Hidden feedback loops in machine learning systems: a simulation model and preliminary results

Author: Anton Khritankov

In this concept paper, we explore some of the aspects of quality of continuous learning artificial intelligence systems as they interact with and influence their environment. We study an important problem of implicit feedback loops that occurs in recommendation systems, web bulletins and price estimation systems. We demonstrate how feedback loops intervene with user behavior on an exemplary housing prices prediction system. Based on a preliminary model, we highlight sufficient existence conditions when such feedback loops arise and discuss possible solution approaches.

Vorgetragen von: Anton Khritankov
Unternehmen: Moscow Institute of Physics and Technology

Vortragssprache: Englisch
Level: Fortgeschrittene

Partner der Konferenz 2020

ASQF e.V ATB - Austrian Testing Board Blekinge Institute of Technology CON.ECT Eventmanagement dpunkt.verlag GmbH Fortiss GmbH Heise Medien GmbH & Co. KG IREB GmbH iSQI GmbH IT Verlag für Informationstechnik GmbH IT-Visions.de Österreichische Computer Gesellschaft (OCG) Software Quality Lab GmbH TU Wien, Inst f. Information Systems Eng., CDL-SQ Verband Österreichischer Software Industrie (VÖSI)