ICS 235 Machine Learning Methods Department of Information and Computer Sciences University of Hawai i at M noa Kyungim Baek Lecture 1 Course logistics Brief introduction to Machine learning First class survey 2 1 Assignments Install Python and Jupyter Notebook before next class You may refer to the install python pdf file in Laulima Class information Lectures MW 12 00 pm 1 15 pm in KUY 305 Instructor Kyungim Baek Email kyungim at hawaii edu Please prefix the subject line of your email with ICS 235 or DATA 235 Office POST 303 F Office hours Wednesday 1 30 pm 2 45 pm 4 30 pm 5 30 pm or by appointment Please contact me to schedule an appointment if you can t make it to the regular office hours 3 4 2 Prerequisites MATH 203 MATH 215 MATH 241 or MATH 251A Knowledge of basic computer science principles and skills to be able to write a program in Python Grading criteria and exam schedule Grading criteria and weights 50 Homework assignments 25 Midterm exam 25 Final exam Exam dates and time Midterm exam Monday October 21 12 00 pm 1 15 pm Final exam Monday December 16 12 00 pm 2 00 pm 5 6 3 Grading Grading will not be on a curve Extra credits Extra credit work will not be offered to individual students Extra credits based on class attendance 0 5 points for an attendance rate 80 and 90 1 point for an attendance rate of 90 or higher Extra credits based on CES course evaluation completion rates 0 5 points for a CES completion rate 80 and 90 1 point for a CES completion rate of 90 or higher Grading scale A 98 A 90 and 98 B 88 and 90 B 80 and 88 C 78 and 80 C 70 and 78 D 68 and 70 D 60 and 68 F 60 7 8 4 Homework submission Homework should be turned in electronically via Laulima by 11 55 pm on the due date Late submission 20 reduction in points if submitted within 24 hours past the deadline Submissions made more than 24 hours past the deadline will not be accepted No extensions to the deadline will be granted Submission by e mail will NOT be accepted unless Laulima is down at the time of submission Exams No makeup exams will be given unless the instructor is provided with official documented proof of an emergency that prevented the student from attending the exam If a makeup exam is offered the questions may differ from those on the regular exam and the difficulty level may not be the same 9 10 5 Academic integrity Homework assignments and tests are checked for cheating and plagiarism All occurrences of academic dishonesty will result in zero credit for the assignment exam in question the first time it is detected Any subsequent infraction will result in a grade of F for the course This applies to all involved those who copy other s work and who allow their work to be copied Reference texts Course in ML by Hal Daum III Andreas C M ller and Sarah Guido Introduction to Machine Learning with Python Christopher Bishop Pattern Recognition and Machine Learning 12 14 6 Course website on Laulima Read through the syllabus for detailed information on the course Questions 15 16 7 What is machine learning Field of study that gives computers the ability to learn without being explicitly programmed Samuel 1959 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 Mitchell 1997 Programming computers to optimize a performance criterion using example data or past experience Alpaydin 2010 Computational methods using experience to improve performance or to make accurate predictions Mohri Rostamizadeh Talwalkar 2012 What is machine learning Field of study that gives computers the ability to learn without being explicitly programmed Samuel 1959 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 Mitchell 1997 Programming computers to optimize a performance criterion using example data or past experience Alpaydin 2010 Computational methods using experience to improve performance or to make accurate predictions Mohri Rostamizadeh Talwalkar 2012 17 18 8 Why use machine learning Traditional approaches Rule based or knowledge based Describe a set of rules for making decisions or deductions May require long list of rules complex problems May have no known rules algorithm Hard to come up with robust set of rules Machine learning approaches Data driven Learn from data Can adapt to new data Can provide insight about large amounts of data Figures from Oh 2017 20 ML algorithms Source https www kaggle com code paultimothymooney kaggle survey 2022 all results 23 9 ML algorithms 10 ML algorithms to know in 2024 Linear regression Logistic regression Na ve Bayes Decision tree Random forest K nearest neighbor KNN K means Support vector machine SVM Apriori Gradient boosting Source https www coursera org articles machine learning algorithms Types of ML Categorization based on the type of supervision during the training Supervised learning Unsupervised learning Reinforcement learning 24 25 10 Supervised learning Dataset contains input output pairs xi yi xi d dimensional input or feature vector yi desired output target Learn f xi yi X Model Algorithm y The learned model f is used to predict target values of new examples 27 Classification and regression Typical supervised learning tasks Classification targets are discrete e g spam filter Regression targets are real values e g housing price prediction Figure credit P Sadowski 28 11 Supervised learning algorithms k nearest neighbors Logistic regression Linear regression Decision trees Support vector machines Neural networks 29 Unsupervised learning No labels only given the data points xi Unsupervised learning algorithms Clustering Outlier Anomaly detection Visualization Dimensionality reduction Topic modeling Figure credit P Sadowski 32 12 Reinforcement learning An agent learning to interact with an environment with some ultimate goal policy Agent Environment action a state s reward r Next class Machine learning concepts A simple example Brief introduction to Python with Jupyter notebook 34 35 13 Please complete the first class survey before you leave Thank you 36 14