פרופ' שמואל [שי] אבידן

(The talk will be given in English)

Speaker:     Ofir Weisse
                   PhD student at the University of Michigan, USA

Wednesday, January 3^rd, 2018
15:00 - 16:00

Room 011, Kitot Bldg., Faculty of Engineering

Regaining Lost Cycles with HotCalls: A Fast Interface for SGX Secure Enclaves

Abstract

Intel's SGX secure execution technology allows the running of computations on secret data using untrusted cloud servers. While recent work showed how to port applications and large scale computations to run under SGX, the performance implications of using the technology remains an open question. We present the first comprehensive quantitative study to evaluate the performance of SGX. We show that straightforward use of SGX library primitives for calling functions add between 8,200 - 17,000 cycles overhead, compared to 150 cycles of a typical system call. We quantify the performance impact of these library calls and show that in applications with high system calls frequency, such as memcached, openVPN, and lighttpd, which all have high bandwidth network requirements, the performance degradation may be as high as 79%. We investigate the sources of this performance degradation by leveraging a new set of micro-benchmarks for SGX-specific operations such as entry-calls and out-calls, and encrypted memory I/O accesses. We leverage the insights we gain from these analyses to design a new SGX interface framework, HotCalls: HotCalls provide a 13-27x speedup over the default interface. It can easily be integrated into existing code, making it a practical solution. Compared to a baseline SGX implementation of memcached, openVPN, and lighttpd - we show that using the new interface boosts the throughput by 2.6-3.7x, and reduce application response time by 62-74%.

(The talk will be given in English)

Speaker:     Dr. Dan Benjamini
                   Section on Quantitative Imaging and Tissue Sciences

National Institutes of Health, MD, USA

Monday, December 4^th, 2017
15:00 - 16:00

Room 011, Kitot Bldg., Faculty of Engineering

Breaking the spatial resolution limit of magnetic resonance imaging

Abstract

Microscopic scale changes in brain tissue take place during normal and abnormal brain devel-opment, and following mild to severe traumatic brain injury (TBI). In addition, brain tissue is heterogeneous in nature; an imaging volume will contain an ensemble of different cells and ex-tracellular matrix components. At the microscopic level, these components would vastly differ in their molecular composition, as well as in their microstructure. As a result, the water that resides within the tissue will also interact with different chemical and physical microenviron-ments. By probing these water interactions using magnetic resonance imaging (MRI) one can, theoretically, overcome the limited spatial resolution (∼ 1mm3) and noninvasively detect and distinguish between the different microenvironments within a voxel. However, to date, com-putational instabilities and vast data requirements, leading to clinically infeasible scan times, have mostly relegated these type of measurements to nuclear magnetic resonance (NMR) ap-plications, and prevented them from being widely and successfully used in conjunction with imaging.

I have recently developed a novel experimental design, data acquisition, and signal processing framework, termed MADCO, which allows to combine various spectroscopic NMR experiments with imaging in reasonable scanning times and with excellent prospects for clinical applications. Using these tools one can measure, map, and render 3D images of the dynamics of water protons in a model-free and direct manner, making no assumptions about the underlying microstructure of the tissue. Here I will present two new, and complemen-tary, MADCO-based MR imaging framework: (1) Magnetic resonance microdynamic imaging (MRMI), which permits the simultaneous noninvasive and model-free quantification of the volume fraction of multiple subcellular, cellular, and interstitial tissue components within a voxel (e.g., axons, neurons, glia, interstitial space, and myelin in brain tissue). (2) Acceler-ated diffusion exchange spectroscopic imaging (ADEXSI), which is able to noninvasively map dynamic migration of water from one microenvironment to another (e.g., molecular exchange rate of water between intra- and extracellular spaces).

Apart from its known biological and clinical applications, MR is routinely used by chemists and biochemists to quantitatively describe materials with sub-molecular resolution; it is used by physicists to quantify flow and velocity in microscopic capillaries to measure mass transport; and it is used by geologists as logging tools for in situ oil, gas, and water underground reservoir evaluation. Integrating, migrating, and translating these approaches and methods from a broad range of scientific fields is what I believe can help us to better understand such a complex system as the human brain.

(The talk will be given in English)

Speaker:  Dr. Roy Lederman
                   Applied & Computational Mathematics, Princeton University

Sunday, December 3^rd, 2017
15:00 - 16:00

Room 011, Kitot Bldg., Faculty of Engineering

Inverse Problems and Unsupervised Learning with applications to Cryo-Electron Microscopy (cryo-EM)

Abstract

Cryo-EM is an imaging technology that is revolutionizing structural biology; the Nobel Prize in Chemistry 2017 was recently awarded to Jacques Dubochet, Joachim Frank and Richard Henderson “for developing cryo-electron microscopy for the high-resolution structure determination of biomolecules in solution".

Cryo-electron microscopes produce a large number of very noisy two-dimensional projection images of individual frozen molecules. Unlike related methods, such as computed tomography (CT), the viewing direction of each image is unknown. The unknown directions, together with extreme levels of noise and additional technical factors, make the determination of the structure of molecules challenging.

While other methods for structure determination, such as x-ray crystallography and nuclear magnetic resonance (NMR), measure ensembles of molecules together, cryo-EM produces measurements of individual molecules. Therefore, cryo-EM could potentially be used to study mixtures of different conformations of molecules. Indeed, current algorithms have been very successful at analyzing homogeneous samples, and can recover some distinct conformations mixed in solutions, but, the determination of multiple conformations, and in particular, continuums of similar conformations (continuous heterogeneity), remains one of the open problems in cryo-EM.

I will discuss a one-dimensional discrete model problem, Heterogeneous Multireference Alignment, which captures many of the properties of the cryo-EM problem. I will then discuss different components which we are introducing in order to address the problem of continuous heterogeneity in cryo-EM: 1. “hyper-molecules,” the first mathematical formulation of truly continuously heterogeneous molecules, 2. The optimal representation of objects that are highly concentrated in both the spatial domain and the frequency domain using high-dimensional Prolate spheroidal functions, and 3. Bayesian algorithms for inverse problems with an unsupervised-learning component for recovering such hyper-molecules in cryo-EM.

Short Bio

Roy Lederman is a postdoc at the Program in Applied and Computational Mathematics at Princeton University, working with Amit Singer. Before joining Princeton, he was a postdoc at Yale University, where he had completed his PhD in applied mathematics, working with Vladimir Rokhlin and Ronald Coifman. Roy holds a BSc in Electrical Engineering and a BSc in Physics from Tel Aviv University.