Prof. Manfred K. Warmuth
Computer Science, UC Santa Cruz
Last x to a greater extent than or less pubs
All pubs
Classes
Other Interests
Coordinates
Fall 2017
Organisational
Class: Tu Thursday 3:20-4:55, Thieman Lecture Hall 1 Instructor business office hours: We 2-3, Fr 1-2 E2 357 Course piazza TAs Ehsan Amid as well as Tianyi Luo Ehsan give-and-take sections: We 10:40-11:45 Phys Sc. 140 Fr 12:00-1:05 populace as well as Marine B214 Ehsan business office hours: We 3:00-4:00 BE-118 Fr 2:00-3:00 BE-118 Tianyi give-and-take sections: Tu 1:30-2:35 Soc Sci ii - 165 We 1:20-2:25 Baskin 165 Tianyi business office hours: Tu 6:00-7:00 pm BE-312 C/D Thursday 5:30-6:30 pm BE-119 Prerequisite: Probability Theory, Linear Algebra Book: Pattern Recognition as well as Machine Learning past times Christopher Bishop Syllabus Summary of lectures1 Intro to Machine Learning using plication plumbing equipment equally an instance curve plumbing equipment Overfitting, complexity control, regularization Experimental setups w. training, validation as well as seek out sets Experimental setups w. training, validation as well as seek out sets Read: Section 1.1 of textbook Hw1 - Due inwards a week: Thursday 10-5-17 induce down of course of didactics ii Conditional Probabilities, Bayes rule Bayes inwards the tube Frequentist versus Bayes three Discuss Hw1 solution sol 1 w. source code past times Andrew Stolman & Zayd Hammoudeh sol 2 code past times Tianyi as well as Ehsan Coding Theory, relative entropy Relevant chapters from Cover/Thomas book Visualizations of relative entropies Maple file Finish reading Chapter 1 of textbook Convexity as well as Jensen's inequality Hw2 - Due inwards a week: Thursday 10-12-17 induce down of course of didactics (Exercises of Chapter 1) four Basic distributions, exponential families Duality for exponential families five Overview of how to build convex loss functions an regularizer & grooming of parameters Linear as well as logistic regression equally examples Bregman diversions for constructing regularizers as well as losses Hw2 Solution Tianyi as well as Ehsan Hw3 - Due Tu 10-24-17 induce down of course of didactics vi seven 8 HW3 due HW4 (programming a uncomplicated Neural Net)