機器學習之網路應用

Nowadays, machine learning is a popular field and has been widely used in several applications such as pattern recognition, computer vision, natural language processing, computer games, data analytics, etc. This new graduate-level course focuses on applying machine learning methods to computer networks research. It is specifically offered to those graduate students who want to conduct research on applying machine learning methods to computer networks. In this course, we will start with a short introduction to the ideas, methods, and current status of machine learning. Then, we will move on to study many high-quality research papers that have successfully applied machine learning to different aspects of computer networks. Students will do two labs to get hands-on experiences on using the Scikit-learn machine learning package to solve classification/prediction problems. Coursework: • Read and study assigned paper (each student should study one paper) • Present assigned paper in the class (by each student) • Attend classes and discuss papers • Do two labs • Conduct a project (Two students form a team to do this task.) • Present the final project results in the class (by the team) • Write the mid-term and final project reports (by the team) • Take the mid-term and final exams

1. Having taken the "Introduction to Computer Networks" course is a must. 2. Having taken the "Introduction to Machine Learning" course is a plus but not required. 3. Know how to write Python programs (or can learn it by yourself quickly) This course is not intended to be a course that uses the whole semester to teach machine learning. Instead, its main focus is on applying machine learning to computer networks. We will only briefly present the ideas of the machine learning methods that have been applied to computer networks, and about 2/3 of the semester will be used to study how to apply machine learning to many aspects of computer networks to solve important problems.

• 15% Assigned paper reading and in-class presentation • 10% Class participation and discussions • 10% Two labs (5% each) • 15% Mid-term exam • 10% Mid-term project report (5 pages long) • 15% In-class presentation of the final project results • 10% Final project report (10 pages long) • 15% Final exam

Currently, there is no good textbook on applying machine learning to networks and communications. Thus, this course does not use a textbook. Instead, students taking this course will need to read many high-quality papers in this field. The following lists some machine learning books that students may reference: • Ian H. Witten, Eibe Frank, Mark A. Hall, Christopher J. Pal, “Data Mining: Practical Machine Learning Tools and Techniques,” Fourth edition, Morgan Kaufmann Publishers, 2016. • John D. Kelleher, Brian Mac Namee, and Aoife D'arcy, “Fundamentals of Machine Learning for Predictive Data Analytics,” The MIT press, Cambridge, Massachusetts, 2015. • Ethem Alpaydin, “Introduction to Machine Learning,” Third edition, The MIT press, Cambridge, Massachusetts, 2014. • Andreas C. Muller and Sarah Guido, “Introduction to Machine Learning with Python,” First edition, O'Reilly, 2016. • Ian Goodfellow, Yoshua Bengio, Aaron Courville, “Deep Learning,” First edition, The MIT press, Cambridge, Massachusetts, 2016. • Francois Chollet, “Deep Learning with Python,” First edition, Manning Publications Co., Shelter Island, New York, 2018. Recommended language, platforms, and tools: • Python 3 • Anaconda • Jupyter notebook • Scikit-learn (or Keras/Tensorflow)

地點

EC413

時間

Monday C&D time slots

聯絡方式

shieyuan@cs.nctu.edu.tw