The Taub Faculty of Computer Science Events and Talks

We consider the 0-Extension problem, where we are given an undirected graph G=(V,E) equipped with non-negative edge weights w: E→ℝ+, a collection T={t1,…,tk}⊆V of k special vertices called terminals, and a semi-metric D over T. The goal is to assign every non-terminal vertex to a terminal while minimizing the sum over all edges of the weight of the edge multiplied by the distance in D between the terminals to which the endpoints of the edge ...
[Full version]

Thin structures are ubiquitous, such as ropes, cables, tree branches, wire arts, and wire-frame furniture. It is well known that 3D thin structures are challenging to reconstruct due to their narrow width and lack of distinct features. However, their reconstruction has received relatively little research attention. Clearly, the detection and reconstruction of these thin objects have many applications in computer graphics, computer vision, and robotics. In this talk I will present our recent works on ...
[Full version]

Our research is focused on two distributed models that allow global communication. The first one is the well-known Congested Clique model and another is the recently introduced Hybrid network model. In this talk, I will present three scheduling algorithms for the Congested Clique model, which we developed during my M.Sc studies. These algorithms run a set of distributed jobs in near-optimal time in a black-box manner, without a priori knowledge of their communication pattern.
[Full version]

Correlation Clustering is an elegant model where given a graph with edges labeled + or −, the goal is to produce a clustering that agrees the most with the labels: + edges should reside within clusters and − edges should cross between clusters. In this work we study the MaxCorr objective, aiming to find a clustering that maximizes the difference between edges classified correctly and incorrectly. We focus on the case of bipartite graphs and ...
[Full version]

Storage processors are a new type of acceleration device that offloads centerpiece storage stack functions from legacy software to high-performance hardware. The legacy software paradigms waste flash capacity and performance, pushing the users to deploy expensive and overprovisioned SSDs. For example, the high write amplification induced by inefficient software limits the applicability of the next-generation low-cost flash technologies like QLC and PLC, as well as ZNS SSDs. Storage processors are purpose-built to exploit flash devices ...
[Full version]

3D computer vision has significantly advanced over the past several decades, with modern algorithms successfully reconstructing entire urban cities. However, many questions remain unexplored, as geometric reasoning alone cannot fully infer the connections among images capturing different parts of the scene or semantic relationships between images captured at distant geographic locations. In this talk, I will present an ongoing line of research that leverages powerful deep networks to address new and exciting problems in 3D ...
[Full version]

We introduce a new deep learning method for point cloud comparison. Our approach, named Deep Point Cloud Distance (DPDist), measures the distance between the points in one cloud and the estimated surface from which the other point cloud is sampled. The surface is estimated locally and efficiently using the 3D modified Fisher vector representation. The local representation reduces the complexity of the surface, enabling efficient and effective learning, which generalizes well between object categories. We ...
[Full version]

Model checking is an automatic verification method that accepts a system model and a specification, and checks whether the model satisfies the specification. CTL is a branching temporal logic suitable for specifying behaviors of both software and hardware systems. In this work, we present a novel approach that combines on-the-fly verification with abstraction in order to obtain an efficient model checking. On-the-fly verification ensures that only parts that are needed for determining the satisfaction of ...
[Full version]

Evolving applications such as autonomous vehicle teleoperation call for high-quality and low latency video transmission. Existing technology allows high-quality video to be transmitted over several parallel cellular channels, with suitable rate control. Leveraging these systems to vehicles in motion involves further challenges due to the higher variability of the cellular channels. Major players in these technologies are the Israeli-based company LiveU and its spinoff DriveU. Here, we present a reinforcement learning based flow control algorithm ...
[Full version]

Programmable switches are advanced network switches that can be programmed to perform certain actions besides routing packets. Programmable switches enable combining the logic and flexibility of software with the speed and efficiency of hardware and thus enjoy the benefits of both. Interstingly, programmable switches are also 2-3 orders of magnitude more power to compute efficient than GPUs and FPGAs. Alas, they offer a very restricted and non-intuitive programming model. Sorting is a fundamental and well ...
[Full version]

We study the social efficiency of contracts as economic mechanisms when multiple principals simultaneously manage a common agent. We consider an incomplete-information setting: The agent chooses an unobservable action that induces both a cost for the agent and an expected value for each principal. The sum of these terms is referred to as the resulting social welfare. The agent’s choice is incentivized by payment schemes (“contracts”) set forth by the principals, who have different private ...
[Full version]

We study the problem of maximizing a monotone submodular function subject to a Multiple Knapsack constraint (SMKP). The input is a set I of items, each associated with a non-negative weight, and a set of bins, each having a capacity. Also, we are given a submodular, monotone and non-negative function f over subsets of the items. The objective is to find a subset of items A⊆I and a packing of the items in the bins, ...
[Full version]

Almost two centuries ago Pierre-Joseph Proudhon proposed social contracts -- voluntary agreements among free people -- as a foundation from which an egalitarian and just society can emerge. A digital social contract is the novel incarnation of this concept for the digital age. Digital social contracts are: Conceptually, voluntary agreements among genuinely-identified people, specified, undertaken, and fulfilled in the digital realm. Mathematically, a novel computational model specifying concurrent, non-deterministic asynchronous agents updating a monotonic shared ...
[Full version]

Recent advances in material science and digital fabrication provide promising opportunities for product design, mechanical and biomedical engineering, robotics, architecture, art, and science. Engineered materials and personalized fabrication are revolutionizing manufacturing culture and having a significant impact on various scientific and industrial works. As new fabrication technologies emerge, effective computational tools are needed to fully exploit the potential of digital fabrication. In this talk, I will discuss how we use the insights from discrete differential ...
[Full version]

The ability to promptly and efficiently detect arbitrarily complex patterns in massive real-time data streams is a crucial requirement for a wide range of modern applications. The ever-growing scale of these applications and the sophistication of the patterns involved makes it imperative to employ advanced solutions that can optimize pattern detection. One of the most prominent and well-established ways to achieve the above goal is to apply complex event processing (CEP) in a parallel manner, ...
[Full version]

Bayesian persuasion, as introduced by Kamenica and Gentzkow in 2011, is the study of information sharing policies among strategic agents. A prime example is signaling in online ad auctions: what information should a platform signal to an advertiser regarding a user when selling the opportunity to advertise to her? Practical considerations such as preventing discrimination, protecting privacy or acknowledging limited attention of the information receiver impose constraints on information sharing. Despite their importance in real-life ...
[Full version]

A streaming algorithm is given a long sequence of items and seeks to compute or approximate some function of this sequence using a small amount of memory. A body of work has been developed over the last two decades, resulting in optimal streaming algorithms for a wide range of problems. While randomized streaming algorithms are well-studied, the vast majority of them are defined and analyzed in the static setting, where the stream is assumed to ...
[Full version]

In this talk i'll be covering several works in the topic of 3D deep learning on pointclouds for scene understanding tasks. First, I'll describe VoteNet (ICCV 2019): a method for object detection from 3D pointclouds input, inspired by the classical generalized Hough voting technique. I'll then explain how we integrated image information into the voting scheme to further boost 3D detection (ImVoteNet, CVPR 2020). In the second part of my talk i'll describe a recent ...
[Full version]

Mutation score is widely accepted to be a reliable measurement for the effectiveness of software tests. Recent studies, however, show that mutation analysis is extremely costly and hard to use in practice. We present a novel direct prediction model of mutation score using neural networks. Relying solely on static code features that do not require compilation or execution of the tests, we predict mutation score with an accuracy better than a quintile. When we include ...
[Full version]

You are invited to the See Beyond Reality Workshop 2020 Workshop  - collaboration of RAFAEL and the Faculty of Computer Science at the Technion, which will take place on Wednesday, December 2, 2020, starting at 09:00, in an online meeting. The seminar will deal with research in the fields of Image Processing, Computer Vision, Computer Graphics. More details in the attached poster and a link to the event will be sent upon registration. Looking forward ...
[Full version]

We are living in an era where the digital world is becoming an inevitable part of our professional and daily lives. Digital creation tools are essential for many professions including design, entertainment, gaming etc. In our daily lives, we all take many pictures or capture many videos each day to record and share our memories. There is a stronger demand to transform such digital workflows into life-like experiences. My research focuses on enabling such a ...
[Full version]

Motivated by applications in machine learning, such as subset selection and data summarization, we consider the problem of maximizing a monotone submodular function subject to mixed packing and covering constraints. We present a tight approximation algorithm that for any constant $\eps >0$ achieves a guarantee of $1-\nicefrac[ ]{1}{\e}-\eps$ while violating only the covering constraints by a multiplicative factor of $1-\eps$. Our algorithm is based on a novel enumeration method, which unlike previously known enumeration techniques, ...
[Full version]

In the "20 questions" game Alice picks a probability distribution over the set {1,...,n}, and draws a secret element x from it. Bob's goal is to construct a strategy to reveal x using only "Yes or No" questions, while minimizing the expected number of questions asked. Bob's success is guaranteed if he uses Huffman code, but then he might ask any possible question. In the paper "Twenty (simple) questions", Dagan et al. showed that the ...
[Full version]

Genome sequencing data is rising exponentially at a rate (125%/year) that is far higher than our current transistor performance scaling (currently 3%/year). New medical and comparative genomics applications have emerged that ensure that the demand for more sequencing will continue to rise, which threatens to overwhelm our current compute capacities. Domain-specific acceleration (DSA), i.e. using specialized hardware for accelerating a narrow domain of algorithms, will enable us to tap the vast potential of this data ...
[Full version]

Image sensors capable of detecting individual photons are typically used in specialized low-light imaging applications. In this talk I will make a case for using time-resolved single-photon sensors as general-purpose image sensors, not limited to photon-starved scenarios. This may seem counterintuitive. Why would one want to use a single-photon sensor for situations other than imaging in low light? What additional information can a single-photon sensor provide that a conventional camera sensor cannot? I will present ...
[Full version]

Kirigami, the traditional Japanese art of paper cutting and folding generalizes origami and has initiated new research in material science as well as graphics. Jere we use its capabilities to perform geometric modeling with corrugated surface representations possessing an isometric unfolding into a planar domain after appropriate cuts are made. We initialize our box-based kirigami structures from orthogonal networks of curves, compute a first approximation of their unfolding via mappings between meshes, and complete the ...
[Full version]

You are invited to the first CS CTF - Capture The Flag competition, in various subjects in programming, information security, algorithmics, artificial intelligence and more. The competition will be held in groups that will face challenges posed by large companies in the industry, faculty members, students and alumni on Friday to Sunday, November 20-22, 2020. Among the first places will be awarded cash prizes courtesy of the companies: Cadence Design Systems Israel CheckPoint Elbit Systems ...
[Full version]

The notion of list-decoding, introduced by Elias in the 1950s, is a natural generalization of unique decoding, where the decoder is allowed to output a list of codewords that contains the transmitted codeword instead of a single codeword. Besides being a fundamental concept in coding theory, list decoding has found diverse applications in theoretical computer science. Error-correcting codes of rate R that are list-decodable up to the relative radius 1−R−ε are said to achieve the ...
[Full version]

State sharding is a common solution to the scalability problem in blockchain systems, allowing nodes to hold a partial view of the system state. With sharding, the processing of a transaction might not be completed locally within a node and potentially requires involvement of multiple shards. Such cross-shards transactions have a negative impact on system performance and are frequent with traditional state partition solutions which are often based on a simple mapping of data into ...
[Full version]

In this talk we explore two contributions for shape analysis. In a first case, we consider surfaces and how local analysis of the angular oscillations and polynomial radial behavior around surface points leads to accurate normal estimation and new integral invariants. A direct application of these integral invariants is geometric detail exaggeration. This however assumes that the shape can be represented as a height function over some parameterization plane in a neighborhood of fixed radius ...
[Full version]

RNNs are a class of Machine learning models used to solve tasks with sequential structure. Since most neural circuits are recurrent, RNNs are often deployed to explain neural activity during computational tasks with a temporal dimension. Unlike their natural counterparts, an RNN’s entire computation is accessible. Yet, most works treat them as a black box and analyze their generated activity directly. A recent line of work uses RNNs as a hypothesis generation tool; The activity ...
[Full version]

In this paper we introduce randomized branching as a tool for parameterized approximation and develop the mathematical machinery for its analysis. Our algorithms improve the best known running times of parameterized approximation algorithms for Vertex Cover and 3-Hitting Set for a wide range of approximation ratios. One notable example is a simple parameterized random 1.5-approximation algorithm for Vertex Cover, whose running time of O*(1.01657k) substantially improves the best known running time of O*(1.0883k) [Brankovic and ...
[Full version]

Monitoring network traffic is an important building block for various management and security systems. In typical settings, the number of active flows in a network node is much larger than the number of available monitoring resources and there is no practical way to maintain "per-flow" state at the node. This situation gave rise to the recent interest in streaming algorithms where complex data structures are used to perform monitoring tasks like identifying the top-$k$ flows, ...
[Full version]

A major bottleneck in the evaluation of music generation is that music appreciation is a highly subjective matter. When considering an average appreciation as an evaluation metric, user studies can be helpful. The challenge of generating personalized content, however, has been examined only rarely in the literature. In this paper, we address generation of personalized music and propose a novel pipeline for music generation that learns and optimizes user-specific musical taste. We focus on the ...
[Full version]

A channel in the Lightning Network allows two parties to secure bitcoin payments and escrow holdings between them. Designed to increase transaction immediacy and reduce blockchain congestion, this has the potential to solve many issues associated with Bitcoin. In this talk, we study the economics of the Lightning Network. We present conditions under which two parties optimally establish a channel and give explicit formulas for channels’ costs. Using these, we derive implications for the network’s ...
[Full version]

A common observation in data-driven applications is that high dimensional data has a low intrinsic dimension, at least locally. Thus, when one wishes to work with data that is not governed by a clear set of equations, but still wishes to perform statistical or other scientific analysis, an optional model is the assumption of an underlying manifold from which the data was sampled. This manifold, however, is not given explicitly but we can obtain samples ...
[Full version]

The potential benefits of taking surface curvature into account in engineering design have not yet been fully utilized. One of the problems is that differential geometry is often not offered in the undergraduate engineering program. In this presentation, I focus on a couple of research topics that demonstrate how curvatures play important roles in engineering design. The first topic is the effect of curvature on the energy absorption characteristics of cylindrical corrugated tubes under compression. ...
[Full version]

Quotient spaces are a natural mathematical tool to describe a variety of algorithmic problems in computer vision and related fields, where different objects are to be compared while their natural symmetries are to be ignored. One popular approach for working in quotient spaces is cancelling out symmetries by finding a canonical alignment of the objects at hand. This approach typically involves solving a challenging optimization problem over the symmetry group. We will review the tractability ...
[Full version]

You are invited to "Exposure to Industry Jobs Workshop - Putting in order Buzz Words" which aims to classify the jobs in the industry, interpret its concepts and products and illustrate a typical work day in the various roles it offers. The workshop will be held online, in two sessions: The first meeting will be held on Thursday, November 5, 2020 at 5:30 pm and will discuss topics lasting 15 minutes each: - Web Track ...
[Full version]

In a polynomial protocol a prover sends messages that are polynomials, and the verifier is allowed to check polynomial identities between these polynomials. The prover complexity is measured as the sum of degrees of the polynomials sent. The motivation for the definition is to capture prover complexity in zero knowledge proofs systems based on polynomial commitment schemes. We will present and illustrate this notion; and present an open question on improved protocols for “range proofs” ...
[Full version]

You are invited to a lecture and conversation with Itai Rosenblatt, scaleIO technical Team Leader and co-founder and VP of R&D at a new start-up, which will deal with the technical interview and the components of success and the initial impression it gives: • What the interviewer is looking for in each part of the interview • What is your chance as a candidate • How to prepare The lecture will take place on Tuesday, ...
[Full version]

The complexity of Iterated Matrix Multiplication is a central theme in Computational Complexity theory, as the problem is closely related to the problem of separating various complexity classes within P. In this talk, we will talk about the complexity of computing an entry in the product of d many n×n generic matrices (denoted IMMn,d) by depth four circuits of bounded individual degree. We show that for all r ≤ na for a fixed constant a, ...
[Full version]

CS 2020-21 orientation day for new students will be held online by zoom broadcast on Tuesday, October 20, 2020 at 10:30, and will include lectures by CS Dean, Vice Dean for Undergraduate Studies, CS Student Committee representatives and more, to be followed by A&Q sessions in small groups with senior UG, Magister and Doctoral students, and with Faculty and administrative staff. More details in UG secretariat. Welcome to CS, good luck and a successful academic ...
[Full version]

You are all invited to the 2019 CS graduation ceremony of , which will be broadcast live on YouTube today, Thursday, October 15, 2020, at 18:30. Looking forward to seeing you!
[Full version]

The planted clique problem plays an important role in the field of average-case complexity. The problem consists of detecting graphs that contain a hidden clique of size $k$ that was planted in graphs from the Erdos-Renyi model $G_{n,0.5}$. The planted clique conjecture suggests that there is no polynomial-time algorithm that can solve the planted clique problem when $2\log_2(n) \ll k \ll \sqrt{n}$. This conjecture has served as the underlying hardness assumption in several different works ...
[Full version]

Reliability and time-efficiency are two key elements to consider in network design. Commonly, each is measured per service - availability probability of a specific service, the latency of a specific service, and overall - system average reliability and system average latency, considering the demand for every service. Intuitively, minimizing latency requires minimizing the number of network elements a service makes use of. In a non-redundant environment, this would also guarantee the maximal reliability of a ...
[Full version]

A statistical zero-knowledge proof for a problem Pi enables a computationally unbounded prover to convince a polynomial-time verifier that x belongs to Pi without revealing any additional information about x to the verifier, in a strong information-theoretic sense. Suppose, however, that the prover wishes to convince the verifier that k separate inputs x_1,...,x_k all belong to Pi (without revealing anything else). A naive way of doing so is to simply run the SZK protocol separately ...
[Full version]

In blockchain applications, transaction latency is a key factor in the quality of service (QoS). Each application that uses the blockchain network has the motivation to reduce the latency for its own transactions. The block proposer selects which pending transaction to include in the proposed block and would like to prioritize its own application transactions over those of other applications. To maintain fairness, a block proposer can be asked to select the transactions randomly among ...
[Full version]

The fields of signal and image processing have been deeply influenced by the introduction of deep neural networks. These are successfully deployed in a wide range of real-world applications, obtaining state of the art results and surpassing well-known and well-established classical methods. Despite their impressive success, the architectures used in many of these neural networks come with no clear justification. As such, these are usually treated as ``black box'' machines that lack any kind of ...
[Full version]

A functional PIR array code is a coding scheme which encodes some s information bits into a tXm array such that every linear combination of the s information bits has k mutually disjoint recovering sets. Every recovering set consists of some of the array's columns while it is allowed to read at most l encoded bits from every column in order to receive the requested linear combination of the information bits. Functional batch array codes ...
[Full version]

We tackle the task of Term Set Expansion (TSE): given a small seed set of example terms from a semantic class, finding more members of that class. The task is of great practical utility, and also of theoretical utility as it requires generalization from few examples. Previous approaches to the TSE task can be characterized as either distributional or pattern-based. We harness the power of neural masked language models (MLM) and propose a novel TSE ...
[Full version]

Shay Moran, Technion CS graduate, will give a lecture on his work in the theory of learning.
[Full version]

A new family of codes, called clustering-correcting codes, is presented in this paper. This family of codes is motivated by the special structure of the data that is stored in DNA-based storage systems. The data stored in these systems has the form of unordered sequences, also called strands, and every strand is synthesized thousands to millions of times, where some of these copies are read back during sequencing. Due to the unordered structure of the ...
[Full version]

Machine learning has made considerable progress over the past decade, matching and even surpassing human performance on a varied set of narrow computational tasks. This progress has been enabled by the widespread availability of large datasets, as well as by improved algorithms and models. Distribution, implemented either through single-node concurrency or through multi-node parallelism, has been the third key ingredient to these advances. The goal of this talk is to provide an overview of the ...
[Full version]

The Technion Ph.D. Students Graduation Ceremony 2020, will be held online TODAY, Wednesday evening, September 16, 2020, and will be broadcast on Facebook. This year, a substantial number of students - 231 Technion graduates in 2019-2020 - will receive their Ph.D. diplomas, among them 14 graduates of The Henry and Marilyn Taub Faculty of Computer Science:   Ben Ezair Titled: Advanced Geometric Methods in Machining and Additive Manufacturing Advised by: Gershon Elber Danielle Ezuz Titled: ...
[Full version]

The Hiroshi Fujiwara Cyber Security Research Center will hold the 8th Summer School on Cyber and Computer Security: "Privacy in Challenging Times". The event will take place on Monday-Thursday, September 7th-10th, 2020, broadcasted by Technion Zoom. Keynote Speakers: Ross Anderson, University of Cambridge, UK Ann Cavoukian, Global Privacy & Security by Design Centre, Canada Susan Landau, Tufts University, USA Confirmed Speakers: Alessandro Acquisti, Carnegie Mellon University, USA Michael Birnhack, Tel Aviv University, Israel Sofía Celi, ...
[Full version]

A proof of work (PoW) blockchain protocol distributes rewards to its participants, called miners, according to their share of the total computational power. Sufficiently large miners can perform selfish mining - deviate from the protocol to gain more than their fair share. Such systems are thus secure if all miners are smaller than a threshold size so their best response is following the protocol. To find the threshold, one has to identify the optimal strategy ...
[Full version]

You are invited to a lecture and conversation with Dr. Yonatan Yaniv, CS graduate and head of the research group at YOTPO, which will deal with the questions: Is it worth working during the degree? How to integrate into the industry, especially now in the Corona times? How do you find the right job for you? How to build a career path? Working in a large company compared to a start-up, and more. The lecture ...
[Full version]

The Shapley value is a conventional and well-studied function for determining the contribution of a player to the coalition in a cooperative game. Among its applications in a plethora of domains, it has recently been proposed to use the Shapley value for quantifying the contribution of a tuple to the result of a database query. In particular, we have a thorough understanding of the tractability frontier for the class of Conjunctive Queries (CQs) and aggregate ...
[Full version]

You are invited to participate in the Virtual Workshop on Machine Learning & Hardware Security that will be held August on Wednesday-Thursday, 26-27, 2020 @ Zoom. Organizers: Prof. Avi Mendelson, Technion, Israel Dr. Shivam Bhasin, NTU, Singapore For more Information, program and registration please visit https://cyber.technion.ac.il Participation is FREE, but requires advance registration.
[Full version]

Scene Text Recognition (STR), the task of recognizing text against complex image backgrounds, is an active area of research. Current state-of-the-art (SOTA) methods still struggle to recognize text written in arbitrary shapes. In this paper, we introduce a novel architecture for STR, named Selective Context ATtentional Text Recognizer (SCATTER). SCATTER utilizes a stacked block architecture with intermediate supervision during training, that paves the way to successfully train a deep BiLSTM encoder, thus improving the encoding ...
[Full version]

Deduplication decreases the physical occupancy of files in a storage volume by removing duplicate copies of data chunks,but creates data-sharing dependencies that complicate standard storage management tasks. Specifically, data migration plans must consider the dependencies between files that are remapped to new volumes and files that are not. Thus far, only greedy approaches have been suggested for constructing such plans, and it is unclear how they compare to one another and how much they can ...
[Full version]

We inspect the fine-grained complexity of answering queries over relational databases. With the ideal guarantees, linear time is required before the first answer to read the input and determine its existence, and then we need to print the answers one by one. Consequently, we wish to identify the queries that can be solved with linear preprocessing time and constant or logarithmic delay between answers. A known dichotomy classifies CQs into those that admit such enumeration ...
[Full version]

Optical character recognition (OCR) systems performance have improved significantly in the deep learning era. This is especially true for handwritten text recognition (HTR ), where each author has a unique style, unlike printed text, where the variation is smaller by design. That said, deep learning based HTR is limited, as in every other task, by the number of training examples. Gathering data is a challenging and costly task, and even more so, the labeling task ...
[Full version]

King and Saia were the first to break the quadratic word complexity bound for Byzantine Agreement in synchronous systems against an adaptive adversary, and Algorand broke this bound with near-optimal resilience (first in the synchronous model and then with eventual-synchrony). Yet the question of asynchronous sub-quadratic Byzantine Agreement remained open. To the best of our knowledge, we are the first to answer this question in the affirmative. A key component of our solution is a ...
[Full version]

Although Magnetic Resonance Imaging (MRI) is a superb medical imaging modality, its clinical use is limited by its long acquisition time. The acquisition time can be shortened by sampling data with a sub-Nyquist rate; however, this requires suitable methods for solving the ill-posed inverse problem of accurate image reconstruction from highly subsampled data. In this seminar, four novel methods for solving the MRI reconstruction problem will be presented. These methods build upon the well-established Compressed ...
[Full version]

We report a series of robust empirical observations, whereby deep Neural Networks learn the examples in both the training and test sets in a similar order. This phenomenon is observed in all the commonly used benchmarks we evaluated, including many image classification benchmarks, and one text classification benchmark. While this phenomenon is strongest for models of the same architecture, it also crosses architectural boundaries -- models of different architectures start by learning the same examples, ...
[Full version]

In this talk, I will describe a couple of results from my PhD thesis, that studies distributed algorithms for connectivity and distance related graph problems. I will focus on a recent line of work studying distance computation in the Congested Clique model of distributed computing. We design fast algorithms for the fundamental all-pairs shortest paths (APSP) and multi-source shortest paths (MSSP) problems. In particular, we show constant-factor approximation algorithms for weighted APSP and MSSP that ...
[Full version]

One of the fascinating properties of deep learning is the ability of the network to reveal the underlying factors characterizing elements in datasets of different types. Autoencoders represent an effective approach for computing these factors and also have been studied in the context of their ability to interpolate between data points by decoding mixed latent vectors. This interpolation often incorporates disrupting artifacts or produces unrealistic images during reconstruction. We argue that these incongruities are due ...
[Full version]

Finding exact solutions to many fundamental optimization and distance computation problems is known to be a hard task in the distributed CONGEST model of computation. A natural relaxation then is to look for approximate solutions. In this talk, I will discuss some of the research I did during my M.Sc. studies that concerns the trade-off between approximation ratio and round complexity for central optimization and distance computation problems.
[Full version]

In the first part of this two-part talk, I will describe my work in adversarial machine learning (ML). Prior work has demonstrated that ML algorithms are vulnerable to evasion attacks at inference time by so-called adversarial examples. However, the implications of attacks in practice, under real-world constraints, remained largely unknown. To fill the gap, we propose evasion attacks that satisfy multiple objectives (e.g., smoothness and robustness against changes in imaging conditions), and show that these ...
[Full version]

In the directed minimum spanning tree problem (DMST, also called minimum weight arborescence), the network is given a root node r, and needs to construct a minimum-weight directed spanning tree, rooted at r and oriented outwards. In this paper we present the first sub-quadratic DMST algorithms in the distributed CONGEST network model, where the messages exchanged between the network nodes are bounded in size. We consider three versions: a model where the communication links are ...
[Full version]

CS invites you to meet Flutter - a new Google application technology - via ZOOM, Wednesday, June 10, 19:00, with: Barak Menachem, Director of the Flutter Israel Developers Community Tamir Haim Rabia, React Native Expert, Flutter In the program: Cross-platform technologies What is Flutter and why to use it HandsOn Workshop and Technology use How to use it to develop academic projects A link to the lecture will be sent after pre-registration
[Full version]

CYBERDAY 2020 event, organized by Eli Biham, Sara Bitan, and the Technion Hiroshi Fujiwara Cyber Security Research Center, will be held in the cyber space (using Zoom - link will be sent towards the event upon registration), on Wednesday, June 10, 2020. In the program: * Hashomer: A Proposal for a Privacy-Preserving Bluetooth Based Contact Tracing Scheme for Hamagen * Neural Reverse Engineering of Stripped Binaries * Adventures in Cryptographic Standardization * Dragonblood: Analyzing the ...
[Full version]

Automated tools and algorithms for synthesizing geometry have been ubiquitous in computer-aided design systems, since their very inception. Beyond the simple example of sketching tools for designers, algorithms for synthesizing geometry take many forms, such as design-by-constraints, optimization, and toolpath generation. In this thesis, we will present three problems in the field on geometry processing, and examine them both within the context of geometry synthesis, and in their broader applications. The first problem we discuss ...
[Full version]

CS invites graduate students to a meeting on the research career in the hi-tech industry: Is there research in the Israeli industry and what are the types of jobs and career paths available, to be discussed by a panel with: Lian Levin Eitan, Senior Manager of Research, Amazon Tal Mizrahi, Principal Architect and Researcher, Huawei Nadav Orbach, General Manager, Imaging & Camera Group, Intel Tamar (Tami) Avraham, Co-Founder and CTO, Coral Detection Systems The meeting ...
[Full version]

Computer system performance has improved due to creatively using more transistors (Moore’s Law) in parallel via bit-, instruction-, thread-, and data-level parallelism. With the slowing of technology scaling, a way to further improve computer system performance under energy constraints is to employ hardware accelerators. Each accelerator is a hardware component that executes a targeted computation class faster and usually with (much) less energy. Already today, many chips in mobile, edge and cloud computing concurrently employ ...
[Full version]

Licklider and Taylor (1968) envisioned computational machinery that could enable better communication between humans than face-to-face. In the last fifty years, we have used computing to develop various means of communication, such as mails, messaging, video conversation, or virtual reality. These are, however, a proxy of face-to-face communication that aims at encoding words, expressions, emotions, and body language at source and try to decode them reliably at the destination. The true potential of computing to ...
[Full version]

Optimization in machine learning, both theoretical and applied, is presently dominated by first-order gradient methods such as stochastic gradient descent. Second-order optimization methods---that involve second-order derivatives and/or second-order statistics of the data---have become far less prevalent despite strong theoretical properties, due to their impractical computation, memory and communication costs. I will present some recent theoretical, algorithmic and infrastructural advances that allow for overcoming these challenges in using second-order methods and obtaining significant performance gains in ...
[Full version]

The idea of light field technology is relatively old, however, apart from various attempts of construction of imaging devices which can densely capture scene radiance, only recent availability of devices from companies like Lytro and Raytrix started a significant increase of interest in the light field technology. The core idea of the light field is in the dense sampling of viewpoints where the dense collection of views distinguishes is from the sparse sampling in multiview ...
[Full version]

In this work, we present a method for node embedding in temporal graphs. We propose an algorithm that learns the evolution of a temporal graph's nodes and edges over time and incorporates this dynamics in a temporal node embedding framework for different graph prediction tasks. We present a joint loss function that creates a temporal embedding of a node by learning to combine its historical temporal embeddings, such that it optimizes per given task (e.g., ...
[Full version]

Two-player turn-based zero-sum stochastic games is an interesting family of infinite duration games played on a finite set of states. These games may be seen as an extension of Markov Decisions Processes (MDPs) to a two-player setting. Such games have applications in various areas ranging from Artificial Intelligence, Operations Research and Automatic Verification. From the theoretical point of view they are interesting because they are known to be in NP intersection co-NP, yet no polynomial ...
[Full version]

Hypernetworks, also known as dynamic networks, are neural networks in which the weights of at least some of the layers vary dynamically based on the input. Such networks have composite architectures in which one network predicts the weights of another network. I will briefly describe the early days of dynamic layers and present recent results from diverse domains: 3D reconstruction from a single image, image retouching, electrical circuit design, decoding block codes, graph hypernetworks for ...
[Full version]

The event will be held in live broadcast by ZOOM -  ID MEETING NO. 518018934 The 2020 open day invites outstanding undergraduates from all universities to come to the Technion and learn about the Computer Science and Department and register for Winter Semester 2020.     The event will be held online by ZOOM -  ID MEETING NO. 518018934, on Wednesday, March 25, 2020. between 12 The program will include review on curriculum, research and life at ...
[Full version]

Voice assistants are designed to make our lives easier. By responding to spoken commands and questions, they provide a natural way of interaction with computers and smartphones. The problem starts when these voice assistants run by default even when the device is locked, which requires vendors to balance comfort with security. In this talk, I will take the audience on an amusing journey of discovering more than twenty new security vulnerabilities in popular voice assistants, ...
[Full version]

Deep Learning has always been divided into two phases: Training and Inference. Deep networks are mostly used with large data-sets, both under supervised (Classification, Regression etc.) or unsupervised (Autoencoders, GANs) regimes. Such networks are only applicable to the type of data they were trained for, and do not exploit the internal statistics of a single datum. We introduce Deep Internal Learning; We train a signal-specific network, we do it at test-time and on the test-input ...
[Full version]

Recently, there is an increase in delegating computations to powerful remote servers. This appearance raises questions of computational integrity and efficient verification. For instance, the server can be unreliable, therefore it may output erroneous results. In other contexts, where preserving information privacy is crucial, as in medical and forensic data, other issues might arise. The veils of secrecy, that are designed to keep the data hidden from the public, may be abused to cover up ...
[Full version]

Contemporary presentation software provides sufficient tools for simple presentation scenarios. However, such tools have not changed for about two decades and are unsuitable for many complex modern needs. Content is typically limited to text and images only and cannot be expressed well in terms of 2D slides and bullets quantization. Such limitations are ill-suited for many modern needs where linear sequences of fixed-sized 2D slides consisting of text and images is not nearly sufficient. In ...
[Full version]

Contemporary presentation software provides sufficient tools for simple presentation scenarios. However, such tools have not changed for about two decades and are unsuitable for many complex modern needs. Content is typically limited to text and images only and cannot be expressed well in terms of 2D slides and bullets quantization. Such limitations are ill-suited for many modern needs where linear sequences of fixed-sized 2D slides consisting of text and images is not nearly sufficient. In ...
[Full version]

Online Portfolio Selection, aiming to optimize the allocation of wealth across a set of assets, is a fundamental research problem in computational finance and machine learning. Despite the theoretical challenges, the implementation of a real-world trading system is extremely challenging. This issue has been extensively studied across several research communities, including finance, statistics, coding and information theory and machine learning. This long-standing problem, however, still poses many challenges. We begin with a discussion of the ...
[Full version]

Applications such as distributed learning and edge computing strive to maximize the number of service requests (e.g., for data access) that can be concurrently executed by the system. Redundancy, in the form of simple replication and erasure coding, has emerged as an efficient and robust way to enable simultaneous access of different data objects by many users competing for the system’s resources. Here, replication and coding of data affect the rates at which users can ...
[Full version]

Bluetooth is a widely deployed standard for wireless communications between mobile devices. It uses authenticated Elliptic Curve Diffie-Hellman for its key exchange. In this paper we show that the authentication provided by the Bluetooth pairing protocols is insufficient and does not provide the promised MitM protection. We present a new attack that modifies the y-coordinates of the public keys (while preserving the x-coordinates). The attack compromises the encryption keys of all of the current Bluetooth ...
[Full version]

In this talk, we will discuss recent advances in multitask regression and their application to flood forecasting. We will begin with a brief overview of Google’s flood forecasting initiative in India. The project involves time series prediction in different geographical locations based on a limited number of heavy tailed samples. To address these challenges, we will consider non-convex robust multitask models based on elliptical distributions. Next, we will present a spectral algorithm for learning shared ...
[Full version]

Magnetic Resonance Imaging (MRI) has long been considered to be among "the gold standards" of diagnostic medical imaging. The long acquisition times, however, render MRI prone to motion artifacts, let alone their adverse contribution to the relative high costs of MRI examination. Over the last few decades, multiple studies have focused on the development of both physical and post-processing methods for accelerated acquisition of MRI scans. These two approaches, however, have so far been addressed ...
[Full version]

The manual detection, examination and repair of computer bugs are all notoriously difficult tasks that programmers face daily. Automated program repair receives a program with a bug, and outputs a set of changes to the program that would make it bug-free. We focus on formal program repair, which means that programs are repaired with respect to a formal specification. Formal program repair, in contrast to test-based repair, guarantees that programs meet the specification for all ...
[Full version]

Medical ultrasound is a widespread imaging modality owing its popularity to cost efficiency, portability, speed, and lack of harmful ionizing radiation. In this work, we demonstrate that replacing the traditional ultrasound processing pipeline with a data-driven, learnable counterpart leads to significant improvement in image quality. Moreover, we demonstrate that greater improvement can be achieved through a learning-based design of the transmitted beam patterns simultaneously with learning an image reconstruction pipeline. We evaluate our method on ...
[Full version]

The recent and on-going Deep-Learning (DL) revolution, introduces a paradigm shift in almost all disciplines of signal processing. Traditionally, Computer Vision (CV) and Image Processing (IP) methods were based on hand-crafted feature extraction from the initial optical image, and then some hand-crafted classifier/filter was defined to achieve the desired result IP/CV. The machine learning approach seeks to learn the 'classifier' stage using data (either labeled or unlabeled), i.e. the decision rule is not explicitly derived ...
[Full version]

We introduce a novel neural network-based partial differential equations solver for forward and inverse problems. The solver is grid free, mesh free and shape free, and the solution is approximated by a neural network. We employ an unsupervised approach such that the input to the network is a points set in an arbitrary domain, and the output is the set of the corresponding function values. The network is trained to minimize deviations of the learned ...
[Full version]

As data analytics becomes more crucial to digital systems, so grows the importance of characterizing the database queries that admit a more efficient evaluation. We consider the tractability yardstick of answer enumeration with a polylogarithmic delay after a linear-time preprocessing phase. Such an evaluation is obtained by constructing, in the preprocessing phase, a data structure that supports polylogarithmic-delay enumeration. In this paper, we seek a structure that supports the more demanding task of a “random ...
[Full version]

We are interested in the following question: How could one create a digital personal assistant, capable of acquiring and applying new skills through spoken interaction? The most rudimentary requirement of such skill acquisition would be Turing complete programming. Other such acquisitions might be the deduction of routines from user actions and their automation thereof. Modern digital personal assistants do not possess the ability to learn through Voice User Interface (VUI), which is their primary communication ...
[Full version]

The recent push towards ultra-high (magnetic) fields in Magnetic Resonance Imaging (MRI) is expected to change the face of biomedical imaging. Recent studies employing state of the art 7 T MRI scanners achieved submillimeter resolution in 3D anatomical imaging of the human brain. However, the higher field also comes with new challenges, especially since we want to acquire the images with high spatial or temporal resolution to study the human brain function. These challenges include ...
[Full version]

We prove that for all constants a, NQP = NTIME[n^polylog(n)] cannot be (1/2 + 2^(-log^a n) )-approximated by 2^(log^a n)-size ACC^0 circuits. Previously, it was even open whether E^NP can be (1/2+1/sqrt(n))-approximated by AC^0[2] circuits. As a straightforward application, we obtain an infinitely often non-deterministic pseudorandom generator for poly-size ACC^0 circuits with seed length 2^{log^eps n}, for all eps > 0. More generally, we establish a connection showing that, for a typical circuit class C, ...
[Full version]

Reconstructing observed images from fMRI brain recordings is challenging. Unfortunately, acquiring sufficient "labeled" pairs of {Image, fMRI} (i.e., images with their corresponding fMRI responses) to span the huge space of natural images is prohibitive for many reasons. We present a novel approach which, in addition to the scarce labeled data (training pairs), allows to train fMRI-to-image reconstruction networks also on "unlabeled" data (i.e., images without fMRI recording, and fMRI recording without images). The proposed model ...
[Full version]

Suppose that in a given planar circular region, there are some smart mobile evaders and we would like to find them using swarms of sweeping agents. A smart evader is a target that detects and responds to the motions of searchers by performing evasive maneuvers, to avoid interception. We assume various search configurations for the sweeping swarm of agents, and present guaranteed search techniques for single agent and multi agent swarms. These search procedures enable ...
[Full version]

In this talk I will review some major advances in the theory of distributed computing in the past decade and discuss key research challenges for the 2020s, with the main focus on distributed graph algorithms. I will present promising new approaches for doing research in the field, and I will also highlight examples of seemingly elementary questions that are still beyond the reach of our current techniques.
[Full version]

CS Labs: Systems and Software Development Laboratory (ICST), Cyber and Computer Security Laboratory (CYBER), The Laboratory for Computer Communication and Networking (LCCN), Geometric Image Processing Laboratory (GIP), invite you to visit the Spring Project Fair inProject Fair in IoT, Software, Android Apps, AI, Cyber, Computer Security, Networks, Computer Vision and Virtual Reality, including demos and presentations by 60 undergraduate teams who will answer your questions on their research. The event will be held on Tuesday, ...
[Full version]

Least-mean squares (LMS) solvers such as Linear / Ridge / Lasso-Regression, SVD and Elastic-Net not only solve fundamental machine learning problems, but are also the building blo cks in a variety of other methods, such as decision trees and matrix factorizations. We suggest an algorithm that gets a finite set of n d-dimensional real vectors and returns a weighted subset of d + 1 vectors whose sum is exactly the same. The proof in Caratheodory’s ...
[Full version]

We propose a method to automatically match two shape collections with a similar shape space structure, e.g. two characters in similar poses, and compute the inter-maps between the collections. Given the intra-maps in each collection, we extract the corresponding shape difference operators, and use them to construct an embedding of the shape space of each collection. We then align the two shape spaces, and use the knowledge gained from the alignment to compute the inter-maps. ...
[Full version]

The task of synthetic data generation can be, roughly, stated as follows: A learner gets to observe examples, sampled from some unknown distributions, and needs to output "synthetic" examples that "look similar". For example, think of an algorithm that receives examples of music tunes from some genre (e.g. jazz, classical, etc.) and in turn generates new synthetic pieces of music from that genre. This task has won considerable attention lately, due to the introduction of ...
[Full version]

We address the problem of binary code search in stripped executables (with no debug information). The main challenge is establishing binary code similarity even when the binary code has been compiled using different compilers, optimization levels and targeting diverse architectures. Moreover, the source code being compiled might be from another version of the software package or another implementation altogether. Overcoming these challenges, while avoiding false-positives, is invaluable to guiding other more costly tasks in the ...
[Full version]

The study of permutations is motivated by their application in coding for flash memories, and their relevance in different applications of networking technologies and various channels. We study the multidimensional constrained systems of $\mathbb{Z}^d$-permutations with restricted movement. During the talk, we will show a correspondence between these restricted permutations and perfect matchings. We will use the theory of perfect matchings to investigate several two-dimensional cases, for which we compute the exact capacity of the constrained ...
[Full version]

Modern machine learning algorithms have achieved remarkable performance in a myriad of applications, and are increasingly used to make impactful decisions in the hiring process, criminal sentencing, healthcare diagnostics and even to make new scientific discoveries. The use of data-driven algorithms in high-stakes applications is exciting yet alarming: these methods are extremely complex, often brittle, notoriously hard to analyze and interpret. Naturally, concerns have raised about the reliability, fairness, and reproducibility of the output of ...
[Full version]

Inmodern applications of graph algorithms, where the graphs of interest arelarge and dynamic, it is unrealistic to assume that an input representationcontains the full information of a graph being studied. For example, consider asocial network, where a vertex corresponds to a user and an edge corresponds toa friendship relation. It is reasonable to assume that users do not alwaysupdate new friendships on the social network, and that sometimes theydo not fully disclose their friendship relations ...
[Full version]

We are living in the era of massive data. Some estimate that more than 90% of the worlds data was generated in recent years. This large amount of data comes with technological challenges such as: distributed storage systems, synchronization issues, density constraints, power constraints and more. The talk will be partitioned into two, independent parts in which some solutions to those challenges will be discussed. Specifically, we will touch upon dynamical distributed storage systems, and ...
[Full version]

We study the online problems related to controlling a linear time-invariant systems with noisy dynamics where either: (1) onlinevadversarially chosen quadratic losses and known dynamics, or (2) unknown dynamics but known quadratic losses. We present an efficient online learning algorithms for both setting with vanishing regret. Based on joint works with Alon Cohen, Avinatan Hassidim, Tomer Koren, Nevena Lazic, and Kunal Talwar Bio: Prof. Yishay Mansour got his PhD from MIT in 1990, following it ...
[Full version]

Formal verification of infinite-state systems, and distributed systems in particular, is a long standing research goal. I will describe a series of works that develop a methodology for verifying distributed algorithms and systems using decidable logics, employing decomposition, abstraction, and user interaction. This methodology is implemented in an open-source tool, and has resulted in the first mechanized proofs of several important distributed protocols. I will also describe a novel approach to the problem of invariant ...
[Full version]

This talk presents new computational and statistical barriers in machine learning, along with the algorithmic developments that they inspire. The computational barriers arise in nonconvex optimization: we prove lower bounds on the (oracle) complexity of finding stationary points using (stochastic) gradient methods, showing that gradient descent is unimprovable for a natural class of problems. We bypass this barrier by designing an algorithm that outperforms gradient descent for a large subclass of problems with high-order smoothness. ...
[Full version]

We propose a method for tracing the flowlines of a discrete tangent vector field on a triangle mesh. Our method makes use of functional vector fields, namely modeling of vector fields as operators acting on functions. In particular, we use the recently proposed adjoint representations of directional derivatives. Our method is characterized by making use of mostly global solutions, as opposed to iterative local algorithms that trace the vector triangle-by-triangle. We compare our approach to ...
[Full version]

We propose a method for tracing the flowlines of a discrete tangent vector field on a triangle mesh. Our method makes use of functional vector fields, namely modeling of vector fields as operators acting on functions. In particular, we use the recently proposed adjoint representations of directional derivatives. Our method is characterized by making use of mostly global solutions, as opposed to iterative local algorithms that trace the vector triangle-by-triangle. We compare our approach to ...
[Full version]

Compilers are designed today to use fixedAbstract -cost models that are based on heuristics to make different code optimizations. However, these models are unable to capture the data dependency, the computation graph, or the organization of instructions. Therefore, with these models compilers achieve suboptimal performance. In his talk, Ameer will show how to use deep reinforcement learning to overcome such hard compiler challenges. This includes the NP-Hard compiler phase ordering challenge (AutoPhase, FCCM2019, SysML2020) and ...
[Full version]

We develop differentially private mechanisms that achieve nearly instance-optimal losses, achieving lower loss than all appropriately unbiased mechanisms for any possible instance. We show that our mechanisms, with a modest increase in sample size (logarithmic or constant), are instance-optimal for a large family of functions. In contrast to existing mechanisms, which use the global or local sensitivity of the function being estimated, and so are necessarily instance suboptimal, the key to our construction is to ...
[Full version]

One of the unique aspects of quantum mechanics is the no-cloning theorem. This theorem has direct applications in quantum crpytography: quantum money is very similar to the cash we use, but in addition, it is physically impossible to forge. In this talk, I will discuss quantum money, and some of its extensions (quantum coins, semi quantum money, tokens for digital signatures, quantum copy protection, and quantum lightning). I will also descrube a way to use ...
[Full version]

Dynamical systems are everywhere, from the flow of particles in the air to the evolution of people's personalities. Recently, the big data revolution has made increasing amounts of observations of such systems readily available. Unfortunately, in many practical scenarios, the governing equations of the dynamics are unknown. For instance, what would be the mathematical model for periodic questionnaire data that describes personalities? Despite the lack of existing models, scientists and engineers are required to address ...
[Full version]

This talk presents a novel end-to-end approach to program repair based on sequence-to-sequence learning. We devise, implement, and evaluate a system, called SequenceR, for fixing bugs based on sequence-to-sequence learning on source code. This approach uses the copy mechanism to overcome the unlimited vocabulary problem that occurs with big code. Our system is data-driven; we train it on 35,578 samples, carefully curated from commits to open-source repositories. We evaluate it on 4,711 independent real bug ...
[Full version]

We consider the problem of training and evaluating generative models with a Generative Adversarial Network (GAN). Although GANs can accurately model complex distributions, they are known to be difficult to train due to instabilities caused by a difficult minimax optimization problem. In this work, we view the problem of training GANs as finding a mixed strategy in a zero-sum game. This leads to a new training method that builds upon ideas from game theory and ...
[Full version]

Recent developments in Natural Language Processing (NLP) allow models to leverage large, unprecedented amounts of raw text, culminating in impressive performance gains in many of the field's long-standing challenges, such as machine translation, question answering, or information retrieval. In this talk, I will show that despite these advances, state-of-the-art NLP models often fail to capture crucial aspects of text understanding. Instead, they excel by finding spurious patterns in the data, which lead to biased and ...
[Full version]

The network line rate is constantly on the rise to support the exploding amounts of data. This means that we have less time to process individual packets, despite a growing demand for better network telemetry. Moreover, CPU speeds are not rising at the same rate as we near the end of Moore’s law, making it harder to rely on software computations. Programmable hardware switches are an emerging technology that enables flexible packet processing while optimizing ...
[Full version]