Speaker: Oleg Kolosov

M.Sc. student under the supervision of Dr. Itzhak Tamo and Dr. Gala Yadgar

Wednesday, May 9^th, 2018 at 15:00

Room 011, Kitot Bldg., Faculty of Engineering

On Fault Tolerance, Locality, and Optimality in Locally Repairable Codes

Abstract

            Erasure codes are used in large-scale storage systems to allow recovery of data from a failed node. A recently developed class of erasure codes, termed locally repairable codes (LRCs), offers tradeoffs between storage overhead and repair cost. LRCs facilitate more efficient recovery scenarios by storing additional parity blocks in the system, but these additional blocks may eventually increase the number of blocks that must be reconstructed. Existing codes differ in their use of the additional parity blocks, but also in their locality semantics and in the parameters for which they are defined. As a result, existing theoretical models cannot be used to directly compare different LRCs to determine which code will offer the best recovery performance, and at what cost.

In this study, we perform the first systematic comparison of existing LRC approaches. We analyze Xorbas, Azure's LRCs, and the recently proposed Optimal-LRCs in light of two new metrics: the average degraded read cost, and the normalized repair cost.

We show the tradeoff between these costs and the code's fault tolerance, and that different approaches offer different choices in this tradeoff. Our experimental evaluation on a Ceph cluster deployed on Amazon EC2 further demonstrates the different effects of realistic network and storage bottlenecks on the benefit from each examined LRC approach. Despite these differences, the normalized repair cost metric can reliably identify the LRC approach that would achieve the lowest repair cost in each setup.

(The talk will be given in English)

Speaker:     Dr. Ran Gilad-Bachrach
                              Microsoft 

Monday, May 7^th, 2018
15:00 - 16:00

Room 011, Kitot Bldg., Faculty of Engineering

Private Collaborative Learning of Intelligible Models on Vertically Split Data

Abstract

A common privacy scenario in machine learning is one in which possibly several different parties hold fragments of a training dataset, and want to train a model together without revealing any unnecessary information to each other about their data. We present methods for training intelligible models, namely Generalized Additive Models (GAMs), on datasets that are vertically split across multiple parties. Our protocol uses fully homomorphic encryption to guarantee data privacy, and is proved to be secure in the semi-honest security model under mild non-collusion assumptions. We prove the privacy properties of our solution as well as demonstrate its practicality in empirical evaluation.

Bio

Ran Gilad-Bachrach is a researcher at Microsoft Research. He studies multiple aspects of machine learning, currently he focuses on Private-AI and Precision-Psychology. Private-AI is the study of privacy preserving and collaboration enabling machine learning. Precision-Psychology is the study of the applications of data science to improving outcomes in helping people going through behavioral changes or fighting disorders. Ran earned his Ph.D. from the Hebrew University of Jerusalem and worked at Intel Research and Bing before joining Microsoft Research.

School of Mechanical Engineering Seminar
Monday, May 7, 2018 at 14:00
Wolfson Building of Mechanical Engineering, Room 206

Solitary waves on ferrofluids jet

Prof. Emilian Parau

 Univ. of East Anglia UK

'Travelling axisymmetric waves on the surface of a cylindrical ferrofluid jet are investigated. An azimuthal magnetic field is generated by an electric current flowing along a stationary metal rod which is mounted along the axis of the moving jet. Solitary waves are computed and their time evolution is discussed. Comparisons with weakly nonlinear theories such as KdV, and experiments are presented.'