Machine Learning 10725Instructor: Ryan Tibshirani (ryantibs at cmu dot edu)Important note: please direct emails on all course related matters to the Education Associate, not the Instructor. The subject line of all emails should begin with "[10725]". Education Associate: Daniel Bird (dpbird at andrew dot cmu dot edu) TAs: Chen Dan (cdan at andrew dot cmu dot edu) PoWei Wang (powei at andrew dot cmu dot edu) Lingxiao Zhao (lingxia1 at andrew dot cmu dot edu) Lecture times: Mondays and Wednesdays 1:302:50pm, Baker Hall A51 Office hours: RT: Wednesdays 34pm, Baker 229B Syllabus: here. Scribing: sign up, latex template. Discussions: Piazza group. 

Theory I: Fundamentals  
Mon Aug 26  Introduction  Slides,  
Wed Aug 28  Convexity I: Sets and functions  Slides,  
Mon Sept 2  (Labor day, no class)  
Wed Sept 4  Convexity II: Optimization basics  Slides,  
Mon Sept 9  Canonical problem forms  Slides,  Hw 1 due Fri Sept 13  
Algorithms I: Firstorder methods  
Wed Sept 11  Gradient descent  
Mon Sept 16  Subgradients  
Wed Sept 18  Subgradient method  
Mon Sept 23  Proximal gradient descent  
Wed Sept 25  Stochastic gradient descent  Hw 2 due Fri Sept 27  
Theory II: Optimality and duality  
Mon Sept 30  Duality in linear programs  
Wed Oct 2  Duality in general programs  
Mon Oct 7  KKT conditions  
Wed Oct 9  Duality uses and correspondences  Hw 3 due Fri Oct 11  
Algorithms II: Secondorder methods  
Mon Oct 14  Newton's method  
Wed Oct 16  Barrier method  
Mon Oct 21  Primaldual interiorpoint methods  
Wed Oct 23  QuasiNewton methods  Hw 4 due Fri Oct 25  
Advanced topics  
Mon Oct 28  Numerical linear algebra  
Wed Oct 30  Coordinate descent  
Mon Nov 4  Dykstra's algorithm  
Wed Nov 6  Dual decomposition, ADMM  
Mon Nov 11  Distributed ADMM  Hw 5 due Fri Nov 15  
Wed Nov 13  FrankWolfe method  
Mon Nov 18  Implicit regularization  
Wed Nov 20  Guest lecture I  
Mon Nov 25  Guest lecture II  
Wed Nov 27  (Thanksgiving break, no class)  
Mon Dec 2  Guest lecture III or recap  Hw 6 due Mon Dec 2  
Wed Dec 4  Little test  Little test 