Download presentation
Presentation is loading. Please wait.
Published byMerilyn Benson Modified over 9 years ago
1
Peter Richtárik Parallel coordinate NIPS 2013, Lake Tahoe descent methods
2
I.Introduction II.Regularized Convex Optimization III.Parallel CD (PCDM) IV.Accelerated Parallel CD (APCDM) V.ESO? Good ESO? VI.Experiments in 10^9 dim VII.Distributed CD (Hydra) VIII.Nonuniform Parallel CD (‘NSync) IX.Mini-batch SDCA for SVMs OUTLINE POSTER
3
Part I. Introduction
4
What is Randomized Coordinate Descent ?
5
Find the minimizer of 2D Optimization Contours of a function Goal:
6
Randomized Coordinate Descent in 2D N S E W
7
1 N S E W
8
1 N S E W 2
9
1 2 3 N S E W
10
1 2 3 4 N S E W
11
1 2 3 4 N S E W 5
12
1 2 3 4 5 6 N S E W
13
1 2 3 4 5 N S E W 6 7 S O L V E D !
14
Convergence of Randomized Coordinate Descent Strongly convex F Smooth or ‘simple’ nonsmooth F ‘difficult’ nonsmooth F Focus on n (big data = big n)
15
Parallelization Dream Depends on to what extent we can add up individual updates, which depends on the properties of F and the way coordinates are chosen at each iteration SerialParallel What do we actually get? WANT
16
How (not) to Parallelize Coordinate Descent
17
“Naive” parallelization Do the same thing as before, but for MORE or ALL coordinates & ADD UP the updates
18
Failure of naive parallelization 1a 1b 0
19
Failure of naive parallelization 1 1a 1b 0
20
Failure of naive parallelization 1 2a 2b
21
Failure of naive parallelization 1 2a 2b 2
22
Failure of naive parallelization 2 O O P S !
23
1 1a 1b 0 Idea: averaging updates may help S O L V E D !
24
Averaging can be too conservative 1a 1b 0 1 2a 2b 2 a n d s o o n...
25
Averaging may be too conservative 2 WANT BAD!!! But we wanted:
26
What to do? Averaging: Summation: Update to coordinate i i-th unit coordinate vector Figure out when one can safely use:
27
Part II. Regularized Convex Optimization
28
Problem Convex (smooth or nonsmooth) Convex (smooth or nonsmooth) - separable - allow Loss Regularizer
29
Regularizer: examples No regularizerWeighted L1 norm Weighted L2 norm Box constraints e.g., SVM dual e.g., LASSO
30
Loss: examples Quadratic loss L-infinity L1 regression Exponential loss Logistic loss Square hinge loss BKBG’11 RT’11b TBRS’13 RT ’13a FR’13a
31
3 models for f with small 1 2 3 Smooth partially separable f [RT’11b ] Nonsmooth max-type f [FR’13] f with ‘bounded Hessian’ [BKBG’11, RT’13a ]
32
Part III. Parallel Coordinate Descent P.R. and Martin Takáč Parallel Coordinate Descent Methods for Big Data Optimization arXiv:1212.0873, 2012 [IMA Leslie Fox Prize 2013] P.R. and Martin Takáč Parallel Coordinate Descent Methods for Big Data Optimization arXiv:1212.0873, 2012 [IMA Leslie Fox Prize 2013]
33
Randomized Parallel Coordinate Descent Method Random set of coordinates (sampling) Current iterateNew iteratei-th unit coordinate vector Update to i-th coordinate
34
ESO: Expected Separable Overapproximation Definition [RT’11b] 1. Separable in h 2. Can minimize in parallel 3. Can compute updates for only Shorthand: Minimize in h
35
Convergence rate: convex f average # updated coordinates per iteration # coordinatesstepsize parameter error tolerance # iterations implies Theorem [RT’11b]
36
Convergence rate: strongly convex f implies Strong convexity constant of the regularizer Strong convexity constant of the loss f Theorem [RT’11b]
37
Part IV. Accelerated Parallel CD Olivier Fercoq and P.R. Accelerated, Parallel and Proximal Coordinate Descent Manuscript, 2013 Olivier Fercoq and P.R. Accelerated, Parallel and Proximal Coordinate Descent Manuscript, 2013
38
3 x YES
39
The Algorithm Parallel CD step ESO parameters
40
Complexity average # updated coordinates per iteration # coordinates error tolerance # iterations implies Theorem [FR’13b]
41
Part V. ESO? Good ESO? (e.g., partially separable f & doubly uniform S)
42
Serial uniform sampling Probability law:
43
-nice sampling Probability law: Good for shared memory systems
44
Doubly uniform sampling Probability law: Can model unreliable processors / machines
45
ESO for partially separable functions and doubly uniform samplings Theorem [RT’11b] 1 Smooth partially separable f [RT’11b ]
46
Theoretical speedup Much of Big Data is here! degree of partial separability # coordinates # coordinate updates / iter WEAK OR NO SPEEDUP: Non-separable (dense) problems LINEAR OR GOOD SPEEDUP: Nearly separable (sparse) problems
48
n = 1000 (# coordinates) Theory
49
Practice n = 1000 (# coordinates)
50
Part VI. Experiment with a 1 billion-by-2 billion LASSO problem
51
Optimization with Big Data * in a billion dimensional space on a foggy day Extreme* Mountain Climbing =
52
Coordinate Updates
53
Iterations
54
Wall Time
55
L2-reg logistic loss on rcv1.binary
56
Part VII. Distributed CD P.R. and Martin Takáč Distributed Coordinate Descent Method for Learning with Big Data arXiv:1310.2059, 2013 P.R. and Martin Takáč Distributed Coordinate Descent Method for Learning with Big Data arXiv:1310.2059, 2013
57
Distributed -nice sampling Probability law: Machine 2Machine 1Machine 3 Good for a distributed version of coordinate descent
58
ESO: Distributed setting Theorem [RT’13b] 3 f with ‘bounded Hessian’ [BKBG’11, RT’13a ] spectral norm of the data
59
Bad partitioning at most doubles # of iterations spectral norm of the partitioning Theorem [RT’13b] # nodes # iterations = implies # updates/node
61
LASSO with a 3TB data matrix 128 Cray XE6 nodes with 4 MPI processes (c = 512) Each node: 2 x 16-cores with 32GB RAM = # coordinates
62
References
63
Shai Shalev-Shwartz and Ambuj Tewari, Stochastic methods for L1-regularized loss minimization. JMLR 2011. Yurii Nesterov, Efficiency of coordinate descent methods on huge-scale optimization problems. SIAM Journal on Optimization, 22(2):341-362, 2012. [RT’11b] P.R. and Martin Takáč, Iteration complexity of randomized block-coordinate descent methods for minimizing a composite function. Mathematical Prog., 2012. Rachael Tappenden, P.R. and Jacek Gondzio, Inexact coordinate descent: complexity and preconditioning, arXiv: 1304.5530, 2013. Ion Necoara, Yurii Nesterov, and Francois Glineur. Efficiency of randomized coordinate descent methods on optimization problems with linearly coupled constraints. Technical report, Politehnica University of Bucharest, 2012. Zhaosong Lu and Lin Xiao. On the complexity analysis of randomized block- coordinate descent methods. Technical report, Microsoft Research, 2013. References: serial coordinate descent
64
[BKBG’11] Joseph Bradley, Aapo Kyrola, Danny Bickson and Carlos Guestrin, Parallel Coordinate Descent for L1-Regularized Loss Minimization. ICML 2011 [RT’12] P.R. and Martin Takáč, Parallel coordinate descen methods for big data optimization. arXiv:1212.0873, 2012 Martin Takáč, Avleen Bijral, P.R., and Nathan Srebro. Mini-batch primal and dual methods for SVMs. ICML 2013 [FR’13a] Olivier Fercoq and P.R., Smooth minimization of nonsmooth functions with parallel coordinate descent methods. arXiv:1309.5885, 2013 [RT’13a] P.R. and Martin Takáč, Distributed coordinate descent method for big data learning. arXiv:1310.2059, 2013 [RT’13b] P.R. and Martin Takáč, On optimal probabilities in stochastic coordinate descent methods. arXiv:1310.3438, 2013 References: parallel coordinate descent Good entry point to the topic (4p paper)
65
P.R. and Martin Takáč, Efficient serial and parallel coordinate descent methods for huge-scale truss topology design. Operations Research Proceedings 2012. [FR’13b] Olivier Fercoq and P.R., Accelerated, Parallel and Proximal Coordinate Descent. Manuscript, 2013. Rachael Tappenden, P.R. and Burak Buke, Separable approximations and decomposition methods for the augmented Lagrangian. arXiv:1308.6774, 2013. Indranil Palit and Chandan K. Reddy. Scalable and parallel boosting with MapReduce. IEEE Transactions on Knowledge and Data Engineering, 24(10):1904-1916, 2012. Shai Shalev-Shwartz and Tong Zhang, Accelerated mini-batch stochastic dual coordinate ascent. NIPS 2013. References: parallel coordinate descent
Similar presentations
© 2025 SlidePlayer.com. Inc.
All rights reserved.