Machine Learning with Limited Data (2023)

When using and developing Machine Learning methods, one is frequently faced with scenarios in which the training data available is highly limited. For instance, in Natural Language Processing (NLP), it is common that there might be rich amounts of data for certain languages (e.g., English, Spanish, Chinese) but much less for others (e.g., Danish, Faroese, Haitian Creole). Importantly, however, when faced with limited data there is often rich domain knowledge available. For instance, when classifying medical images, there is an abundance of knowledge from the medical field that can be used, and when dealing with natural languages, the field of linguistics has an abundance of insights that can be used. Whereas many methods for transfer learning simply ignore such domain knowledge, this course highlights the advantages of using it, and the necessity of interdisciplinary collaboration, rather than simply addressing data points without the appropriate domain context.  

This course aims to: (i) provide an overview of methods for dealing with limited data in machine learning settings; (ii) provide concrete interactive code examples for participants to develop and potentially use for their own research; and (iii) highlight the importance of interdisciplinary collaboration and discussion with domain experts.  

As everyone speaks a language, the examples used in the course will focus on NLP, and participants will be challenged to use both their own intuitions of language as well as domain expertise provided by a world-leading linguist to develop and analyze ML/NLP systems. 

Organizer: Johannes Bjerva 

Lecturers:  Johannes Bjerva (AAU)

ECTS:          3

Time:          March 2023 
                    8 - 11 May 2023

                    26-29 September 2023

Place:          Hybrid from Copenhagen
     A. C. Meyers Vænge 15, building A
            Room 1.008


Zip code: 2450

City:             Copenhagen

Number of seats:   50

Deadline:                September 2023


Important information concerning PhD courses:


We have over some time experienced problems with no-show for both project and general courses. It has now reached a point where we are forced to take action. Therefore, the Doctoral School has decided to introduce a no-show fee of DKK 3.000 for each course where the student does not show up. Cancellations are accepted no later than 2 weeks before start of the course. Registered illness is of course an acceptable reason for not showing up on those days. Furthermore, all courses open for registration approximately four months before start. This can hopefully also provide new students a chance to register for courses during the year. We look forward to your registrations.