Display name

title: Behrooz Omidvar-Tehrani

Username (this should match the folder name)

authors:

omidvar-tehrani

Is this the primary user of the site?

superuser: false

Enabling Interactivity between Human and Artificial Intelligence

Interactivity is to enable users to interact with their data and data systems effectively. It goes beyond efficiency measures and aims to make data usable for users. There exist different needs to interact with data and seek relevant information for an improved decision making process, such as shopping online, finding risky insured drivers, and building expert committees. In this talk, we discuss challenges and opportunities to substantiate interactivity in artificial intelligence applications. We begin by introducing the status quo, i.e., AI systems in which the user is left out of the loop, such as automated end-to-end ML pipelines, which lack personalization, customization, and explanation. Then we present a “guidance” approach in the form of a mixed-initiative system, where both human intelligence and artificial intelligence assist each other through iterative interactions to achieve higher levels of effectiveness. Next we describe how Reinforcement Learning helps model these interactions and how learned policies can augment the quality of interactions. Last, we discuss some results obtained so far and some future directions.

Role/position

Unknown

Behrooz Omidvar-Tehrani is a Research Scientist with a focus on interactive data systems. He has held positions in Naver Labs Europe, the Grenoble Alpes University, and the Ohio State University. His research interest is in the area of data management and at the crossroad of data mining, databases, and machine learning. He has published several papers in international conferences, such as VLDB, SIGMOD, and ICDE, and journals such as TKDE, VLDBJ, and FGCS.

Set this to `[]` or comment out if you are not using People widget.

user_groups:

Keynotes