Privacy and Fault-Tolerance in Distributed Optimization. Nitin Vaidya University of Illinois at Urbana-Champaign
|
|
- Anis Kennedy
- 5 years ago
- Views:
Transcription
1 Privacy and Fault-Tolerance in Distributed Optimization Nitin Vaidya University of Illinois at Urbana-Champaign
2 Acknowledgements Shripad Gade Lili Su
3 argmin x2x SX i=1 i f i (x)
4 Applications g f i (x) = cost for robot i to go to location x f 1 (x) x g Minimize total cost of rendezvous x 1 argmin x2x SX i=1 i f i (x) f 2 (x) x 2
5 Applications f 1 (x) f 2 (x) Learning Minimize cost Σ f i (x) i f 3 (x) f 4 (x) 5
6 Outline argmin x2x SX i=1 i f i (x) f & f " f & f " f & f " f % f # f $ f % f # f $ f % f # f $ Distributed Optimization Privacy Fault-tolerance
7 Distributed Optimization Server f & f " f & f # f " f % f # f $ 7
8 Client-Server Architecture Server f 1 (x) f 2 (x) f & f # f " f 3 (x) 8 f 4 (x)
9 Client-Server Architecture g Server maintains estimate x ( g Client i knows f ) (x) x ( Server f & f # f "
10 Client-Server Architecture g Server maintains estimate x ( g Client i knows f ) (x) x ( Server In iteration k+1 f ) (x ( ) g Client i idownload x ( from server iupload gradient f ) (x ( ) f & f # f "
11 Client-Server Architecture g Server maintains estimate x ( g Client i knows f ) (x) Server In iteration k+1 f ) (x ( ) g Client i idownload x ( from server iupload gradient f ) (x ( ) f & f # f " g Server x (-& x ( α ( 2 f ) x ( )
12 Variations g Stochastic g Asynchronous g 12
13 Peer-to-Peer Architecture f 1 (x) f 2 (x) f & f " f % f # f $ f 3 (x) f 4 (x)
14 Peer-to-Peer Architecture g Each agent maintains local estimate x g Consensus step with neighbors g Apply own gradient to own estimate x (-& x ( α ( f ) x ( f & f " f % f # f $
15 Outline argmin x2x SX i=1 i f i (x) f & f " f & f " f & f " f % f # f $ f % f # f $ f % f # f $ Distributed Optimization Privacy Fault-tolerance
16 Server f ) (x ( ) f & f # f "
17 Server f ) (x ( ) f & f # f " Server observes gradients è privacy compromised
18 Server f ) (x ( ) f & f # f " Server observes gradients è privacy compromised Achieve privacy and yet collaboratively optimize
19 Related Work g Cryptographic methods (homomorphic encryption) g Function transformation g Differential privacy 19
20 Differential Privacy Server f ) x ( + ε k f & f # f " 20
21 Differential Privacy Server f ) x ( + ε k f & f # f " Trade-off privacy with accuracy 21
22 Proposed Approach g Motivated by secret sharing g Exploit diversity Multiple servers / neighbors 22
23 Proposed Approach Server 1 Server 2 f & f # f " Privacy if subset of servers adversarial 23
24 Proposed Approach f & f " f % f # f $ Privacy if subset of neighbors adversarial 24
25 Proposed Approach g Structured noise that cancels over servers/neighbors 25
26 Intuition x 1 x 2 Server 1 Server 2 f & f # f " 26
27 Intuition x 1 x 2 Server 1 Server 2 Each client simulates multiple clients f && f &# f #& f ## f "& f "# 27
28 Intuition x 1 x 2 Server 1 Server 2 f && f &# f #& f ## f "& f "# f && (x) + f&# x = f & x f )8 (x) not necessarily convex 28
29 Algorithm g Each server maintains an estimate In each iteration g Client i idownload estimates from corresponding server iupload gradient of f ) g Each server updates estimate using received gradients
30 Algorithm g Each server maintains an estimate In each iteration g Client i idownload estimates from corresponding server iupload gradient of f ) g Each server updates estimate using received gradients g Servers periodically exchange estimates to perform a consensus step
31 Claim g Under suitable assumptions, servers eventually reach consensus in argmin x2x SX i=1 i f i (x) 31
32 Privacy f && + f #& +f "& f #& + f ## +f "# Server 1 Server 2 f && f &# f #& f ## f "& f "# 32
33 Privacy f && + f #& +f "& f #& + f ## +f "# Server 1 Server 2 f && f &# f #& f ## f "& f "# g Server 1 may learn f &&, f #&, f "&, f #& + f ## +f "# g Not sufficient to learn f ) 33
34 f && (x) + f&# x = f& x g Function splitting not necessarily practical g Structured randomization as an alternative 34
35 Structured Randomization g Multiplicative or additive noise in gradients g Noise cancels over servers 35
36 Multiplicative Noise x 1 x 2 Server 1 Server 2 f & f # f " 36
37 Multiplicative Noise x 1 x 2 Server 1 Server 2 f & f # f " 37
38 Multiplicative Noise x 1 x 2 Server 1 Server 2 α f & (x 1 ) β f & (x 2 ) f & f # f " α+β=1 38
39 Multiplicative Noise x 1 x 2 Server 1 Server 2 α f & (x 1 ) β f & (x 2 ) f & f # f " α+β=1 Suffices for this invariant to hold over a larger number of iterations
40 Multiplicative Noise x 1 x 2 Server 1 Server 2 α f & (x 1 ) β f & (x 2 ) f & f # f " α+β=1 Noise from client i to server j not zero-mean
41 Claim g Under suitable assumptions, servers eventually reach consensus in argmin x2x SX i=1 i f i (x) 41
42 Peer-to-Peer Architecture f & f " f % f # f $
43 Reminder g Each agent maintains local estimate x g Consensus step with neighbors g Apply own gradient to own estimate x (-& x ( α ( f ) x ( f & f " f % f # f $
44 Proposed Approach g Each agent shares noisy estimate with neighbors Scheme 1 Noise cancels over neighbors Scheme 2 Noise cancels network-wide f & f " f % f # f $
45 Proposed Approach g Each agent shares noisy estimate with neighbors Scheme 1 Noise cancels over neighbors Scheme 2 Noise cancels network-wide x + ε 1 ε 1 + ε 2 = 0 (over iterations) f & f " f % f # f $ x + ε 2
46 Peer-to-Peer Architecture g Poster today Shripad Gade
47 Outline argmin x2x SX i=1 i f i (x) f & f " f & f " f & f " f % f # f $ f % f # f $ f % f # f $ Distributed Optimization Privacy Fault-tolerance
48 Fault-Tolerance g Some agents may be faulty g Need to produce correct output despite the faults 48
49 Byzantine Fault Model g No constraint on misbehavior of a faulty agent g May send bogus messages g Faulty agents can collude 49
50 Peer-to-Peer Architecture g f i (x) = cost for robot i to go to location x f 1 (x) x g Faulty agent may choose arbitrary cost function x 1 f 2 (x) x 2
51 Peer-to-Peer Architecture f & f " f % f # f $ 51
52 Client-Server Architecture f ) (x ( ) Server f & f # f "
53 Fault-Tolerant Optimization g The original problem is not meaningful argmin x2x SX i=1 i f i (x) 53
54 Fault-Tolerant Optimization g The original problem is not meaningful argmin x2x SX i=1 i f i (x) g Optimize cost over only non-faulty agents argmin x2x SX f i (x) i=1 i good
55 Fault-Tolerant Optimization g The original problem is not meaningful argmin x2x SX i=1 i f i (x) g Optimize cost over only non-faulty agents Impossible! argmin x2x SX f i (x) i=1 i good
56 Fault-Tolerant Optimization g Optimize weighted cost over only non-faulty agents argmin x2x SX i=1 i good f i (x) α i g With α i as close to 1/ good as possible
57 Fault-Tolerant Optimization g Optimize weighted cost over only non-faulty agents argmin x2x SX i=1 i good f i (x) α i With t Byzantine faulty agents: t weights may be 0
58 Fault-Tolerant Optimization g Optimize weighted cost over only non-faulty agents argmin x2x SX i=1 i good f i (x) α i t Byzantine agents, n total agents At least n-2t weights guaranteed to be > 1/2(n-t)
59 Centralized Algorithm g Of the n agents, any t may be faulty g How to filter cost functions of faulty agents? X
60 Centralized Algorithm: Scalar argument x Define a virtual function G(x) whose gradient is obtained as follows 60
61 Centralized Algorithm: Scalar argument x Define a virtual function G(x) whose gradient is obtained as follows At a given x g Sort the gradients of the n local cost functions 61
62 Centralized Algorithm: Scalar argument x Define a virtual function G(x) whose gradient is obtained as follows At a given x g Sort the gradients of the n local cost functions g Discard smallest t and largest t gradients 62
63 Centralized Algorithm: Scalar argument x Define a virtual function G(x) whose gradient is obtained as follows At a given x g Sort the gradients of the n local cost functions g Discard smallest t and largest t gradients g Mean of remaining gradients = Gradient of G at x 63
64 Centralized Algorithm: Scalar argument x Define a virtual function G(x) whose gradient is obtained as follows At a given x g Sort the gradients of the n local cost functions g Discard smallest t and largest t gradients g Mean of remaining gradients = Gradient of G at x Virtual function G(x) is convex
65 Centralized Algorithm: Scalar argument x Define a virtual function G(x) whose gradient is obtained as follows At a given x g Sort the gradients of the n local cost functions g Discard smallest t and largest t gradients g Mean of remaining gradients = Gradient of G at x Virtual function G(x) is convex à Can optimize easily
66 Peer-to-Peer Fault-Tolerant Optimization g Gradient filtering similar to centralized algorithm require rich enough connectivity correlation between functions helps g Vector case harder redundancy between functions helps 66
67 Summary argmin x2x SX i=1 i f i (x) f & f " f & f " f & f " f % f # f $ f % f # f $ f % f # f $ Distributed Optimization Privacy Fault-tolerance
68 Thanks! disc.ece.illinois.edu
69 69
70 70
71 Distributed Peer-to-Peer Optimization g Each agent maintains local estimate x In each iteration g Compute weighted average with neighbors estimates f & f " f % f # f $
72 Distributed Peer-to-Peer Optimization g Each agent maintains local estimate x In each iteration g Compute weighted average with neighbors estimates g Apply own gradient to own estimate x (-& x ( α ( f ) x ( f & f " f % f # f $
73 Distributed Peer-to-Peer Optimization g Each agent maintains local estimate x In each iteration g Compute weighted average with neighbors estimates g Apply own gradient to own estimate g Local estimates converge to f & f " x (-& x ( α ( f ) x ( SX argmin f i (x) x2x i i=1 f % f # f $
74 RSS Locally Balanced Perturbations g Add to zero (locally per node) g Bounded ( Δ) Algorithm g Node j selects A,B such that d@ A,B B = 0 and d ( 8,) Δ g Share A,B = x@ A + d@ A,B with node i g Consensus and (Stochastic) Gradient Descent 74
75 RSS Network Balanced Perturbations g Add to zero (over network) g Bounded ( Δ) Algorithm g Node j computes perturbation A - sends s A,B to i - add received s B,A and subtract sent s A,B A = rcvd sent A A A g Obfuscate state = + d@ shared with neighbors g Consensus and (Stochastic) Gradient Descent 75
76 Convergence Let xp A Q = Q A / Q and = 1/ k f xp A Q f x O log T T + O Δ# log T T g Asymptotic convergence of iterates to optimum g Privacy-Convergence Trade-off g Stochastic gradient updates work too 76
77 Function Sharing g Let f B (x) be bounded degree polynomials Algorithm g Node j shares s A,B x with node i g Node j obfuscates using p A x = s B,A x s A,B (x) g Use f^a x = f A x + p A (x) and use distributed gradient descent 77
78 Function Sharing - Convergence g Function Sharing iterates converge to correct optimum ( f^b x = f(x)) g Privacy: If vertex connectivity of graph f then no group of f nodes can estimate true functions f ) (or any good subset) g p A (x) is also similar to f A (x) then it can hide f B x well 78
Deterministic Consensus Algorithm with Linear Per-Bit Complexity
Deterministic Consensus Algorithm with Linear Per-Bit Complexity Guanfeng Liang and Nitin Vaidya Department of Electrical and Computer Engineering, and Coordinated Science Laboratory University of Illinois
More informationResilient Asymptotic Consensus in Robust Networks
Resilient Asymptotic Consensus in Robust Networks Heath J. LeBlanc, Member, IEEE, Haotian Zhang, Student Member, IEEE, Xenofon Koutsoukos, Senior Member, IEEE, Shreyas Sundaram, Member, IEEE Abstract This
More informationThe Complexity of a Reliable Distributed System
The Complexity of a Reliable Distributed System Rachid Guerraoui EPFL Alexandre Maurer EPFL Abstract Studying the complexity of distributed algorithms typically boils down to evaluating how the number
More informationConsensus-Based Distributed Optimization with Malicious Nodes
Consensus-Based Distributed Optimization with Malicious Nodes Shreyas Sundaram Bahman Gharesifard Abstract We investigate the vulnerabilities of consensusbased distributed optimization protocols to nodes
More informationMulti-Robotic Systems
CHAPTER 9 Multi-Robotic Systems The topic of multi-robotic systems is quite popular now. It is believed that such systems can have the following benefits: Improved performance ( winning by numbers ) Distributed
More informationFault-Tolerant Consensus
Fault-Tolerant Consensus CS556 - Panagiota Fatourou 1 Assumptions Consensus Denote by f the maximum number of processes that may fail. We call the system f-resilient Description of the Problem Each process
More informationResilient Distributed Optimization Algorithm against Adversary Attacks
207 3th IEEE International Conference on Control & Automation (ICCA) July 3-6, 207. Ohrid, Macedonia Resilient Distributed Optimization Algorithm against Adversary Attacks Chengcheng Zhao, Jianping He
More informationIntroduction to Modern Cryptography Lecture 11
Introduction to Modern Cryptography Lecture 11 January 10, 2017 Instructor: Benny Chor Teaching Assistant: Orit Moskovich School of Computer Science Tel-Aviv University Fall Semester, 2016 17 Tuesday 12:00
More informationDo we have a quorum?
Do we have a quorum? Quorum Systems Given a set U of servers, U = n: A quorum system is a set Q 2 U such that Q 1, Q 2 Q : Q 1 Q 2 Each Q in Q is a quorum How quorum systems work: A read/write shared register
More informationScientific Computing: An Introductory Survey
Scientific Computing: An Introductory Survey Chapter 5 Nonlinear Equations Prof. Michael T. Heath Department of Computer Science University of Illinois at Urbana-Champaign Copyright c 2002. Reproduction
More informationAGREEMENT PROBLEMS (1) Agreement problems arise in many practical applications:
AGREEMENT PROBLEMS (1) AGREEMENT PROBLEMS Agreement problems arise in many practical applications: agreement on whether to commit or abort the results of a distributed atomic action (e.g. database transaction)
More information12. LOCAL SEARCH. gradient descent Metropolis algorithm Hopfield neural networks maximum cut Nash equilibria
12. LOCAL SEARCH gradient descent Metropolis algorithm Hopfield neural networks maximum cut Nash equilibria Lecture slides by Kevin Wayne Copyright 2005 Pearson-Addison Wesley h ttp://www.cs.princeton.edu/~wayne/kleinberg-tardos
More informationByzantine Vector Consensus in Complete Graphs
Byzantine Vector Consensus in Complete Graphs Nitin H. Vaidya University of Illinois at Urbana-Champaign nhv@illinois.edu Phone: +1 217-265-5414 Vijay K. Garg University of Texas at Austin garg@ece.utexas.edu
More informationCoordination. Failures and Consensus. Consensus. Consensus. Overview. Properties for Correct Consensus. Variant I: Consensus (C) P 1. v 1.
Coordination Failures and Consensus If the solution to availability and scalability is to decentralize and replicate functions and data, how do we coordinate the nodes? data consistency update propagation
More informationProvable Security for Program Obfuscation
for Program Obfuscation Black-box Mathematics & Mechanics Faculty Saint Petersburg State University Spring 2005 SETLab Outline 1 Black-box Outline 1 2 Black-box Outline Black-box 1 2 3 Black-box Perfect
More informationDecoupling Coupled Constraints Through Utility Design
1 Decoupling Coupled Constraints Through Utility Design Na Li and Jason R. Marden Abstract The central goal in multiagent systems is to design local control laws for the individual agents to ensure that
More informationFinite-Time Resilient Formation Control with Bounded Inputs
Finite-Time Resilient Formation Control with Bounded Inputs James Usevitch, Kunal Garg, and Dimitra Panagou Abstract In this paper we consider the problem of a multiagent system achieving a formation in
More informationSection 6 Fault-Tolerant Consensus
Section 6 Fault-Tolerant Consensus CS586 - Panagiota Fatourou 1 Description of the Problem Consensus Each process starts with an individual input from a particular value set V. Processes may fail by crashing.
More informationDistributed Systems Byzantine Agreement
Distributed Systems Byzantine Agreement He Sun School of Informatics University of Edinburgh Outline Finish EIG algorithm for Byzantine agreement. Number-of-processors lower bound for Byzantine agreement.
More informationBroadcast and Verifiable Secret Sharing: New Security Models and Round-Optimal Constructions
Broadcast and Verifiable Secret Sharing: New Security Models and Round-Optimal Constructions Dissertation submitted to the Faculty of the Graduate School of the University of Maryland, College Park in
More informationDistributed Optimization over Networks Gossip-Based Algorithms
Distributed Optimization over Networks Gossip-Based Algorithms Angelia Nedić angelia@illinois.edu ISE Department and Coordinated Science Laboratory University of Illinois at Urbana-Champaign Outline Random
More informationDegradable Agreement in the Presence of. Byzantine Faults. Nitin H. Vaidya. Technical Report #
Degradable Agreement in the Presence of Byzantine Faults Nitin H. Vaidya Technical Report # 92-020 Abstract Consider a system consisting of a sender that wants to send a value to certain receivers. Byzantine
More informationBenny Pinkas. Winter School on Secure Computation and Efficiency Bar-Ilan University, Israel 30/1/2011-1/2/2011
Winter School on Bar-Ilan University, Israel 30/1/2011-1/2/2011 Bar-Ilan University Benny Pinkas Bar-Ilan University 1 What is N? Bar-Ilan University 2 Completeness theorems for non-cryptographic fault-tolerant
More informationr-robustness and (r, s)-robustness of Circulant Graphs
r-robustness and (r, s)-robustness of Circulant Graphs James Usevitch and Dimitra Panagou Abstract There has been recent growing interest in graph theoretical properties known as r- and (r, s)-robustness.
More information10. Smooth Varieties. 82 Andreas Gathmann
82 Andreas Gathmann 10. Smooth Varieties Let a be a point on a variety X. In the last chapter we have introduced the tangent cone C a X as a way to study X locally around a (see Construction 9.20). It
More informationNetwork Newton. Aryan Mokhtari, Qing Ling and Alejandro Ribeiro. University of Pennsylvania, University of Science and Technology (China)
Network Newton Aryan Mokhtari, Qing Ling and Alejandro Ribeiro University of Pennsylvania, University of Science and Technology (China) aryanm@seas.upenn.edu, qingling@mail.ustc.edu.cn, aribeiro@seas.upenn.edu
More informationFrom Secure MPC to Efficient Zero-Knowledge
From Secure MPC to Efficient Zero-Knowledge David Wu March, 2017 The Complexity Class NP NP the class of problems that are efficiently verifiable a language L is in NP if there exists a polynomial-time
More informationNon-Convex Optimization. CS6787 Lecture 7 Fall 2017
Non-Convex Optimization CS6787 Lecture 7 Fall 2017 First some words about grading I sent out a bunch of grades on the course management system Everyone should have all their grades in Not including paper
More informationNeed for Deep Networks Perceptron. Can only model linear functions. Kernel Machines. Non-linearity provided by kernels
Need for Deep Networks Perceptron Can only model linear functions Kernel Machines Non-linearity provided by kernels Need to design appropriate kernels (possibly selecting from a set, i.e. kernel learning)
More informationStatistical Machine Learning
Statistical Machine Learning Christoph Lampert Spring Semester 2015/2016 // Lecture 12 1 / 36 Unsupervised Learning Dimensionality Reduction 2 / 36 Dimensionality Reduction Given: data X = {x 1,..., x
More informationThe Weighted Byzantine Agreement Problem
The Weighted Byzantine Agreement Problem Vijay K. Garg and John Bridgman Department of Electrical and Computer Engineering The University of Texas at Austin Austin, TX 78712-1084, USA garg@ece.utexas.edu,
More informationCommunication-efficient and Differentially-private Distributed SGD
1/36 Communication-efficient and Differentially-private Distributed SGD Ananda Theertha Suresh with Naman Agarwal, Felix X. Yu Sanjiv Kumar, H. Brendan McMahan Google Research November 16, 2018 2/36 Outline
More informationA n = A N = [ N, N] A n = A 1 = [ 1, 1]. n=1
Math 235: Assignment 1 Solutions 1.1: For n N not zero, let A n = [ n, n] (The closed interval in R containing all real numbers x satisfying n x n). It is easy to see that we have the chain of inclusion
More informationAustralian National University WORKSHOP ON SYSTEMS AND CONTROL
Australian National University WORKSHOP ON SYSTEMS AND CONTROL Canberra, AU December 7, 2017 Australian National University WORKSHOP ON SYSTEMS AND CONTROL A Distributed Algorithm for Finding a Common
More informationOn Acceleration with Noise-Corrupted Gradients. + m k 1 (x). By the definition of Bregman divergence:
A Omitted Proofs from Section 3 Proof of Lemma 3 Let m x) = a i On Acceleration with Noise-Corrupted Gradients fxi ), u x i D ψ u, x 0 ) denote the function under the minimum in the lower bound By Proposition
More informationAppendix A.1 Derivation of Nesterov s Accelerated Gradient as a Momentum Method
for all t is su cient to obtain the same theoretical guarantees. This method for choosing the learning rate assumes that f is not noisy, and will result in too-large learning rates if the objective is
More informationGeneralized Consensus and Paxos
Generalized Consensus and Paxos Leslie Lamport 3 March 2004 revised 15 March 2005 corrected 28 April 2005 Microsoft Research Technical Report MSR-TR-2005-33 Abstract Theoretician s Abstract Consensus has
More informationOptimal and Player-Replaceable Consensus with an Honest Majority Silvio Micali and Vinod Vaikuntanathan
Computer Science and Artificial Intelligence Laboratory Technical Report MIT-CSAIL-TR-2017-004 March 31, 2017 Optimal and Player-Replaceable Consensus with an Honest Majority Silvio Micali and Vinod Vaikuntanathan
More informationOutline. Scientific Computing: An Introductory Survey. Nonlinear Equations. Nonlinear Equations. Examples: Nonlinear Equations
Methods for Systems of Methods for Systems of Outline Scientific Computing: An Introductory Survey Chapter 5 1 Prof. Michael T. Heath Department of Computer Science University of Illinois at Urbana-Champaign
More informationConsensus Algorithms for Camera Sensor Networks. Roberto Tron Vision, Dynamics and Learning Lab Johns Hopkins University
Consensus Algorithms for Camera Sensor Networks Roberto Tron Vision, Dynamics and Learning Lab Johns Hopkins University Camera Sensor Networks Motes Small, battery powered Embedded camera Wireless interface
More informationDistributed Randomized Algorithms for the PageRank Computation Hideaki Ishii, Member, IEEE, and Roberto Tempo, Fellow, IEEE
IEEE TRANSACTIONS ON AUTOMATIC CONTROL, VOL. 55, NO. 9, SEPTEMBER 2010 1987 Distributed Randomized Algorithms for the PageRank Computation Hideaki Ishii, Member, IEEE, and Roberto Tempo, Fellow, IEEE Abstract
More informationSecure Multiparty Computation from Graph Colouring
Secure Multiparty Computation from Graph Colouring Ron Steinfeld Monash University July 2012 Ron Steinfeld Secure Multiparty Computation from Graph Colouring July 2012 1/34 Acknowledgements Based on joint
More informationArchitectures and Algorithms for Distributed Generation Control of Inertia-Less AC Microgrids
Architectures and Algorithms for Distributed Generation Control of Inertia-Less AC Microgrids Alejandro D. Domínguez-García Coordinated Science Laboratory Department of Electrical and Computer Engineering
More informationNeed for Deep Networks Perceptron. Can only model linear functions. Kernel Machines. Non-linearity provided by kernels
Need for Deep Networks Perceptron Can only model linear functions Kernel Machines Non-linearity provided by kernels Need to design appropriate kernels (possibly selecting from a set, i.e. kernel learning)
More informationAgreement Protocols. CS60002: Distributed Systems. Pallab Dasgupta Dept. of Computer Sc. & Engg., Indian Institute of Technology Kharagpur
Agreement Protocols CS60002: Distributed Systems Pallab Dasgupta Dept. of Computer Sc. & Engg., Indian Institute of Technology Kharagpur Classification of Faults Based on components that failed Program
More informationAsynchronous Convex Consensus in the Presence of Crash Faults
Asynchronous Convex Consensus in the Presence of Crash Faults Lewis Tseng 1, and Nitin Vaidya 2 1 Department of Computer Science, 2 Department of Electrical and Computer Engineering, University of Illinois
More informationLinear Regression. CSL603 - Fall 2017 Narayanan C Krishnan
Linear Regression CSL603 - Fall 2017 Narayanan C Krishnan ckn@iitrpr.ac.in Outline Univariate regression Multivariate regression Probabilistic view of regression Loss functions Bias-Variance analysis Regularization
More informationLinear Regression. CSL465/603 - Fall 2016 Narayanan C Krishnan
Linear Regression CSL465/603 - Fall 2016 Narayanan C Krishnan ckn@iitrpr.ac.in Outline Univariate regression Multivariate regression Probabilistic view of regression Loss functions Bias-Variance analysis
More informationBELIEF-PROPAGATION FOR WEIGHTED b-matchings ON ARBITRARY GRAPHS AND ITS RELATION TO LINEAR PROGRAMS WITH INTEGER SOLUTIONS
BELIEF-PROPAGATION FOR WEIGHTED b-matchings ON ARBITRARY GRAPHS AND ITS RELATION TO LINEAR PROGRAMS WITH INTEGER SOLUTIONS MOHSEN BAYATI, CHRISTIAN BORGS, JENNIFER CHAYES, AND RICCARDO ZECCHINA Abstract.
More informationOn the complexity of maximizing the minimum Shannon capacity in wireless networks by joint channel assignment and power allocation
On the complexity of maximizing the minimum Shannon capacity in wireless networks by joint channel assignment and power allocation Mikael Fallgren Royal Institute of Technology December, 2009 Abstract
More information12. LOCAL SEARCH. gradient descent Metropolis algorithm Hopfield neural networks maximum cut Nash equilibria
Coping With NP-hardness Q. Suppose I need to solve an NP-hard problem. What should I do? A. Theory says you re unlikely to find poly-time algorithm. Must sacrifice one of three desired features. Solve
More informationMath 473: Practice Problems for Test 1, Fall 2011, SOLUTIONS
Math 473: Practice Problems for Test 1, Fall 011, SOLUTIONS Show your work: 1. (a) Compute the Taylor polynomials P n (x) for f(x) = sin x and x 0 = 0. Solution: Compute f(x) = sin x, f (x) = cos x, f
More informationIterative Approximate Byzantine Consensus in Arbitrary Directed Graphs
Iterative Approximate Byzantine Consensus in Arbitrary Directed Graphs Nitin Vaidya 1,3, Lewis Tseng,3, and Guanfeng Liang 1,3 1 Department of Electrical and Computer Engineering, Department of Computer
More informationAsynchronous Non-Convex Optimization For Separable Problem
Asynchronous Non-Convex Optimization For Separable Problem Sandeep Kumar and Ketan Rajawat Dept. of Electrical Engineering, IIT Kanpur Uttar Pradesh, India Distributed Optimization A general multi-agent
More informationInference in Bayesian Networks
Andrea Passerini passerini@disi.unitn.it Machine Learning Inference in graphical models Description Assume we have evidence e on the state of a subset of variables E in the model (i.e. Bayesian Network)
More informationLecture 1. 1 Introduction. 2 Secret Sharing Schemes (SSS) G Exposure-Resilient Cryptography 17 January 2007
G22.3033-013 Exposure-Resilient Cryptography 17 January 2007 Lecturer: Yevgeniy Dodis Lecture 1 Scribe: Marisa Debowsky 1 Introduction The issue at hand in this course is key exposure: there s a secret
More informationZangwill s Global Convergence Theorem
Zangwill s Global Convergence Theorem A theory of global convergence has been given by Zangwill 1. This theory involves the notion of a set-valued mapping, or point-to-set mapping. Definition 1.1 Given
More informationSecret Sharing CPT, Version 3
Secret Sharing CPT, 2006 Version 3 1 Introduction In all secure systems that use cryptography in practice, keys have to be protected by encryption under other keys when they are stored in a physically
More informationLeast Mean Squares Regression. Machine Learning Fall 2018
Least Mean Squares Regression Machine Learning Fall 2018 1 Where are we? Least Squares Method for regression Examples The LMS objective Gradient descent Incremental/stochastic gradient descent Exercises
More informationEarly stopping: the idea. TRB for benign failures. Early Stopping: The Protocol. Termination
TRB for benign failures Early stopping: the idea Sender in round : :! send m to all Process p in round! k, # k # f+!! :! if delivered m in round k- and p " sender then 2:!! send m to all 3:!! halt 4:!
More informationConstrained Consensus and Optimization in Multi-Agent Networks
Constrained Consensus Optimization in Multi-Agent Networks The MIT Faculty has made this article openly available. Please share how this access benefits you. Your story matters. Citation As Published Publisher
More informationGradient Descent. Sargur Srihari
Gradient Descent Sargur srihari@cedar.buffalo.edu 1 Topics Simple Gradient Descent/Ascent Difficulties with Simple Gradient Descent Line Search Brent s Method Conjugate Gradient Descent Weight vectors
More informationVariance Reduction and Ensemble Methods
Variance Reduction and Ensemble Methods Nicholas Ruozzi University of Texas at Dallas Based on the slides of Vibhav Gogate and David Sontag Last Time PAC learning Bias/variance tradeoff small hypothesis
More informationPrivacy of Numeric Queries Via Simple Value Perturbation. The Laplace Mechanism
Privacy of Numeric Queries Via Simple Value Perturbation The Laplace Mechanism Differential Privacy A Basic Model Let X represent an abstract data universe and D be a multi-set of elements from X. i.e.
More informationCS599: Convex and Combinatorial Optimization Fall 2013 Lecture 24: Introduction to Submodular Functions. Instructor: Shaddin Dughmi
CS599: Convex and Combinatorial Optimization Fall 2013 Lecture 24: Introduction to Submodular Functions Instructor: Shaddin Dughmi Announcements Introduction We saw how matroids form a class of feasible
More informationDifferential Privacy
CS 380S Differential Privacy Vitaly Shmatikov most slides from Adam Smith (Penn State) slide 1 Reading Assignment Dwork. Differential Privacy (invited talk at ICALP 2006). slide 2 Basic Setting DB= x 1
More informationCSC321 Lecture 8: Optimization
CSC321 Lecture 8: Optimization Roger Grosse Roger Grosse CSC321 Lecture 8: Optimization 1 / 26 Overview We ve talked a lot about how to compute gradients. What do we actually do with them? Today s lecture:
More informationComparison of Modern Stochastic Optimization Algorithms
Comparison of Modern Stochastic Optimization Algorithms George Papamakarios December 214 Abstract Gradient-based optimization methods are popular in machine learning applications. In large-scale problems,
More information3E4: Modelling Choice. Introduction to nonlinear programming. Announcements
3E4: Modelling Choice Lecture 7 Introduction to nonlinear programming 1 Announcements Solutions to Lecture 4-6 Homework will be available from http://www.eng.cam.ac.uk/~dr241/3e4 Looking ahead to Lecture
More information2 = = 0 Thus, the number which is largest in magnitude is equal to the number which is smallest in magnitude.
Limits at Infinity Two additional topics of interest with its are its as x ± and its where f(x) ±. Before we can properly discuss the notion of infinite its, we will need to begin with a discussion on
More informationSolutions of Equations in One Variable. Newton s Method
Solutions of Equations in One Variable Newton s Method Numerical Analysis (9th Edition) R L Burden & J D Faires Beamer Presentation Slides prepared by John Carroll Dublin City University c 2011 Brooks/Cole,
More informationa 2 = ab a 2 b 2 = ab b 2 (a + b)(a b) = b(a b) a + b = b
Discrete Structures CS2800 Fall 204 Final Solutions. Briefly and clearly identify the errors in each of the following proofs: (a) Proof that is the largest natural number: Let n be the largest natural
More informationAn Unconditionally Secure Protocol for Multi-Party Set Intersection
An Unconditionally Secure Protocol for Multi-Party Set Intersection Ronghua Li 1,2 and Chuankun Wu 1 1 State Key Laboratory of Information Security, Institute of Software, Chinese Academy of Sciences,
More informationCOMS W4995 Introduction to Cryptography October 12, Lecture 12: RSA, and a summary of One Way Function Candidates.
COMS W4995 Introduction to Cryptography October 12, 2005 Lecture 12: RSA, and a summary of One Way Function Candidates. Lecturer: Tal Malkin Scribes: Justin Cranshaw and Mike Verbalis 1 Introduction In
More informationDual Decomposition for Inference
Dual Decomposition for Inference Yunshu Liu ASPITRG Research Group 2014-05-06 References: [1]. D. Sontag, A. Globerson and T. Jaakkola, Introduction to Dual Decomposition for Inference, Optimization for
More informationSimple Techniques for Improving SGD. CS6787 Lecture 2 Fall 2017
Simple Techniques for Improving SGD CS6787 Lecture 2 Fall 2017 Step Sizes and Convergence Where we left off Stochastic gradient descent x t+1 = x t rf(x t ; yĩt ) Much faster per iteration than gradient
More informationSymmetric Rendezvous in Graphs: Deterministic Approaches
Symmetric Rendezvous in Graphs: Deterministic Approaches Shantanu Das Technion, Haifa, Israel http://www.bitvalve.org/~sdas/pres/rendezvous_lorentz.pdf Coauthors: Jérémie Chalopin, Adrian Kosowski, Peter
More information1. Method 1: bisection. The bisection methods starts from two points a 0 and b 0 such that
Chapter 4 Nonlinear equations 4.1 Root finding Consider the problem of solving any nonlinear relation g(x) = h(x) in the real variable x. We rephrase this problem as one of finding the zero (root) of a
More informationStochastic Gradient Descent in Continuous Time
Stochastic Gradient Descent in Continuous Time Justin Sirignano University of Illinois at Urbana Champaign with Konstantinos Spiliopoulos (Boston University) 1 / 27 We consider a diffusion X t X = R m
More informationCommunication constraints and latency in Networked Control Systems
Communication constraints and latency in Networked Control Systems João P. Hespanha Center for Control Engineering and Computation University of California Santa Barbara In collaboration with Antonio Ortega
More informationJim Lambers MAT 460 Fall Semester Lecture 2 Notes
Jim Lambers MAT 460 Fall Semester 2009-10 Lecture 2 Notes These notes correspond to Section 1.1 in the text. Review of Calculus Among the mathematical problems that can be solved using techniques from
More informationMS 2001: Test 1 B Solutions
MS 2001: Test 1 B Solutions Name: Student Number: Answer all questions. Marks may be lost if necessary work is not clearly shown. Remarks by me in italics and would not be required in a test - J.P. Question
More informationLinear Codes, Target Function Classes, and Network Computing Capacity
Linear Codes, Target Function Classes, and Network Computing Capacity Rathinakumar Appuswamy, Massimo Franceschetti, Nikhil Karamchandani, and Kenneth Zeger IEEE Transactions on Information Theory Submitted:
More informationIntroduction to Optimization
Introduction to Optimization Konstantin Tretyakov (kt@ut.ee) MTAT.03.227 Machine Learning So far Machine learning is important and interesting The general concept: Fitting models to data So far Machine
More informationStochastic Variance Reduction for Nonconvex Optimization. Barnabás Póczos
1 Stochastic Variance Reduction for Nonconvex Optimization Barnabás Póczos Contents 2 Stochastic Variance Reduction for Nonconvex Optimization Joint work with Sashank Reddi, Ahmed Hefny, Suvrit Sra, and
More informationBenny Pinkas Bar Ilan University
Winter School on Bar-Ilan University, Israel 30/1/2011-1/2/2011 Bar-Ilan University Benny Pinkas Bar Ilan University 1 Extending OT [IKNP] Is fully simulatable Depends on a non-standard security assumption
More informationPrivate and Verifiable Interdomain Routing Decisions. Proofs of Correctness
Technical Report MS-CIS-12-10 Private and Verifiable Interdomain Routing Decisions Proofs of Correctness Mingchen Zhao University of Pennsylvania Andreas Haeberlen University of Pennsylvania Wenchao Zhou
More informationPrivacy in Statistical Databases
Privacy in Statistical Databases Individuals x 1 x 2 x n Server/agency ( ) answers. A queries Users Government, researchers, businesses (or) Malicious adversary What information can be released? Two conflicting
More informationAgreement algorithms for synchronization of clocks in nodes of stochastic networks
UDC 519.248: 62 192 Agreement algorithms for synchronization of clocks in nodes of stochastic networks L. Manita, A. Manita National Research University Higher School of Economics, Moscow Institute of
More informationLecture Notes 20: Zero-Knowledge Proofs
CS 127/CSCI E-127: Introduction to Cryptography Prof. Salil Vadhan Fall 2013 Lecture Notes 20: Zero-Knowledge Proofs Reading. Katz-Lindell Ÿ14.6.0-14.6.4,14.7 1 Interactive Proofs Motivation: how can parties
More informationCS 781 Lecture 9 March 10, 2011 Topics: Local Search and Optimization Metropolis Algorithm Greedy Optimization Hopfield Networks Max Cut Problem Nash
CS 781 Lecture 9 March 10, 2011 Topics: Local Search and Optimization Metropolis Algorithm Greedy Optimization Hopfield Networks Max Cut Problem Nash Equilibrium Price of Stability Coping With NP-Hardness
More informationEarly-Deciding Consensus is Expensive
Early-Deciding Consensus is Expensive ABSTRACT Danny Dolev Hebrew University of Jerusalem Edmond Safra Campus 9904 Jerusalem, Israel dolev@cs.huji.ac.il In consensus, the n nodes of a distributed system
More informationPERFECTLY secure key agreement has been studied recently
IEEE TRANSACTIONS ON INFORMATION THEORY, VOL. 45, NO. 2, MARCH 1999 499 Unconditionally Secure Key Agreement the Intrinsic Conditional Information Ueli M. Maurer, Senior Member, IEEE, Stefan Wolf Abstract
More informationRandomized Protocols for Asynchronous Consensus
Randomized Protocols for Asynchronous Consensus Alessandro Panconesi DSI - La Sapienza via Salaria 113, piano III 00198 Roma, Italy One of the central problems in the Theory of (feasible) Computation is
More informationmin f(x). (2.1) Objectives consisting of a smooth convex term plus a nonconvex regularization term;
Chapter 2 Gradient Methods The gradient method forms the foundation of all of the schemes studied in this book. We will provide several complementary perspectives on this algorithm that highlight the many
More informationCSE 417T: Introduction to Machine Learning. Lecture 11: Review. Henry Chai 10/02/18
CSE 417T: Introduction to Machine Learning Lecture 11: Review Henry Chai 10/02/18 Unknown Target Function!: # % Training data Formal Setup & = ( ), + ),, ( -, + - Learning Algorithm 2 Hypothesis Set H
More informationDistributed Algorithms for Consensus and Coordination in the Presence of Packet-Dropping Communication Links
COORDINATED SCIENCE LABORATORY TECHNICAL REPORT UILU-ENG-11-2208 (CRHC-11-06) 1 Distributed Algorithms for Consensus and Coordination in the Presence of Packet-Dropping Communication Links Part II: Coefficients
More informationIntroduction to Machine Learning Prof. Sudeshna Sarkar Department of Computer Science and Engineering Indian Institute of Technology, Kharagpur
Introduction to Machine Learning Prof. Sudeshna Sarkar Department of Computer Science and Engineering Indian Institute of Technology, Kharagpur Module 2 Lecture 05 Linear Regression Good morning, welcome
More informationSecure Computation. Unconditionally Secure Multi- Party Computation
Secure Computation Unconditionally Secure Multi- Party Computation Benny Pinkas page 1 Overview Completeness theorems for non-cryptographic faulttolerant distributed computation M. Ben-Or, S. Goldwasser,
More informationNetwork Algorithms and Complexity (NTUA-MPLA) Reliable Broadcast. Aris Pagourtzis, Giorgos Panagiotakos, Dimitris Sakavalas
Network Algorithms and Complexity (NTUA-MPLA) Reliable Broadcast Aris Pagourtzis, Giorgos Panagiotakos, Dimitris Sakavalas Slides are partially based on the joint work of Christos Litsas, Aris Pagourtzis,
More information