COMPSCI 762 : Advanced Machine Learning

Science

2020 Semester One (1203) (15 POINTS)

Course Prescription

Machine learning is a branch of artificial intelligence concerned with making accurate, interpretable, computationally efficient, and robust inferences from data to solve a given problem. Students should understand the foundations of machine learning, and introduce practical skills to solve different problems. Students will explore research frontiers in machine learning. Recommended preparation: COMPSCI 220, 225 and STATS 101

Course Overview

This course will provide the foundations of machine learning, and provide practical skills to solve different problems. Students will explore research frontiers in machine learning while learning  about the theoretical underpinnings of machine learning.

This course provides a broad introduction to machine learning. Topics include: (i) Supervised learning (decision trees, support vector machines, neural networks). (ii) Unsupervised learning (clustering, association rule mining, anomaly detection). (iii) fundamental practices in machine learning (bias/variance theory).

Course Requirements

Prerequisite: Approval of Academic Head or nominee Restriction: COMPSCI 361

Capabilities Developed in this Course

Capability 1: Disciplinary Knowledge and Practice
Capability 2: Critical Thinking
Capability 3: Solution Seeking
Capability 4: Communication and Engagement
Capability 5: Independence and Integrity
Capability 6: Social and Environmental Responsibilities

Learning Outcomes

By the end of this course, students will be able to:
  1. Demonstrate an understanding - Demonstrate technical knowledge of the underlying principals and concepts of machine learning science. (Capability 1 and 4)
  2. Use and apply - Apply efficient machine learning algorithms on a problem. (Capability 1, 2 and 4)
  3. Evaluate and reflect - Design evaluation procedures to evaluate a model. (Capability 1, 2, 3 and 5)
  4. Understand and use - Interpret the results of machine learning run on real data. (Capability 2, 3, 4 and 5)
  5. Understand and reflect - Assess the benefits/drawbacks of competing models and algorithms, relevant to real problems. (Capability 1 and 4)
  6. Demonstrate an understanding - Demonstrate your knowledge about cutting edge research streams and developments in machine learning. (Capability 1, 2 and 3)
  7. Identify and critically evaluate - Recognise real-world problems suitable to machine learning. (Capability 1, 2, 4, 5 and 6)

Assessments

Assessment Type Percentage Classification
Assignments 30% Individual Coursework
Test 22% Individual Test
Final Exam 48% Individual Examination
Assessment Type Learning Outcome Addressed
1 2 3 4 5 6 7
Assignments
Test
Final Exam

To pass the course, students must pass both the Practical component (Assignments) and the Theory component (Exam plus Test) separately, as well as obtaining 50% in their overall final mark.

Key Topics

  • Hypothesis Bias/ Regression
  • Decision Trees
  • Ensembles
  • Statistics
  • Neural Network
  • Biological Inspired Techniques: Genetics Algorithm, PSO
  • Feature Selection
  • Bayesian
  • Reinforcement Learning
  • Support Vector Machine
  • kNN
  • k-Means and Partition based Clustering
  • Hierarchal based Clustering
  • DBScan
  • Data Streams, Concept Drift
  • Association Rule Mining
  • Pattern Mining
  • Anomaly Detection

Learning Resources


Mitchell, T. M. (1997). Machine learning. New York: McGraw-Hill.

Witten, I. H., Frank, E., Hall, M., & Pal, C. (2016). Data Mining (4th edition). Retrieved from https://ebookcentral.proquest.com/lib/auckland/detail.action?docID=4708912#

Žliobaitė, I., Pechenizkiy, M., & Gama, J. (2015). An Overview of Concept Drift Applications. In N. Japkowicz & J. Stefanowski (Eds.), Big data analysis: new algorithms for a new society: Vol. Studies in big data (pp. 91–114). https://doi.org/10.1007/978-3-319-26989-4_4

Gama, J. (2010). Knowledge discovery from data streams. Retrieved from http://marc.crcnetbase.com/isbn/9781439826126

Bifet, A., Gavaldà, R., Holmes, G., & Pfahringer, B. (2018). Machine Learning for Data Streams with Practical Examples in MOA. Retrieved from https://mitpress.mit.edu/books/machine-learning-data-streams

Tan, P.-N., Steinbach, M., & Kumar, V. (2005). Introduction to data mining (1st ed). Boston: Pearson Addison Wesley.

Han, J., Kamber, M., & Pei, J. (2014). Data Mining (3rd Revised ed.). Morgan Kaufmann Publishers.
Bishop, C. M. (2006). Pattern Recognition and Machine Learning (1st ed. 2006. Corr. 2nd printing 2011). New York, NY: Springer-Verlag New York Inc.

Special Requirements

You must pass both the theory [the exam and test] and the practical [i.e., assignments] components to pass this course.

Workload Expectations

This course is a standard 15 point course and students are expected to spend 10 hours per week involved in each 15 point course that they are enrolled in.

For this course, you can expect 3 hours of lectures, a 1 hour tutorial, 3 hours of reading and thinking about the content and 3 hours of work on assignments and/or test preparation.

Digital Resources

Course materials are made available in a learning and collaboration tool called Canvas which also includes reading lists and lecture recordings (where available).

Please remember that the recording of any class on a personal device requires the permission of the instructor.

Copyright

The content and delivery of content in this course are protected by copyright. Material belonging to others may have been used in this course and copied by and solely for the educational purposes of the University under license.

You may copy the course content for the purposes of private study or research, but you may not upload onto any third party site, make a further copy or sell, alter or further reproduce or distribute any part of the course content to another person.

Academic Integrity

The University of Auckland will not tolerate cheating, or assisting others to cheat, and views cheating in coursework as a serious academic offence. The work that a student submits for grading must be the student's own work, reflecting their learning. Where work from other sources is used, it must be properly acknowledged and referenced. This requirement also applies to sources on the internet. A student's assessed work may be reviewed against online source material using computerised detection mechanisms.

Inclusive Learning

All students are asked to discuss any impairment related requirements privately, face to face and/or in written form with the course coordinator, lecturer or tutor.

Student Disability Services also provides support for students with a wide range of impairments, both visible and invisible, to succeed and excel at the University. For more information and contact details, please visit the Student Disability Services’ website at http://disability.auckland.ac.nz

Special Circumstances

If your ability to complete assessed coursework is affected by illness or other personal circumstances outside of your control, contact a member of teaching staff as soon as possible before the assessment is due.

If your personal circumstances significantly affect your performance, or preparation, for an exam or eligible written test, refer to the University’s aegrotat or compassionate consideration page: https://www.auckland.ac.nz/en/students/academic-information/exams-and-final-results/during-exams/aegrotat-and-compassionate-consideration.html.

This should be done as soon as possible and no later than seven days after the affected test or exam date.

Student Feedback

During the course Class Representatives in each class can take feedback to the staff responsible for the course and staff-student consultative committees.

At the end of the course students will be invited to give feedback on the course and teaching through a tool called SET or Qualtrics. The lecturers and course co-ordinators will consider all feedback.

Your feedback helps to improve the course and its delivery for all students.

Student Charter and Responsibilities

The Student Charter assumes and acknowledges that students are active participants in the learning process and that they have responsibilities to the institution and the international community of scholars. The University expects that students will act at all times in a way that demonstrates respect for the rights of other students and staff so that the learning environment is both safe and productive. For further information visit Student Charter (https://www.auckland.ac.nz/en/students/forms-policies-and-guidelines/student-policies-and-guidelines/student-charter.html).

Disclaimer

Elements of this outline may be subject to change. The latest information about the course will be available for enrolled students in Canvas.

In this course you may be asked to submit your coursework assessments digitally. The University reserves the right to conduct scheduled tests and examinations for this course online or through the use of computers or other electronic devices. Where tests or examinations are conducted online remote invigilation arrangements may be used. The final decision on the completion mode for a test or examination, and remote invigilation arrangements where applicable, will be advised to students at least 10 days prior to the scheduled date of the assessment, or in the case of an examination when the examination timetable is published.

Published on 18/12/2019 11:42 p.m.