Machine Learning - Spring 2018

Name of course:
Machine Learning 


ECTS credits:
The course will be offered in two versions: one where assignments are not mandatory (3 ECTS points) and another version where assignments are mandatory (4 ECTS). 


Course parameters:
Language: English.
Level of course: PhD course.
Time of year: Spring semester, 2018.
No. of contact hours/hours in total incl. preparation, assignment(s) or the like:
Preparation (reading): 45 hours
Self-study lectures: 5 hours
Assignments: 25 hours
Lectures/sessions: 7 sessions of 4 hours = 28 hours
Capacity limits: 10 participants 


Objectives of the course:
The main point of the course is to learn about the most effective machine learning techniques and gain experience applying/implementing them. The materials studied provide a broad introduction to machine learning, data-mining and statistical pattern recognition. The following topics will be covered: 

·          Supervised learning
·          Unsupervised learning
·          Best machine learning practices
·          Case studies and applications (e.g. web search and anti-spam) 


Learning outcomes and competences:
At the end of the course, the student should be able to

·          Explain the key concepts of supervised and unsupervised learning
·          Explain the best machine learning practices
·          Apply and implement machine learning techniques 


Compulsory programme:
The course covers 13 weeks, divided into seven sessions. There are two weeks between the different sessions, except for session 7 which is only a week from session 6. The course plan is shown below. 

The course participants are expected to read the relevant materials provided prior to a given session. Furthermore, video lectures marked as ”self-study” must be watched by the participants independently (i.e. they are not included in the sessions). 


Course contents:
The following assignments will be covered: 

·          A1: “Linear Regression”
·          A2: “Logistic Regression”
·          A3: “Multi-class Classification and Neural Networks”
·          A4: “Neural Network Learning”
·          A5: “Regularized Linear Regression and Bias/Variance”
·          A6: “Support Vector Machines”
·          A7: “K-Means Clustering and PCA”
·          A8: “Anomaly Detection and Recommender Systems” 

For overview of sessions see HERE 


Prerequisites:
Programming experience and basic understanding of statistics and linear algebra. 


Name of lecturers:
The course follows an on-line course. 

 

Type of course/teaching methods:
The course follows the structure of an existing machine learning course offered by Coursera, which provides most of the materials that are needed (video lectures, assignments etc.) to run the course. The particpants will meet and watch the Coursera video lectures and work on assignments. The link to the Coursera course is https://www.coursera.org/learn/machine-learning


Literature:
The literature is available on the Coursera course website: https://www.coursera.org/learn/machine-learning 


Course homepage:
https://github.com/peterwvj/au-machine-learning-2018

 

Course assessment:
Assessment is based on active participation. 

 

Provider:
Department of Engineering. 

 

Special comments on this course:
None. 

 

Time:
We meet bi-weekly (except for a single session, see the course plan above) for 4 hours to watch the video lectures and work on assignments. This time slot does not include the time spent on reading/self-study/completing assignments. However, the course does permit a single special sessions for local needs (see session 6). This session may cover new topics (presented by a guest lecturer, for example), or existing topics/assignments that are considered difficult. 


Place:
TBD. 


Registration:
Deadline for registration is 5 days before the course starts. Information regarding admission will be sent out no later than 2 days later. 

For registration: email sha@eng.au.dk

If you have any questions, please contact Stefan Hallerstede. e-mail: sha@eng.au.dk