CS4780 5780 Machine Learning Fall 2011 Thorsten Joachims Cornell University Department of Computer Science Outline of Today Who we are Prof Thorsten Joachims TAs Karthik Raman Chenhao Tan Adith Swaminathan Consultants Mevlana Gemici Anthony Chang Nic Williamson Heran Yang Boiar Qin What is learning Why should a computer be able to learn Examples of machine learning What it takes to build a learning system Syllabus Administrivia One Definition of Learning Definition Mitchell A computer program is said to learn from experience E with respect to some class of tasks T and performance measure P if its performance at tasks in T as measured by P improves with experience E Syllabus Concept Learning Hypothesis space version space Instance Based Learning k nearest neighbor collaborative filtering Decision Trees TDIDT attribute selection pruning and overfitting ML Experimentation hypothesis tests resampling estimates Linear Rules Perceptron duality mistake bound Support Vector Machines optimal hyperplane kernels stability Generative Models Na ve Bayes linear discriminant analysis Hidden Markov Models probabilistic model estimation Viterbi Structured Output Prediction predicting sequences rankings etc Learning Theory PAC learning mistake bounds Clustering HAC Clustering k means mixture of Gaussians Textbook and Course Material Main Textbooks Main Tom Mitchell Machine Learning McGraw Hill 1997 Cristianini Shawe Taylor Introduction to Support Vector Machines Cambridge University Press 2000 online Schoelkopf Smola Learning with Kernels MIT Press 2001 online Course pack one chapter Additional References optional Ethem Alpaydin Introduction to Machine Learning MIT Press 2004 See other references on course web page Course Notes Slides available on course homepage Material on blackboard Pre Requisites and Related Courses Pre Requisites Programming skills e g CS 2110 Basic linear algebra e g MATH2940 Basic probability theory e g CS 2800 Short exam to test prereqs Related Courses CS4700 Foundations of Artificial Intelligence CS4758 Robot Learning CS4300 Information Retrieval CS6780 Advanced Machine Learning CS6784 Advanced Topics in Machine Learning CS6740 Advanced Language Technologies Homwork Assignments Assignments 5 homework assignments Some problem sets some programming and experiments Policies Assignments are due at the beginning of class on the due date in hardcopy Code must be submitted via CMS by the same deadline Assignments turned in late will drop 5 points for each period of 24 hours for which the assignment is late Everybody had 3 free late days Use them wisely No assignments will be accepted after the solutions have been made available typically 3 days after deadline Typically collaboration of two students see each assignment for detailed collaboration policy We run automatic cheating detection Must state all sources of material used in assignments or project Please review Cornell Academic Integrity Policy Exams and Quizzes In class Quizzes A few per semester No longer than 5 minutes You can miss one of them Exams Two Prelim exams In class No final exam Final Project Organization Self defined topic related to your interests and research Groups of 3 4 students Each group has TA as advisor Deliverables Project proposal 2 weeks after fall break Meetings with TA to discuss progress Short presentation in class last week of classes Project report exam period Grading Deliverables 2 Prelim Exams Final Project Homeworks 5 assignments Quizzes in class PreReq Exam Participation 40 of Grade 15 of Grade 35 of Grade 5 of Grade 2 of Grade 3 of Grade Outlier elimination For homeworks and quizzes the lowest grade is replaced by the second lowest grade How to Get in Touch Online http www cs cornell edu Courses cs4780 2011fa Piazza forum Videonote Email Addresses Thorsten Joachims tj cs cornell edu Karthik Raman karthik cs cornell edu Chenhao Tan chenhao cs cornell edu Adith Swaminathan adith cs cornell edu Mevlana Gemici Anthony Chang Nic Williamson Heran Yang Boiar Qin Office Hours Thorsten Joachims Tuesdays 2 40pm 3 30pm 4153 Upson Hall Other office hours TBD