arvutiteaduse instituudi lõputööde teemade register


Validating models that require eXplainable Artificial Intelligence (XAI) at Swedbank.
Organisatsiooni nimiSwedbank
KokkuvõteAs AI's ground truth needs to be firmly established, models are often run with the help of a digital underclass and a digital overclass. Depending on the mundanity of the job, the members of the former are the 15 000 Facebook content moderators, while the latter includes doctors ensuring that the AI decisions are spot on, or bank officials overseeing the legitimacy of loan decisions that are undertaken or that Anti Money-Laundering (AML) signals are noise-free.

Banks are employing AI model creators, but also independent validators - who use statistical methods for ensuring that all the pertinent risks arising in their creation have been addressed.

In this Masters project, you will help us to push forward the introduction of more machine learning at Swedbank, by coming up with innovative ways of applying XAI for our models (by mastering SHAP, LIME, LRP)!

A remuneration will be offered to a suitable candidate for this project.
Lõputöö kaitsmise aasta2020-2021
JuhendajaJaan Übi
Suhtlemiskeel(ed)eesti keel, inglise keel
Nõuded kandideerijale
Tase Magister
Märksõnad
Kandideerimise kontakt
Nimi Jaan Übi
Tel
E-mail jaan.ubi@swedbank.ee


ati.study@lists.ut.ee