ...
[ Full version ]

...
[ Full version ]

digital sensors) wireless sensor networks (WSNs). The spatial physical phenomena are observed by a heterogeneous WSN, meaning that it partially consists of analog sensors and partially of digital sensors. The analog sensors transmit their (continuous) noisy observations to the Fusion Centre (FC), while the digital sensors first perform a simple thresholding and then transmit their binary values over imperfect wireless channels to the FC. The resulting observations are mixed conti...
[ Full version ]

Concurrent systems of all types are notoriously hard to program. First one had to understand how to express the algorithm in a concurrent manner. Then, one has to managing the concurrency so that the implementation is correct. Finally the resulting program needs to be both efficient and scalable. These formidable tasks are similar for all concurrent systems---ranging from distributed cyber-physical systems such as claytronics to multicore processors. Traditionally, proposed ...
[ Full version ]

Random dynamic distortions naturally affect images taken through atmospheric turbulence or wavy water. We show how computer vision can function under such effects, and even exploit them, relying on physical, geometric and statistical models of refractive disturbances. We make good use of distortions created by atmospheric turbulence: distorted multi-view videos lead to tomographic reconstruction of large-scale turbulence fields, outdoors. We also demonstrate several approaches to...
[ Full version ]

Cloud technology is a dynamic and innovative field, which brings with it many changes in the way we perform large-scale computing, supply and consume data, and manage our data online. Recently, Cloud technology has begun to show it's impact on the manner in which Service Providers (SPs) manage their networks. Today's leading SPs have recently announced their goal to move towards Network Function Virtualization (NFV), a major shift which means moving their services onto a large, se...
[ Full version ]

(1) Censorship in the Wild, Analyzing Internet Filtering in Syria. Several government authorities worldwide enforce Internet censorship, however, due to the lack of publicly available information and the inherent risks of performing active measurements, it is often hard for the research community to investigate censorship practices in the wild. In this talk I will present the methodology and the results of a measurement analysis of 600GB worth of leaked logs from 7 Blue Coat SG...
[ Full version ]

...
[ Full version ]

In this talk I will describe two recent research efforts addressing challenges in autonomous operation in unknown environments. First, I will present an approach for multi-robot distributed inference over variables of interest, such as robot trajectories, considering the initial relative pose between the robots and multi-robot data association are both unknown. Assuming robots share with each other informative observations, this inference problem is formulated within...
[ Full version ]

In this talk, I will survey some recent developments of mathematical morphology on graphs. A specific focus will be made on some trees that can be built from an image or a dataset, how such trees can be seen as spaces of shapes, and how they can be linked to some classical optimization approaches common in computer vision. I will try as much as possible to hide the mathematical part and to concentrate on clarifying the underlying ideas. ...
[ Full version ]

Bitcoin is the first currency in human history that is decentralized and does not need Trust of a third party (like Banks, Credit Cards, PayPal etc…). It enables people to send money and pay one another directly and instantly, as with cash or gold, over the Internet. It is based on military-grade cryptography, distributed peer-to-peer networks, and the revolutionary BlockChain technology. The talk provides a basic overview of Bitcoin structure, politics and business perspectiv...
[ Full version ]

We present a subdivision based algorithm to compute the solution of an under-constrained piecewise polynomial system of n−2 equations with n unknowns, exploiting properties of B-spline basis functions. The solution of such systems is, typically, a two-manifold in R^n. To guarantee the topology of the approximated solution in each sub-domain, we provide subdivision termination criteria, based on the (known) topology of the univariate solution on the domain’s boundary, and the e...
[ Full version ]

I will present some results of my PhD thesis on online packet scheduling. I'll focus on: - "Buffer Management with Bounded Delay'', which is the time-online variant of single machine weighted throughput maximization for unit-sized jobs; specifically, I will present (or mention) almost all known results on randomized algorithms. Surprisingly, the best known algorithm and its analysis are simple, which contrasts with the best known deterministic algorithm(s). - "Colle...
[ Full version ]

Amir will describe challenges he faces working on large scale app identification, and why is that even important. This talk links to many other things we do at Facebook TLV, including the Onavo apps and our work as part of Internet.org, a Facebook initiative to make internet connectivity available for every human being on this plant. Bio: Amir Gershman is a software engineer in the Onavo data team at Facebook, where he focuses on building big data solutions for...
[ Full version ]

Peer-to-peer swarming, as used by BitTorrent, is one of the \emph{de facto} solutions for content dissemination in today's Internet. By leveraging resources provided by users, peer-to-peer swarming is a simple, scalable and efficient mechanism for content distribution. Although peer-to-peer swarming has been widely studied for a decade, prior work has focused on the dissemination of one commodity (a single file). This work focuses on the multi-commodity case. We...
[ Full version ]

Cyber-Physical Systems (CPS) are embedded systems in which computational and physical systems are heavily interconnected for performing critical applications including avionics, automotive electronics, manufacturing systems, power networks, medical devices and transportation systems. The high-profile and critical nature of CPShas fostered a large and growing body of works on formal methods for supporting model-based system design in the systems development life cycle. The ph...
[ Full version ]

A lecture with a life demo! General Agenda ...
[ Full version ]

Balanced Graph Partitioning is a hard problem. Doing it at large scale on graphs of millions of nodes and edges is even harder. Doing it in a distributed way makes the problem even more challenging. And finally, dong it with linear or even sub linear time and space complexity may sound like pushing the limits too far. In this talk I will present our practical approach to this hard problem motivated by the systems we build and describe two algorithms that solve it in two different ...
[ Full version ]

Cellular metabolism represents fundamental biochemical activities that enable cells to break down food nutrients, generate energy, and produce molecular building blocks required for cell replication. Metabolic processes in living cells involve thousands of enzymes, whose joint activity can be represented via metabolic networks. In these networks, nodes represent small molecules called metabolites, and edges represent biochemical reactions that transform substrate metabolites to pr...
[ Full version ]

It is presently still unknown how to show hardness of learning problems. There are huge gaps between our upper and lower bounds in the area. The main obstacle is that standard NP-reductions do not yield hardness of learning. All known lower bounds rely on (unproved) cryptographic assumptions. We introduce a new technique to this area, using reductions from problems that are hard on average. We put forward a natural generalization of Feige's assumption about the complexi...
[ Full version ]

...
[ Full version ]

Current datacenter networks inherit the principles that went into the design of the Internet, where packet transmission and path selection decisions are distributed among the endpoints and routers. Instead, we propose that each sender should delegate control—to a centralized arbiter—of when each packet should be transmitted and what path it should follow. Fastpass is a datacenter network architecture built using this principle. Fastpass incorporates two fast algorit...
[ Full version ]

Parameterization is a useful tool for handling NP-hard problems in the real world. It aims to reduce the running times of algorithms for such problems, by confining the combinatorial explosion to some parameter k. As this parameter is often significantly smaller than the input size, it allows to develop practical algorithms for non-trivial classes of instances for these problems. In this talk we present a novel framework for developing parameterized algorithms, using construction...
[ Full version ]

It is quite common that multiple human observers attend to a single point of interest. Mutual awareness activity (MAWA) refers to the dynamic of this social phenomena. A peak of a MAWA is known as a mutual awareness event (MAWE) and can be interpreted as a "buzz" event, which draws the attention of many observers. A preferred way to monitor those social phenomenon is with a camera that captures the human observers while they observe the activity in the scene. Our work studies the ...
[ Full version ]

In light of the rapidly increasing amount of data and demand for fast query processing, increasing the efficiency of database operations continues to be a challenging and important task. XML is based on a tree-structured data model. Naturally, the most popular XML querying language (XPath) uses patterns of selection predicates on multiple hierarchically structured elements. These patterns are often abstracted by twig patterns. Finding all occurrences of such a (XML query) twig pat...
[ Full version ]

...
[ Full version ]

In our work we introduce novel methods for processing signals defined on graphs and high dimensional data clouds. Finding efficient methods to represent and process such signals is quite a challenging problem since they are non-uniformly sampled. As such, traditional signal processing methods are usually not helpful, since they are designed for data defined on regular Euclidean grids. Our solution consists of changing the way we look at the data by ''putting an order in the disord...
[ Full version ]

Overlays play a central role in the scalability of many peer-to-peer (P2P) networks and large scale data-center systems. The eclipse attack has been identified as one of the major potential attacks against overlays. In eclipse attacks, an attacker that controls a portion of the nodes in the system eclipses a large fraction of the correct nodes. By eclipsing correct nodes, attackers isolate correct nodes from the rest of the system, and thereby can completely control what these nod...
[ Full version ]

Programmers developing GPGPU applications to run on hybrid systems are mainly facing two types of challenges, achieving better performance and saving energy. One type of hybrid systems is recent SoC platforms. When it comes to such integrated, SoC based systems, there is a high need to optimize for both, energy and performance. Prior research works already suggested approaches and techniques to schedule computations for parallel execution on the different compute devices (hen...
[ Full version ]

...
[ Full version ]

...
[ Full version ]

I will begin by discussing what makes a patch salient. Most previous work assert that distinctness is the dominating factor. The difference between the various algorithms is in the way they compute distinctness. Some focus on the patterns, others on the colors, and several add high-level cues and priors. We propose a simple, yet powerful, algorithm that integrates these three factors. Our key contribution is a novel and fast approach to compute pattern distinctness. ...
[ Full version ]

...
[ Full version ]

Paravirtual I/O allows the host to interpose on the I/O activity of its guest VMs. Recent studies show significant performance benefits when dividing host cores into "VMcores" dedicated to running VMs and "IOcores" dedicated to processing their paravirtual I/O. In the context of rack-scale computing, we propose to push this design further by designating physical host servers to be either "VMhosts" or "IOhosts", such that VMhosts offload their paravirtual I/O processing to remote I...
[ Full version ]

In this talk I will present a cryptanalytic attack, which we call Bug Attacks. This attack utilizes bugs in the hardware implementation of computer instructions. The best known example of such a bug is the Intel division bug, which resulted in slightly inaccurate results for extremely rare inputs. Whereas in most applications such bugs can be viewed as a minor nuisance, we show that in the case of RSA (even when protected by OAEP), Pohlig-Hellman and several other schemes such bug...
[ Full version ]

Today, most computing platforms are parallel, and the amount of parallelism available for the computation is ever increasing. This growing popularity of parallel platforms creates a need for efficient algorithms that utilize the growing degree of potential parallelism. This study presents specific algorithms as well as general techniques for creating fast practical data structures, focusing on data structures that provide a progress guarantee(e.g., wait-free data structures). Addi...
[ Full version ]

We present a simple scheme for width-independent multiplicative approximation algorithms for generalized covering and packing problems. We then present our main contribution- a novel sampling and acceleration technique for this scheme, which enables near-linear time algorithms. Applications of this result include near-linear time and width-free multiplicative approximation algorithms for fractional packing and covering problems and for normalized covering semi-definite programm...
[ Full version ]

The Approximate Nearest Neighbor Field (ANNF) assigns to each patch in one image, a similar patch from another image. It has recently become a basic building block in numerous applications, e.g., denoising, colorization, alignment and image editing. The increase in popularity of ANNF is largely due to the introduction of PatchMatch, an ANNF algorithm that wisely uses image coherency to speed up the computation. More recent algorithms further improve efficiency by...
[ Full version ]

In this talk we would cover the basics of performance profiling using the Linux perf tool, and tracing using ftrace. The talk will consist mainly of demos....
[ Full version ]

The problem of finding the placement of certain number of facilities so that they can serve all the demands efficiently is a very important subject of research. In this talk, we present some fundamental facility location problems in memory-constrained environment. Here the input is considered to be given in a read-only memory and only constant amount of work-space is available for the computation. This model is well-motivated for handling big-data as well as for computing in sma...
[ Full version ]

The topic of the lecture is methods of vision-based navigation for UAV and visual navigation of ground robots from above. The visual navigation of robots is based on the same principles as human visual navigation. Three main methods for visual navigation exist: navigation without any data about terrarium, navigation according to digital terrarium map, navigation according to previously made photos (videos) of terrarium. New patented methods for the visual navigation of ground robo...
[ Full version ]

Secure multiparty computation is an extremely strong and important tool for making distributed computing more secure. General solutions to the problem allow us to carry out any desired computation among a set of players, while making sure that the result is correct and furthermore that this result is the only new information released. This should even hold if some of the players have been corrupted by an adversary. An important issue in this connection is how the adversary chooses...
[ Full version ]

...
[ Full version ]

Feature tracking in video is a crucial task in computer vision. Usually, the tracking problem is handled one feature at a time, using a single-feature tracker like the Kanade-Lucas-Tomasi algorithm, or one of its derivatives. While this approach works quite well when dealing with high-quality video and "strong" features, it often falters when faced with dark and noisy video containing low-quality features. We present a framework for jointly tracking a set of features, which enable...
[ Full version ]

Subspace clustering is the unsupervised learning problem of clustering a collection of data samples drawn from a union of subspaces, according to their spanning subspaces. State-of-the-art algorithms employ a self-expressive data model in order to construct a graph of relations between data samples, and partition the graph using spectral clustering. These algorithms provide excellent performance for small to medium data collections, however, their polynomial complexity (i...
[ Full version ]

We construct a bi-Lipschitz bijection from the Boolean cube to the Hamming ball of equal volume. More precisely, we show that for all even n, there exists an explicit bijection f from the n-dimensional Boolean cube to the Hamming ball of equal volume embedded in $(n+1)$-dimensional Boolean cube, such that for all $x$ and $y$ it holds that $\mathrm{distance}(x,y) / 5 \leq \mathrm{distance}(f(x),f(y)) \leq 4 \mathrm{distance}(x,y)$, where $\mathrm{distance}(,)$ denotes the Hamming d...
[ Full version ]

The problem of securely outsourcing computation to an untrusted server gained momentum with the recent penetration of cloud computing services. The ultimate goal in this setting is to design efficient protocols that minimize the computational overhead of the clients and instead rely on the extended resources of the server. In this talk, we will focus on the outsourced pattern matching problem which is highly motivated in the context of delegatable computing since it offers storage...
[ Full version ]

...
[ Full version ]

There is resolution loss that is caused strictly by noise, without the presence of imaging blur. Past studies analyze resolution limits in single channel, pan-chromatic systems. There is a need to generalize this analysis to color cameras. We analyze two main types of color sensors: Bayer sensors, which yield color by a filter array on top of a single detector array, and full-field systems (e.g. 3CCD). Bayer sensors compromise resolution. In order to quantify this inherent loss, w...
[ Full version ]

Radix page tables as implemented in the x86-64 micro-architecture incur a penalty of four memory references on each TLB miss. The problem aggravates in virtualized environments with nested page tables where every page walk requires 24 memory references. The virtual memory overhead on guest performance can approach 90\% in servers or scientific applications. Trying to mitigate the cost of TLB misses hardware vendors have added MMU caches that store partial translations....
[ Full version ]

Inferring maps between shapes is a long standing problem in geometry processing. The less similar the shapes are, the harder it is to compute a map, or even define criteria to evaluate it. In many cases, shapes appear as part of a collection, e.g. an animation or a series of faces or poses of the same character, where the shapes are similar enough, such that maps within the collection are easy to obtain. Our main observation is that given two collections of shapes whose “shape s...
[ Full version ]

We study the direct-sum problem for $k$-party ``Number On the Forehead'' (NOF) deterministic communication complexity. We prove several positive results, showing that the complexity of computing a function $f$ in this model, on $\ell$ instances, may be significantly cheaper than $\ell$ times the complexity of computing $f$ on a single instance. Quite surprisingly, we show that this is the case for ``most'' (boolean, $k$-argument) functions. We then formalize sufficient conditions ...
[ Full version ]

Degree d Testing is a probabilistic algorithm that given a function f answer whether f is a degree d polynomial or far from such polynomial using "few" queries to f. One natural such tester is the low dimensional tester. This test checks if the degree of f is d on a low dimensional subspace. Another test is the Gowers norm test. This test checks that the a random d+1 derivative is zero. In this talk, we will study the structure of functions that pass those tests with non-negligi...
[ Full version ]

Abstract: We study a model of fairness in secure computation in which an adversarial party that aborts on receiving output is forced to pay a mutually predefined monetary penalty. We then show how the Bitcoin network can be used to achieve the above notion of fairness in the two-party as well as the multiparty setting (with a dishonest majority). In particular, we propose new ideal functionalities and protocols for fair secure computation and fair lottery in this model. ...
[ Full version ]

...
[ Full version ]

Among the many ways we have to model signals, one approach that has found great popularity in the past decade is based on sparse representations. The main focus in our work is exploring novel sparsity-based signal models that go beyond the classical one. In this talk I will present two such contributions. 1) Statistical models based on sparse representations: In this work we introduce statistical dependencies between various components of the sparsity-based model. Th...
[ Full version ]

Lior Ron, Technion CS M.Sc. graduate and for the last eight years a developer at Google, conducted the Google Maps and Google Now projects, and lately also the first Smartphone in collaboration with Motorola. Lior will talk about the next challenges confronting Google and Motorola, including the future Wearable platform, and will share his experiences during a decade in Silicon Valley. More details about Lior Ron in an ...
[ Full version ]

The PCP theorem (Arora et. al., J. ACM 45(1,3)) says that every NP-proof can be encoded to another proof, namely, a probabilistically checkable proof (PCP), which can be tested by a verifier that queries only a small part of the PCP. A natural question is how large is the blow-up incurred by this encoding, i.e., how long is the PCP. The state-of-the-art work of Ben-Sasson and Sudan (SICOMP 38(2)) and Dinur (J. ACM 54(3)) shows that for any language in NTIME(t), one can construct P...
[ Full version ]

Algebraic Multigrid (AMG) methods have long been recognized for their efficiency as solvers of sparse linear systems of equations, mainly such that arise from discretizations of Partial Differential Equations (PDE). During the past 10 years, a great effort was invested in extending the applicability of AMG methods to other types of problems, mainly by developing adaptive versions of these methods that require fewer assumptions on the underlying systems. Our work is a part of thi...
[ Full version ]

Translucent materials are ubiquitous, and simulating their appearance requires accurate physical parameters. However, physically-accurate parameters for scattering materials are difficult to acquire. We introduce an optimization framework for measuring bulk scattering properties of homogeneous materials (phase function, scattering coefficient, and absorption coefficient) that is more accurate, and more applicable to a broad range of materials. The optimization combines stochastic ...
[ Full version ]

...
[ Full version ]

...
[ Full version ]

...
[ Full version ]

In this seminar talk, we introduce linear quasi-cyclic codes over finite fields. We recall the spectral analysis of Semenov-Trifonov (ST) and explain their BCH-like lower bound on the minimum distance of quasi-cyclic codes. Furthermore, we propose a new bound that generalizes the ST approach and give a syndrome-based algebraic decoding algorithm up to the new bound. Joint work with San Ling....
[ Full version ]

Cardinality estimation algorithms receive a stream of elements that may appear in arbitrary order, with possible repetitions, and return the number of distinct elements. Such algorithms usually seek to minimize the required storage at the price of inaccuracy in their output. In this talk we study the weighted generalization of the cardinality estimation problem, where each item is associated with a weight and the goal is to estimate the total sum of weights. We show how to g...
[ Full version ]

A primary aim of combinatorial game theory (CGT) is to formulate tractable winning strategies for games. I wish to sell you the idea that sometimes the only known method to do so is via a judiciously chosen numeration system, analogously to the choice of an appropriate data structure for optimization problems. We will also see that numeration systems may be conducive to solve elegantly other problems in math. Bio: EE Technion graduate. PhD in math UCLA. A...
[ Full version ]

Linux device drivers typically call kmalloc(), request_mem_region(), ioremap() and other kernel API functions to obtain resources. Failing to release them correctly when the device is removed leads to resource leaks and possibly oopses. Even worse, if the initialization fails in the middle, the allocation must be unrolled, which is typically done with goto statements which are bug-prone. The Device Managed Resources (devres, devm) API was added to the kernel in 20...
[ Full version ]

I will describe some of our recent work on learning systems for image segmentation using a two stage approach: given an image, we first obtain a /diverse/ set of top M most probable segmentations from a discrete probabilistic model, and then rank these using a discriminatively trained ranker that makes use of much more complex features than what could be tractably used in the initial model. The ranking model is learned to minimize the gap with the best segmentation in the set. ...
[ Full version ]

Cancer is diagnosed by a biopsy – a pathologist’s examination of a tissue sample via a microscope. Biopsy image is very rich with data and contains many variations and artifacts, which makes the analysis of it to be very difficult. Most of the previous approaches for segmentation and classification of biopsy images do not understand the meaning of each pixel and the structure of its basic elements. In this talk, I will present a novel biopsy analysis system...
[ Full version ]

Given the popularity of Linux and the increasing adoption of embedded multicore platforms, Linux (and Linux-like RTOSs such as LynxOS and QNX) are increasingly used to host real-time workloads on multiprocessors. This naturally exposes new limitations, both in the actual implementation and in the supporting analytical foundations, and has thus been a driving force for a number of recent advances in multiprocessor real-time scheduling. In this talk, I'm going to present ...
[ Full version ]

The rank modulation scheme has been proposed for efficient writing and storing data in non-volatile memory storage. In this setup data is represented by permutations in Sn, where Sn is the symmetric group of order n. Error-correction in the rank modulation scheme is done by considering codes in Sn, usually under the Kendall's tau-metric. The Kendall's tau-metric between two permutations is the minimum number of adjacent transpositions required in order to change one to the ot...
[ Full version ]

Property testing algorithms are required to discern between inputs with some property and inputs far from having the property, by only making very few queries into the input. The number of queries used by a property testing algorithm is its query complexity. For many properties we have strong lower bounds for the query complexity, but these properties can be decomposed into few sub-properties for which we do have property testing algorithms with low query complexity. In this t...
[ Full version ]

We study linear and quadratic Boolean functions in the context of property testing. We do this by observing that the query complexity of testing properties of linear and quadratic functions can be characterized in terms of complexity in another model of computation called parity decision trees. The observation allows us to characterize testable properties of linear functions in terms of the approximate $l_1$ norm of the Fourier spectrum of an associated function. It also...
[ Full version ]

The choice between Byzantine and crash fault tolerance is viewed as a fundamental design decision when building fault tolerant systems. We show that this dichotomy is not fundamental, and present a unified model of fault tolerance in which the number of tolerated faults of each type is a configuration choice. Additionally, we observe that a single fault is capable of devastating the performance of existing Byzantine fault tolerant replication systems. We argue that fault tolerant ...
[ Full version ]

...
[ Full version ]

The person re-identification (ReID) problem has lately received increasing attention especially due to its important role in surveillance systems, which should be able to keep track of people after they have left the field of view of one camera and entered the field of view of any overlapping or non-overlapping camera. It was shown that person ReID accuracy can be significantly improved given a training set that demonstrates changes in appearances associated with the two non-over...
[ Full version ]

Additive combinatorics is the branch of mathematics whose objects of study are subsets of integers (or other mathematical groups), and which studies the properties and patterns in these subsets that can be expressed via the basic operations of addition or multiplication. One of the central conjectures in additive combinatorics is the polynomial Freiman-Ruzsa conjecture which attempts to classify approximate subgroups of abelian groups. In a recent breakthrough [Sanders, Anal. PDE...
[ Full version ]

Suppose two parties, Alice and Bob, have access to a trusted third party that can do some \emph{simple} computation for them. How can we find out whether the trusted third party's functionality is sufficient to implement arbitrary secure computation from it? The oblivious transfer (OT) functionality just takes as input a tuple of bits $(b_0,b_1)$ from Alice and a choice bit $c$ from Bob and then outputs $b_c$ to Bob. Joe Kilian showed in 1988 that OT is complete in...
[ Full version ]

e study the complexity of computational problems when multiple instances are given, and the instances are *correlated*. Such instances arise for example when they are derived from one "underlying source". Does this make the problem easier? We study this question in the domain of constraint satisfaction problems (CSPs). For example: given several CSP instances with solutions that are 99% the same, and for which the constraints are 95% the same, is it easier to find the ...
[ Full version ]

In contemporary 3-dimensional computer graphics, the graphics pipeline consists of a sequence of affine transformations composed at the end with a single perspective projection. Typically, affine and projective transformations are represented by 4׳4 matrices, and 3-dimensional points and vectors are represented with four homogeneous coordinates. In this framework, quadric surfaces are also usually modeled by (symmetric) 4׳4 matrices. Recently, however, sever...
[ Full version ]

The Alon-Roichman theorem states for any finite group $G$ and $\lambda>0$, a multiset $S$ of size $O(\log |G|/\lambda^2)$ picked uniformly at random is a $\lambda$-spectral expander with high probability. Wigderson and Xiao derandomized the Alon-Roichman theorem when the group $G$ is given as a multiplication table. In this talk we study this problem when the group $G$ is a permutation group and is given by a set of generating elements. We will see an algorithm to construct an e...
[ Full version ]

Massive graph datasets are used operationally by providers of internet, social network and search services. Sampling can reduce storage requirements as well as query execution times, while prolonging the useful life of the data for baselining and retrospective analysis. Here, sampling must mediate between the characteristics of the data, the available resources, and the accuracy needs of queries. Inference methods can be used to fuse datasets which individually provide only an inc...
[ Full version ]

...
[ Full version ]

Oded Cohn, a Technion CS graduate who is now a Vice President at IBM and the Director of IBM Research - Haifa Lab, will share his perspective on technology career development, academia vs. industry, startup vs. large corporation, research and innovation. In particular, he will discuss the question of why is it important to continue studying for advanced degrees, MSc and PhD, in todays employment environment which is happily offering positions for Technion CS/EE Bachelors. Examples...
[ Full version ]

The number of guest virtual machines that can be consolidated on one physical host is typically limited by the memory size, motivating memory overcommitment. Guests are given a choice to either install a "balloon" driver to coordinate the overcommitment activity, or to experience degraded performance due to uncooperative swapping. Ballooning, however, is not a complete solution, as hosts must still fall back on uncooperative swapping in various circumstances. Additionally, balloon...
[ Full version ]

In their seminal work on non-malleable cryptography, Dolev, Dwork and Naor, showed how to construct a non-malleable commitment with logarithmically-many "rounds"/"slots", the idea being that any adversary may successfully maul in some slots but would fail in at least one. Since then new ideas have been introduced, ultimately resulting in constant-round protocols based on any one-way function. Yet, in spite of this remarkable progress, each of the known constructions of non-mall...
[ Full version ]

How much adversarial error can protocols for interactive communication tolerate? This question was examined previously by Braverman and Rao [STOC 2011] for the case of “robust” protocols, where intuitively each party has a fixed and predetermined “order of speaking.” All previous work in coding for interactive communication focused on robust protocols. We consider a new class of protocols for Interactive Communication, namely, adaptive protocols. Such protocol...
[ Full version ]

In this talk we extend the notion of list decoding to the setting of interactive communication and study its limits. In particular, we show that any protocol can be encoded, with a constant rate, into a list-decodable protocol which is resilient to a noise rate of up to $1/2-\varepsilon$, and that this is tight. Using our list-decodable construction, we study a more nuanced model of noise where the adversary can corrupt up-to $\alpha$ fraction of Alice's communication a...
[ Full version ]

Clock gating is very useful for reducing the power consumed by digital systems. Three gating methods are known: synthesis-based, data-driven and auto-gated FFs (AGFF). We present a novel method called Look-Ahead Clock Gating (LACG), which combines all the three. LACG computes the clock enabling signals of each FF one cycle ahead of time, based on the present cycle data of those FFs on which it depends. It avoids the tight timing constraints of AGFF and data-driven by allotting a f...
[ Full version ]

...
[ Full version ]

Image denoising is used to find the best estimate of the original image given its noisy version. The Non-Local Means (NLM) denoising algorithm compares pixel neighborhoods within an extended search region in the image. Each pixel value is estimated as a weighted average of all other pixels in this search region, such that pixels with a similar neighborhood are assigned higher weights. This denoising approach refers to Additive White Gaussian Noise (AWGN). The participation of diss...
[ Full version ]

...
[ Full version ]

The single-graph multiple-flows (SGMF) architecture combines coarse-grain reconfigurable computing with dynamic dataflow to deliver massive thread-level parallelism. The CUDA-compatible SGMF architecture is positioned as an energy efficient design alternative for GPGPUs. The architecture maps a compute kernel, represented as a dataflow graph, onto a coarse-grain reconfigurable fabric composed of a grid of interconnected functional units. Each unit dynamically schedules instances o...
[ Full version ]

The NetFPGA is an open platform enabling researchers and instructors to build high-speed, hardware-accelerated networking systems. The NetFPGA is the de-facto experimental platform for line-rate implementations of network research and it has a family of boards, supporting from 1GE to 100GE. The target audience is not restricted to hardware researchers: the NetFPGA provides the ideal platform for research across a wide range of networking topics from architecture to alg...
[ Full version ]

Is it possible to perform BM3D in real-time? Probably not, but it can be approximated. In this talk we will present an algorithm for video patch-matching that enables real-time video processing for a variety of applications, such as colorization, denoising, or artistic effects. We name our algorithm RIANN - Ring Intersection Approximate Nearest Neighbor - since it finds potential matches by intersecting rings around key points in appearance space. RIANN's real-time performance ...
[ Full version ]

...
[ Full version ]

Obtaining a non-trivial (super-linear) lower bound for computation of the Fourier transform in the linear circuit model has been a long standing open problem for over 40 years. All lower bounds so far have made strong restrictions on the computational model. An early result by Morgenstern from 1973, provides an $\Omega(n \log n)$ lower bound for the unnormalized Fourier transform when the constants used in the computation are bounded. The proof uses a potentia...
[ Full version ]

...
[ Full version ]

The traditional model for computing over a communication network (called LOCAL) allows sending a message of arbitrary size in a single time step. This way, the time complexity is a measure of the locality of algorithms: saying that an algorithm runs in time T is equivalent, under the LOCAL model, to saying that the problem can be solved if each node learns all information the nodes which are reachable within T hops. Therefore, in this model any problem can be solved in time lin...
[ Full version ]

The inventor and former Vice Manager of M-system will lecture on Flash Technology, on the birht of the Disk-on-Key and on the development of "Junior" - the winning chess software. You are all invited....
[ Full version ]

Lightweight process virtualization is a virtualization technology which exists for many years, in Linux and in other OSs, and which matured significantly in the recent year. Linux-based containers are based on this technology. In this lecture, I will discuss the Linux kernel infrastructure for Linux-based containers as well as several interesting userspace projects which use this technology. Following is a list of the main topics which will be discussed: Name...
[ Full version ]

The design and management of current distributed systems is a very complex task. This is mainly due to the fact that typical systems are very large and are often not controlled by a single entity. For example, the Internet is composed of independent administrative entities, called Autonomous Systems (ASs), and the overall behavior is determined by a non-trivial combination of the different policies of each AS and the actions of the end-users. When designing such a system, one mus...
[ Full version ]

Hardware virtualization has long been studied after being introduced to commodity servers. Despite the ongoing research and the developing hardware support, virtual machines incur degraded performance in a wide variety of cases, especially when an unmodified virtual machine operating system is used. One of the major causes of the experienced degraded performance is the lack of physical hardware transparency in virtual machines, since the hypervisor---their controlling software-lay...
[ Full version ]

We show that the existence of a coin-flipping protocol safe against any non-trivial constant bias (e.g., .499), implies the existence of one way functions. This improves upon a recent result of Haitner and Omri [FOCS’11], who proved this implication for protocols with bias .207. Unlike the result of Haitner and Omri, our result holds also for weak coin-flipping protocols. Joint work with Iftach Haitner and Aris Tentes. ...
[ Full version ]

The well known impossibility result of Cleve (STOC 1986) implies that in general it is impossible to securely compute a function with complete fairness without an honest majority. Since then, the accepted belief has been that nothing non-trivial can be computed with complete fairness in the two party setting. The surprising work of Gordon, Hazay, Katz and Lindell (STOC 2008) shows that this belief is false, and that there exist some non-trivial (deterministic, finite-domain) b...
[ Full version ]

The immune system has several roles: protection against parasitism by viruses, bacteria and foreign or aberrant cells; repair of organ and tissue damage; and maintenance of integrity. Hence, beside ongoing routine tasks, it needs to be prepared for unforeseen - even unforeseeable - troubles. However, effective immunity has to be economical; investment in immunity must be balanced with other fitness traits. An organism needs to eat, grow, reproduce and so on. Our immune system is f...
[ Full version ]

...
[ Full version ]

Linux Containers is a technology based on namespaces and cgroups. We will start with an half an hour rehearsal about namespaces and cgroups, which are the building blocks of Linux Containers. Then we will discuss Linux Containers implementation and Checkpoint/Restart....
[ Full version ]

I will discuss new methods and studies that aim to improve eyes-free data entry for blind mobile device users. Currently, mobile devices are generally accessible to blind people, but text entry is almost prohibitively slow. Studies show that blind people enter text on an iPhone at a rate of just 4 words per minute. I will present Perkinput, a chording text entry method where users touch the screen with one to three fingers at a time in patterns based on Braille. In...
[ Full version ]

Abstract: For a string $a \in \{0,1\}^n$ its $k$-fold direct sum encoding is a function $f_a$ that takes as input sets $S \subseteq [n]$ of size $k$ and outputs $f_a(S) = \sum_{i \in S} a_i$. In this paper we are interested in the Direct Sum Testing Problem, where we are given a function $f$, and our goal is to test whether $f$ is close to a direct sum encoding, i.e., whether there exists some $a \in \{0,1\}^n$ such that $f(S) = \sum_{i \in S} a_i$ for most inputs $S$. ...
[ Full version ]

Work stealing is the method of choice for load balancing in task parallel programming languages and frameworks. Yet despite considerable effort invested in optimizing work stealing task queues, existing algorithms issue a costly memory fence when removing a task, and these fences are believed to be necessary for correctness. We will refute this belief, demonstrating fence-free work stealing algorithms for microarchitectures with a bounded total store ordering (TSO) memory model. B...
[ Full version ]

The main role of RNA as a mediator in the process of protein construction out of DNA information, as expressed in Crick's "central dogma of molecular biology", is challenged by recent discoveries regarding the amount of non-protein-coding RNA (ncRNA) being transcribed, and the diverse functionalities of some of these molecules. Some functional RNA molecules are known for several decades now (e.g. tRNA and ribosomal RNA), and other were more recently discovered (e.g. riboswit...
[ Full version ]

Photographs of hazy scenes typically have low-contrast and offer a limited scene visibility. I will describe a new method for single-image dehazing that relies on a generic regularity in natural images in which pixels of small image patches exhibit one-dimensional distributions in pixel space. I will derive a local formation model that explains this formation in the context of hazy scenes and use it for recovering the scene transmission based on the models' offset. Moreover, this ...
[ Full version ]

Time of Flight cameras are an increasingly popular consumer imaging modality that obtain range maps at high framerates. For commercial cameras, such as the new Microsoft Kinect, the basic operating principle is to emit a square wave or pulse signal, and measure the time it takes to reflect off scene objects and return to the camera. In nanosecond codedcameras, we embed a carefully chosen code into the emitted signal, so that when it comes back, we can decode more information. We w...
[ Full version ]

Want to volunteer in a significant network project and receive credit? Come and meet the Wikipedia community representatives and receive tools for academic writing and working in the Wikipedia community. 12:30 Wikipedia - Operation Workshop Yuval Norik, Technion student and Wikipedia system operator 13:30 Academic Writing Lior Golgar, Board Member in Israel Wikimedia Society More details on Wikimedia Poject Course: t...
[ Full version ]

ZeroCash is a new decentralized anonymous payment (DAP) system. It is an extension of decentralized ledger-based system like Bitcoin and solves the anonymity problem in those systems by use of efficient implementations of Zero-Knowledge Proofs. I'll start with a high-lever theoretical overview of Bitcoin and time permitting will go into theoretical considerations of implementing ZK proofs. Joint work with Alessandro Chiesa (MIT), Christina Garman (JHU), Mathew Green (J...
[ Full version ]

Information theoretically secure multi-party computation implies severe communication overhead among the computing participants, as there is a need to reduce the polynomial degree after each multiplication. In particular, when the input is (practically) unbounded, the number of multiplications and therefore the communication bandwidth among the participants may be practically unbounded. In some scenarios the communication among the participants should better be avoided altogether,...
[ Full version ]

The current landscape of proprietary Ethernet switches and e-switches limits the foundation of compute and storage clouds and Web 2.0 infrastructures. The “Generation of Open Ethernet” initiative requires a networking operating system that enables both invocation of independent development while leveraging the strength of an open source community.. Linux as the foundation for most of the commercial networking stacks is a nature candidate for driving both legacy swit...
[ Full version ]

With their ability to image with isotropic resolution of up to 4nm per pixel, Scanning Electron Microscopes (SEM) have become invaluable tools for studying intra-cellular structures and model organelles such as mitochondria, synapses, and vesicles. To acquire image stacks from 3D tissue samples, SEMs scan a rectangular region of the block face several times and average the results to form a clear image of the tissue surface. A thin layer from this surface is then milled and the pr...
[ Full version ]

...
[ Full version ]

...
[ Full version ]

In recent years, social networks have become an important part of our lives. Recent surveys (Facebook Reports Third Quarter 2013 Results) has shown that Facebook alone holds 1.19 Billion monthly active users. 728 Million out of them are daily active users. All these users interact solely with the Facebook cloud. A huge data center located far from their homes. By decentralizing the social network service we not only improve overall network performance but also prevent a massive da...
[ Full version ]

The presentation provides an overview on some short and long term solutions in metro and backbone networks that can facilitate reaching the next level of reliability, which can enable the Internet to become an always operating and fast communication system for the society. In particular I am focusing on fast failure localization and restoration approaches in optical backbone networks, and on the failure recovery techniques at the IP layer. Furthermore, I introduce a future directi...
[ Full version ]

In this talk we concentrate on finding the best migration plan, that is, a partial ordering of live migrations that realizes a move from the current to the desired placement, takes the minimal possible time, and maintains the placement constrains throughout the process. This is not an easy task since additional resources and intermediate migrations may be needed in order to maintain feasibility; in fact, we show that even for a simple model, capturing only the critical aspects of ...
[ Full version ]

Abstract: The graph isomorphism problem is one of the most celebrated computational problems whose complexity status lies somewhere intermediate between P and NP-complete. We revisit two very different-looking relaxations of the problem. In the first relaxation we require the graphs to preserve the number of types of local neighborhoods through the well-known vertex-refinement heuristic and its obvious extension to refinement of $k$-tuples. In the second relaxation we write the na...
[ Full version ]

...
[ Full version ]

In this talk, I discuss the heat method for computing the geodesic distance to a specified subset (e.g., point or curve) of a given domain. The heat method is robust, efficient, and simple to implement since it is based on solving a pair of standard linear elliptic problems. The resulting systems can be prefactored once and subsequently solved in near-linear time. In practice, distance is updated an order of magnitude faster than with state-of-the-art methods, while maintaining a ...
[ Full version ]

MediaWiki is a Free software package for setting up a wiki site. It was originally developed to support Wikipedia, and now it is used by thousands of other sites. The lecture will show the general structure of the software on the level of user and on the level of the developer, it will demonstrate some of its unique features such as broad internationalization support and sophisticated JavaScript loading techniques, and will give the participants an opportunity to contribute to its...
[ Full version ]

Next-Generation Sequencing (NGS) facilitates genetic studies to discover SNPs and indels associated with Mendelian and complex diseases. The measurement process, which generates millions of short reads, creates various data processing and interpretation challenges for which a multitude of software tools are being developed. A common framework used to date to discover variations in sequenced data includes the following steps in a pipeline: First, mapping the sequenced reads to so...
[ Full version ]

In this talk we will describe a method for building surface mesh mappings, given a coarse set of correspondences. The focus will be on how to generate bijective mappings with low isometric distortion. This is a joint work with Noam Aigerman and Roi Poranne...
[ Full version ]

We study the complexity of realizing the "worst" functions in several standard models of information-theoretic cryptography. In particular, for the case of security against passive adversaries, we obtain the following main results. - OT complexity of secure two-party computation: Every function $f:[N]\times [N]\to\{0,1\}$ can be securely evaluated using $O({N^{2/3}})$ invocations of an oblivious transfer oracle. A similar result holds for securely sampling a uniform pa...
[ Full version ]

The recent growing popularity of cloud-based solutions and the variety of new applications present new challenges for cloud management and resource utilization. In this paper we concentrate on the networking aspect and consider the placement problem of virtual machines (VMs) of applications with intense bandwidth requirements. Optimizing the available network bandwidth is far more complex than optimizing resources like memory or CPU, since every network link may be used by many ph...
[ Full version ]

...
[ Full version ]

In today's information era, data centers are facing challenging tasks of managing, analyzing, and storing enormous amounts of data. One such task is recovery, namely, the problem of storing information with added redundancy in order to provide resiliency to erasures. The key requirement of the problem is the locality property, which assumes that it is possible to recover each element of the information by accessing a small of amount of other information blocks. Clearly, the proper...
[ Full version ]

The number of guest virtual machines that can be consolidated on one physical host is typically limited by the memory size, motivating memory overcommitment. Guests are given a choice to either install a “balloon" driver to coordinate the overcommitment activity, or to experience degraded performance due to uncooperative swapping. Ballooning, however, is not a complete solution, as hosts must still fall back on uncooperative swapping in various circumstances. Additionally, ballo...
[ Full version ]

Smartphones suffer from limited computational capabilities. A method to mitigate these problems is code offloading: executing application code on a remote server. We introduce COARA, a middleware platform for code offloading on Android that uses aspect-oriented programming (AOP) with AspectJ. AOP allows COARA to intercept code for offloading without a customized compiler or modification of the operating system. COARA requires minimal changes to application source code. Since state...
[ Full version ]

We study the Lattice Isomorphism Problem (LIP), in which given two lattices $L_1$ and $L_2$ the goal is to decide whether there exists an orthogonal linear transformation mapping $L_1$ to $L_2$. Our main result is an algorithm for this problem running in time $n^{O(n)}$ times a polynomial in the input size, where $n$ is the rank of the input lattices. A crucial component is a new generalized isolation lemma, which can isolate $n$ linearly independent vectors in a given subset of $...
[ Full version ]

Future applications will need to use computational accelerators like GPUs to achieve their performance and power goals. However building efficient systems that use accelerators today is incredibly difficult. The main problem lies in the lack of appropriate OS support -- while OSes provide optimized resource management and Input/Output (I/O) services to CPU applications, they make no such services available to accelerator programs. In this talk I will discu...
[ Full version ]

Surprisingly, ”Internal-Denosing” (using internal noisy patches) usually out- performs ”External-Denoising” (using external clean patches), especially in high noise- levels. We analyze and explain this phenomenon. We further show how the ”fractal” property of natural images (cross-scale patch recurrence) promotes a new powerful internal search-space. Since noise drops dramatically at coarser scales of the noisy image, for almost any noisy patch, its unknown clean versi...
[ Full version ]

...
[ Full version ]

Spectral methods proved to be an important and versatile tool in a wide range of problems in the fields of computer graphics, machine learning, pattern recognition, and computer vision, where many important problems boil down to constructing a Laplacian operator and finding a few of its eigenvalues and eigenfunctions (classical examples include diffusion distances, diffusion maps, and spectral clustering). In this talk, I will show how to generalize spectral geometry t...
[ Full version ]

In this talk, we introduce O-structures, a conceptual architectural memory element that can be used to facilitate parallelism in task-based execution models. Much like register renaming, each write to an O-structure creates a new version of program memory at that location. These versions can be accessed concurrently and out of program order. O-structures provide a set of semantics that match the needs of task-based execution models, specifically allowing tasks to synchronize on sp...
[ Full version ]

A write-once-memory is a type of memory in which cells can only be written once. I.e. we can write “1” to a cell that currently has the value “0”, but not the other way around. It is not hard to show that if one wishes to use the memory to store t messages (i.e. write to the memory t times), then the total number of information bits that can be stored is at most log(t+1)*n, when n is the number of memory cells. It is a challenging task to obtain encoding schemes that achie...
[ Full version ]

Storage-class memory (SCM) technologies such as phase-change memory, spin-transfer torque MRAM, and memristers promise the performance and flexibility of DRAM with the persistence of flash and disk. In this talk, I will discuss two interfaces to persistent data stored in SCM. First, I will talk about Mnemosyne, which is a system that exposes storage-class memory directly to applications in the form of persistent regions. With only minor hardware changes, Mnemosyne suppo...
[ Full version ]

...
[ Full version ]

Finding corresponding points between images is challenging, particularly when objects change their pose non-rigidly, in wide-baseline conditions, or when instances of a perceptual category are compared. In this talk I will present an algorithm for finding a geometrically consistent set of point matches between two images. Given a set of candidate matches that may include many outliers, our method seeks the largest subset of these correspondences that can be aligned perfectly using...
[ Full version ]

This lecture is a quick and unformal guide to scripts and sophisticated commands in Bash. The aim is to supply the listener with tools to use the command-line interface as the strong tool that it is, as well as to write real-life scripts. Rather than explaining the syntax formally, it will be demonstrated in many small examples, showing the use and abuse of Bash. The subjects that were chosen for this lecture are those that the lecturer himself needed for his own tasks....
[ Full version ]

Automatic sign language recognition has close connections with both computer vision and speech recognition. The linguistics of sign languages is less well understood than that of spoken languages, and sign language recognition is much less advanced than speech recognition. We consider American sign language (ASL), and focus on recognition of one constrained but important part of the language: fingerspelling, in which signers spell out a word as a sequence of handshapes or hand tra...
[ Full version ]

You are invited to celebrate Prof. Ron Pinter's 60th birthday and take part in a symposium: "At the interface of Biology and Computation". The symposium will be held on Monday, January 6 2014 from 9:00 to 17:15 at the Technion Coler-California Visitors Center. Participation is free but requires pre-registration. More details and program ...
[ Full version ]

...
[ Full version ]

The availability of three dimensional (3D) protein structures has increased dramatically over the past years. While at the start of the millennium, the Protein DataBank (PDB) held approximately 14,000 (3D) structures of biological macromolecules to-date this database holds 95,000 structures, 25,000 of those added during the last four years. As the total number of structures increased, the number of structures that cannot be classified by established bioinformatics methods ...
[ Full version ]

Over the past three decades, communication complexity has found applications in nearly every area of computer science, and constitutes one of the few known techniques for proving unconditional lower bounds. Developing tools in communication complexity is thus a promising approach for making progress in other computational models such as circuit complexity, streaming, data structures and privacy to mention a few. One striking example of such tool is information th...
[ Full version ]

In this talk, I will discuss results from our recent work of last year (i.e. 2013!) on the classic problem of Leader Election. It is perhaps surprising that this classic problem still yields such rich results. We look at Leader Election in the synchronous message passing setting. Most of this work deals with randomization and we have discovered definitive lower bounds and efficient algorithms. Majority of this talk will focus on [1], in which we introduce the first su...
[ Full version ]