Outgoing students
General information
Students can participate in three different Erasmus programs:
- Erasmus+ Study
- Erasmus+ Traineeship
- Erasmus Extra EU
To participate in the Erasmus Traineeship and Erasmus Extra EU programs, you have to identify the place where you would like to spend a period for the traineeship of at least two months in collaboration with a Professor of DISCo who works on an area of your interest and who can propose you the place (University/Company) where you can spend your Erasmus period.
Notes:
- For Erasmus+ Study (to attend classes and take exams) there is only one selection period during the year, usually in January.
- For Erasmus+ Traineeship there are two selection periods during the year, usually in February and July.
- For Erasmus Extra EU there are two selection periods during the year, usually in February and July.
Detailed information about the Erasmus programs and the calls for applications are available at the University portal.
Destinations for the current Accademic Year may be found in the document titled Destinations.
I suggest to contact the person who is offering the position only if you have a true interest in the topic of the traineeship together with the knowledge and skills that the position requires. Furthermore, when contacting the person offering the position also enclose your CV.
Institution: Aristotle University of Thessaloniki (AUTH)
Website: https://www.auth.gr/en/
Topic: Neurosymbolic AI for algorithmic fairness to support decision making in the healthcare domain
The traineeship is framed within the activity of a large European Project with the objective of making data and AI pipelines compliant by the design (https://datapact.eu/). AUTH DataPact team is working on approaches to support decisions about patient discharge by predicting their discharge date and a few other variables. To this end they are developing solutions based on different types of neural network architectures. However, considering the sensitive domain, it is important to ensure that the decision are fair towards disadvantaged groups. The approach that AUTH is considering together with the UNIMIB team is based on Logic Tensor Network (LTN), a neurosymbolic framework that has demonstrated state-of-the-art debiasing performance on several benchmark datasets. LTN integrates the power of deep neural networks with declarative knowledge expressed as logical formulas. In the application to the patient discharge problem, a number of challenges must be considered though: modelling patient data (from multiple sources and at different time scales), modeling different prediction tasks (regression, classification), integration with privacy-preserving frameworks. The student will acquire knowledge on application of machine learning in the healthcare from the AUTH team, who has substantial experience in this field, and about neurosymbolic approaches by UNIMIB team.
About AUTH and Thessaloniki:
AUTH is the largest university in greek, making Thessaloniki a student and business city on the sea with a vibrant culture.
contact: matteo.palmonari@unimib.it