Machine learning techniques are widely used in many computing applications; for example, in web    search engines, spam filtering, speech and image recognition, computer games, machine vision,       credit card fraud detection, stock market analysis and product marketing applications. Machine          learning implies that there is some improvement that results from the learning program having seen   some data. The improvement can be in terms of some performance program (e.g., learning an expert system or improving the performance of a planning or scheduling program), in terms of finding an      unknown relation in the data (e.g., data mining, pattern analysis), or in terms of customizing adaptive systems (e.g., adaptive user-interfaces or adaptive agents).

This course is research oriented. The practical component involves working on a real-world like       research project developed with the help of the teaching team. The research project involves           definition of research questions, project planning,  data analysis workflow, programming,                  collaboration effort and regular communication of project progress in a oral or written form, including writing a literature review and a final research report. Programming skills are necessary for this course. The practical component of the course expects group work.

Lectures will introduce some of the recent developments in the field of machine learning. Students are expected to attend the lectures as they will be evaluated about their content in the final exam. Although this class is research oriented, students should be aware that the final exam weights      significantly (50%) on the final grade.

Course Requirements

Prerequisite: CS361 or CS762

Lectures topics

Week 1-7: Introduction, Advanced Neural Networks

Week 8-12:  Fairness in Machine Learning, Adversarial Learning

Learning Outcomes

The students will be able to:

Discuss the idea that all machine learning algorithms have a basis and will be able to describe the basis of several algorithms

Discuss the theory that for a particular dataset one algorithm will perform well and for another

dataset a different algorithm will perform well. There is no one algorithm that performs well on all datasets.

Independently develop and carry out to completion a research project addressing real-world

problems using appropriate machine learning methodology and open-source datasets in a group of preferably 5 students.

Design a good set of experiments for determining the answer to some basic research question,

such that they can show that the experiments actually support the question they are asking. Assessments

Your final grade will consist of a number of assignment marks worth 40% combined and an exam    worth 60%. This  is a research based course, so the assignment marks will be based on a research project work, performed in a collaboration with fellow students.

There is a practical and a theoretical pass on this paper. To pass the course, students must pass both the Practical component (Assignments) and the Theory component (Exam) separately, as well as obtaining 50% in their overall final mark. This means you need to have more than 50% of the       exam marks and 50% of the marks for the assignments. Hence make sure you allocate enough time for both the practical assessments and exam preparation.

We use the standard university grade boundaries. >89.5 for A+, then 5 mark increments down to >49.5 for C-.

You can find a list of assignments and the timeline for the semester here: () . We strongly advise students to have a look a this document at the beginning of the semester to get an idea of what types of assignment are given along the semester.

Workload Expectations

This course is a  15 point course and students are expected to spend about  150 hours in total. Expected weekly workload (excluding the teaching break): 1 hour lecture, 1 hour lecture review, 1-2 hour reading and thinking, and 6-7 hours for the research project/assignments.

Students  are  expected  to  spend  additional  30  hours  for  assessment  (e.g.  exam,  presentation) preparation.

Teaching Staff

Thomas Lacombe (lecturer, course coordinator)

Room: 419, Computer Science Building (Building 303)

Email: thomas.lacombe@auckland.ac.nz (mailto:thomas.lacombe@auckland.ac.nz)

Office hours: Only by appointment; For many matters where the issue is generally applicable to other students, you may be best advised to go through a class representative.

Katerina Taskova  (lecturer - second part of the semester)

Room: 493, Computer Science Building (Building 303S)

Email: katerina.taskova@auckland.ac.nz (mailto:katerina.taskova@auckland.ac.nz) Office hours: TBA during mid-term break

Katharina Dost   (tutor/marker)

Email: kdos481@aucklanduni.ac.nz (mailto:kdos481@aucklanduni.ac.nz)

Lecture Times and Schedule

Tentative schedule for lectures and presentations*:

* This may vary depending on the numbers of research group projects and due to the nature of the  practical component. We will inform you as soon as timeslots are fixed or if anything changes during the semester.

Zoom link for lectures and presentations: https://auckland.zoom.us/j/99513913023? pwd=QTNSb0UzcHUvb290MjdrQlkwMHFEUT09 (https://auckland.zoom.us/j/99513913023? pwd=QTNSb0UzcHUvb290MjdrQlkwMHFEUT09)

Class reps

The class rep for this semester is Davy Yang (xyan289@aucklanduni.ac.nz

(mailto:xyan289@aucklanduni.ac.nz) ).

Class reps can act as an intermediary between students in the class and the lecturers and tutors.     You can share with them any suggestions/complaints/remarks about the lectures. The class reps are not a part of the teaching team.

Seeking Assistance

The primary source of assistance is the teaching staff. Please contact us  with any questions or        concerns about the course. We all are available via email. For help with more generic study skills or literacy, the Student Learning Centre and Library both offer many courses designed to help students become more efficient at study.

Piazza

We have set up Piazza for this course. The main purpose of Piazza is for you to interact with other students in the course; while lecturers will monitor Piazza and help if necessary, we believe that the best way for Piazza to work in this class is if you are all collectively responding to each       other's problems!

To encourage student responses, we as lecturers will follow a "24 hour" rule: during the first 24 hours of any post about the material in this course, we will not respond. (Note: this does not mean that we  will respond immediately after 24 hours! Depending on when your question goes up, we may be in    meetings, or it may be after work hours and we're trying to take care of our families, etc. In general    we'll get responses up as soon as is reasonable. If you haven't seen a response in two working days, please repost or email us.)

In addition, we reserve the right to remove posts what we feel are detrimental to the class. Please bear in mind that Piazza posts are not anonymous to the lecturing staff.

Exam

The final exam is worth 50% of your final mark. Please check Student Services Online (SSO) for the exam time and date. The exam is will be designed and conducted as online, non-invigilated, time-    limited examination. Provisional exam results can be obtained from SSO.

You will get your final grade via SSO.  Please also do understand that we are not allowed to be in communication with you in regards to your exam after the exam is written. If you email any of us during this period regarding the exam we won't be able to respond to your email.

If you would like to know more about exams process please see

exams/final-results.html (https://www.auckland.ac.nz/en/students/academic-information/exams- and-final-results/after-exams/final-results.html)

If you feel you need to talk to a person about the exam results, we suggest the science student center or the student adviser relevant to your degree.

people.html (https://www.auckland.ac.nz/en/science/about-the-faculty/school-of-computer- science/our-people.html)

Missed Exam

If you miss the exam for any valid reason, or you sit the exam but believe that your performance was impaired for some reason, then you may be able to apply for an aegrotat, compassionate or special  pass consideration. For more detailed information, contact the science student center.

Course Expectations

The document linked below outlines the School of Computer Science's philosophy of learning and teaching and our expectations for student engagement. Please read it carefully!

()

Academic Integrity

Sharing assignment solutions and source code does not help learning. Consequently, our academic  integrity policy does not permit sharing of solutions or source code leading to solutions. Violation of   this will result in your assignment submission attracting no marks, and you may face disciplinary        actions in addition. Therefore, please do not share assignments, assignment solutions and/or source code leading to assignment solutions. Do not publish or make your assignments or solutions              available in any form online, or you will be liable if someone copies your solution. Please come talk to us if you have any doubt over what is legit and what is not. You can refer to online tutorials and          resources. However, please learn from them and implement the solutions yourself based on what      you've learnt from those sources. Do not blindly copy from online sources. If there is a real need to    copy some code snippet, please ensure (a) you understand what the code snippet does, and (b) cite the source in a comment directly above the snippet. Don't leave your computers, devices, and            belongings unattended — you must secure these at all times to prevent anyone having access to       your assignments or solutions.

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 ()