Distributed Optimization via Alternating Direction Method of Multipliers

  Рет қаралды 30,445

Microsoft Research

Microsoft Research

Күн бұрын

Пікірлер: 43
@baggepinnen
@baggepinnen 8 жыл бұрын
It would be helpful if the slides were shown more. I had to pause every time the slides were shown for only a few seconds to get what he was talking about.
@bryce1361
@bryce1361 Жыл бұрын
Since the video likes to show slides for a few seconds here are some timestamps. I only included the first time a slide showed up Goals: 2:10 Topic: Dual decomposition Dual problem: 6:21 Dual ascent: 8:25 Dual decomposition: 10:28, 14:09 Method of Multipliers: 15:05 Comparing Method of Multipliers to Dual Decomposition:19:19 Alternating Direction Method of Multipliers: 24:39 Convergence: 30:50 ADMM with scaled dual variables: 37:07 Related algorithms: 39:09 Common Patterns: Decomposition: 42:00 Proximal operator: 43:59 Quadratic objective: 44:10 Smooth objective: 49:08 Examples: Constrained convex optimization: 50:35 Lasso: 55:49 Sparse inverse covariance selection: 59:48 Sparse inverse covariance selection via ADMM: 1:00:46 Consensus and exchange: Consensus Optimization: 1:02:29 Consensus SVM example: 1:13:59 Distributed LASSO example: 1:16:48 Consensus Optimization via ADMM: 1:18:28 Exchange ADMM: 1:22:48 Exchange problem: 1:23:19 Solve time scaling: 1:28:07 Conclusions: Summary and conclusions: 1:31:01
@jayesh_yewale
@jayesh_yewale Жыл бұрын
Much helpful.. Thank you !
@TaoPangPang
@TaoPangPang 9 күн бұрын
Slides: web.stanford.edu/class/ee364b/lectures/admm_slides.pdf
@UNBADRO
@UNBADRO Жыл бұрын
Ahmad Bazzis lectures are much more clear and concise.
@Arashyou18
@Arashyou18 9 ай бұрын
Could you please send the link of Ahmad Bazzis lecture on this topic (ADMM)?
@vladdvorkin
@vladdvorkin 8 жыл бұрын
Thanks for the talk. It's quite clear and useful. It would be great if you uploaded the slides, it's not practical to perceive a slide for a couple of seconds only.
@nzdeepak
@nzdeepak 7 жыл бұрын
Thank you :)
@giannidigirolamo8868
@giannidigirolamo8868 7 жыл бұрын
I'm sorry to dislike such video...but...i don't care to see boyd's face. I want to see the slides
@shizk6773
@shizk6773 3 жыл бұрын
Yeah what kind of a MANIAC edited this video
@tomhayward8478
@tomhayward8478 7 жыл бұрын
Can you upload the video from the camera that shows the lecturer and the slides?
@panrel1292
@panrel1292 8 жыл бұрын
Sorry that the recording did not follow the presentation!
@houman1806
@houman1806 3 жыл бұрын
What is the definition of "CG step" (discussed on 47:55 )?
@qr-ec8vd
@qr-ec8vd 2 жыл бұрын
is it conjugate gradient? not sure
@hajerjm
@hajerjm Жыл бұрын
can't see the slides -.-
@haifa6004
@haifa6004 6 жыл бұрын
good explanation. but the camera is only focus on the advisor only. bad thing
@stdjmax
@stdjmax 6 жыл бұрын
Great Lecture One question. What is the code of the "pos()" in matlab from your ADMM examples z = pos(x_hat + u); Basicly what the pos() does ?
@FernandoCrema
@FernandoCrema 6 жыл бұрын
pos(x) = x_+ = max{0, x}
@lukec6886
@lukec6886 Жыл бұрын
(Probably) the slides in the talk: web.stanford.edu/~boyd/papers/pdf/springer_15_lect2.pdf
@JadtheProdigy
@JadtheProdigy 5 жыл бұрын
i like this guy, any more vids with him?
@marsag3118
@marsag3118 4 жыл бұрын
there are 40 videos of his lectures here on yt. look for "convex optimization"
@ehfo
@ehfo 6 жыл бұрын
isn't like an insualt to on Microsoft and use a mac laptop?
@panchajanya91
@panchajanya91 4 жыл бұрын
i wish, someone had told the cameraman to focus on slides when he would explain.
@siliur24
@siliur24 Жыл бұрын
quite interesting, but we're not here to admire the speaker's shirt ... showing the entire room with the slides (and so the laser pointer) would have more meaning ...
@matthewkarikomi1130
@matthewkarikomi1130 4 жыл бұрын
it works if you get the slide number from the right-hand footer next to boyd's head
@sui-chan.wa.kyou.mo.chiisai
@sui-chan.wa.kyou.mo.chiisai 4 жыл бұрын
Is camera sleeping?
@watchmanling
@watchmanling 3 жыл бұрын
It must be a business people doing the edit.
@selmane5315
@selmane5315 2 жыл бұрын
360p, really?
@daxu2684
@daxu2684 6 жыл бұрын
It would be great if the captains are available.
@chinmayaroutray4969
@chinmayaroutray4969 4 жыл бұрын
Camare should be on the slides
@broccoli322
@broccoli322 Жыл бұрын
Great talk but the way the video is presented is simply terrible. The slides need to be shown along with the presenter.
@ehfo
@ehfo 6 жыл бұрын
where is the slides?
@mohdfaizan-lz6xj
@mohdfaizan-lz6xj 7 жыл бұрын
Video making is not good. Only focused to speaker instead of slides.
@Savedbygrace952
@Savedbygrace952 11 ай бұрын
The cameraman is very poor.
@behzadnourani3195
@behzadnourani3195 4 жыл бұрын
This video is almost useless. It shows Stephen Boyd for 95 percent of the time and the formulas 5 percent of the time...
@tomhayward8478
@tomhayward8478 7 жыл бұрын
This is not useful without the slides
@kimiashayestehfard3677
@kimiashayestehfard3677 7 жыл бұрын
I dislike this video only because it did not show the slides!
Lessons from Megaprojects:  The Creators and Destroyers of Capital
59:51
Microsoft Research
Рет қаралды 1,4 М.
Convex Optimization and Applications - Stephen Boyd
2:31:46
The Qualcomm Institute
Рет қаралды 30 М.
99.9% IMPOSSIBLE
00:24
STORROR
Рет қаралды 31 МЛН
Quilt Challenge, No Skills, Just Luck#Funnyfamily #Partygames #Funny
00:32
Family Games Media
Рет қаралды 55 МЛН
BAYGUYSTAN | 1 СЕРИЯ | bayGUYS
36:55
bayGUYS
Рет қаралды 1,9 МЛН
Benjamin Recht: Optimization Perspectives on Learning to Control (ICML 2018 tutorial)
2:05:38
Optimization Part I - Stephen Boyd - MLSS 2015 Tübingen
59:05
Max Planck Institute for Intelligent Systems
Рет қаралды 16 М.
Lecture 12: Variants of ADMM and applications
1:57:46
Jalal Kazempour
Рет қаралды 4,2 М.
Daniel Kuhn: Data-driven and Distributionally Robust Optimization and Applications -- Part 2/2
1:39:18
DTU: Lectures on Power & Energy Systems
Рет қаралды 9 М.
Learn Machine Learning Like a GENIUS and Not Waste Time
15:03
Infinite Codes
Рет қаралды 225 М.
120 Years of Lyapunov's Methods
45:01
Stanford Autonomous Systems Laboratory
Рет қаралды 11 М.
On Gradient-Based Optimization: Accelerated, Stochastic and Nonconvex
1:07:08
Optimization Part II - Stephen Boyd - MLSS 2015 Tübingen
1:31:25
Max Planck Institute for Intelligent Systems
Рет қаралды 7 М.