소식

이벤트

세미나 University of Glasgow, Frank Pollick 교수 강연

페이지 정보

작성자 최고관리자 작성일 22-09-19 17:17

본문

날짜 : 2022년 9월 21일 수요일 15시
장소 : 제4공학관 D510

제목: Calibrating trust between humans and autonomous systems
 

Abstract:As machines become increasingly more intelligent, they become more capable of operating with greater degrees of independence from their users. However, optimal use of these autonomous systems is dependent on appropriate trust from their users. A lack of trust towards an autonomous system will likely lead to the user doubting the capabilities of the system, potentially to the point of disuse. Conversely, too much trust in a system may lead to the user overestimating the capabilities of the system, and potentially result in errors which could have been avoided with proper collaboration. Thus, appropriate trust is trust which is calibrated to reflect the true capabilities of the system. The calibration of trust towards autonomous systems is an area of research of increasing popularity, as more and more intelligent machines are introduced to modern workplaces. In this talk I present the results of two human user experiments that examined trust calibration derived from human interaction with an autonomous image classifier. In the first experiment, we were able to examine the ways that participants placed trust in the classifier during different types of system performance. We also investigated whether users’ trust could be better calibrated by providing different displays of System Confidence Information, to help convey the system’s decision making. In our second experiment we additionally provided participants with another cue of system decision making, Gradient-weighted Class Activation Mapping, which indicated regions the classifier found salient to make its identification. We investigated whether this additional cue could promote greater trust towards the classifier and improve participants’ subjective understanding of the system’s decision making, as a way of exploring how to improve the interpretability of these systems. Results of the two experiments showed that while users preferred interfaces that appeared more rich in information this did not necessarily improve trust calibration. Users also reported greater understanding of the classifier’s decision making when provided with the Gradient-weighted Class Activation Mapping cue. This research contributes to our current understanding of calibrating users’ trust towards autonomous systems, and may be useful when designing Autonomous Image Classifier Systems.


Bio: Prof.Frank Pollick School of Psychology & Neuroscience University of Glasgow 

첨부파일

댓글목록

등록된 댓글이 없습니다.