scholarly journals Interactive Policy Learning through Confidence-Based Autonomy

2009 ◽  
Vol 34 ◽  
pp. 1-25 ◽  
Author(s):  
S. Chernova ◽  
M. Veloso
Keyword(s):  
Task Performance ◽  
Action Selection ◽  
Learning Performance ◽  
Policy Learning ◽  
Learning From Demonstration ◽  
Simulated Driving ◽  
Interactive Algorithm ◽  
Future Task ◽  
Algorithmic Component ◽  
Human Teacher

We present Confidence-Based Autonomy (CBA), an interactive algorithm for policy learning from demonstration. The CBA algorithm consists of two components which take advantage of the complimentary abilities of humans and computer agents. The first component, Confident Execution, enables the agent to identify states in which demonstration is required, to request a demonstration from the human teacher and to learn a policy based on the acquired data. The algorithm selects demonstrations based on a measure of action selection confidence, and our results show that using Confident Execution the agent requires fewer demonstrations to learn the policy than when demonstrations are selected by a human teacher. The second algorithmic component, Corrective Demonstration, enables the teacher to correct any mistakes made by the agent through additional demonstrations in order to improve the policy and future task performance. CBA and its individual components are compared and evaluated in a complex simulated driving domain. The complete CBA algorithm results in the best overall learning performance, successfully reproducing the behavior of the teacher while balancing the tradeoff between number of demonstrations and number of incorrect actions during learning.

scholarly journals AutoQual: task-oriented structural vibration sensing quality assessment leveraging co-located mobile sensing context

2021 ◽  
Author(s):  
Yue Zhang ◽  
Zhizhang Hu ◽  
Susu Xu ◽  
Shijia Pan
Keyword(s):  
Task Performance ◽  
Expert Knowledge ◽  
Vibration Signal ◽  
Absolute Error ◽  
Mobile Sensing ◽  
Learning Performance ◽  
Structural Vibration ◽  
Vibration Sensing ◽  
Assessment Approach ◽  
Task Oriented

AbstractIn this paper, we introduce AutoQual, a mobile-based assessment scheme for infrastructure sensing task performance prediction under new deployment environments. With the growth of the Internet-of-Things (IoT), many non-intrusive sensing systems have been explored for various indoor applications, such as structural vibration sensing. This indirect sensing approach’s learning performance is prone to deployment variance when signals propagate through the environment. As a result, current systems heavily rely on expert knowledge and manual assessment to achieve effective deployments and high sensing task performance. In order to mitigate this expert effort, we propose to systematically study factors that reflect deployment environment characteristics and methods to measure them autonomously. We present AutoQual that measures a series of assessment factors (AFs) reflecting how the deployment environment impacts the system performance. AutoQual outputs a task-oriented sensing quality (TSQ) score by integrating measured AFs trained from known deployments as a prediction of untested system’s performance. In addition, AutoQual achieves this assessment without manual effort by leveraging co-located mobile sensing context to extract structural vibration signal for processing automatically. We evaluate AutoQual by using it to predict untested systems’ performance over multiple sensing tasks. We conduct real-world experiments and investigate 48 deployments in 11 environments. AutoQual achieves less than 0.10 average absolute error when auto-assessing multiple tasks at untested deployments, which shows a $$\le 0.018$$ ≤ 0.018 absolute error difference compared to the manual assessment approach.

scholarly journals Falls Risk and Simulated Driving Performance in Older Adults

2013 ◽  
Vol 2013 ◽  
pp. 1-8 ◽  
Author(s):  
John G. Gaspar ◽  
Mark B. Neider ◽  
Arthur F. Kramer
Keyword(s):  
Older Adults ◽  
Task Performance ◽  
Dual Task ◽  
Response Times ◽  
Cognitive Task ◽  
Driving Performance ◽  
Unexpected Events ◽  
Falls Risk ◽  
Simulated Driving ◽  
Dual Task Performance

Declines in executive function and dual-task performance have been related to falls in older adults, and recent research suggests that older adults at risk for falls also show impairments on real-world tasks, such as crossing a street. The present study examined whether falls risk was associated with driving performance in a high-fidelity simulator. Participants were classified as high or low falls risk using the Physiological Profile Assessment and completed a number of challenging simulated driving assessments in which they responded quickly to unexpected events. High falls risk drivers had slower response times (~2.1 seconds) to unexpected events compared to low falls risk drivers (~1.7 seconds). Furthermore, when asked to perform a concurrent cognitive task while driving, high falls risk drivers showed greater costs to secondary task performance than did low falls risk drivers, and low falls risk older adults also outperformed high falls risk older adults on a computer-based measure of dual-task performance. Our results suggest that attentional differences between high and low falls risk older adults extend to simulated driving performance.

scholarly journals Group social rank is associated with performance on a spatial learning task

2018 ◽  
Vol 5 (2) ◽  
pp. 171475 ◽  
Author(s):  
Ellis J. G. Langley ◽  
Jayden O. van Horik ◽  
Mark A. Whiteside ◽  
Joah R. Madden
Keyword(s):  
Task Performance ◽  
Social Rank ◽  
Learning Task ◽  
Learning Ability ◽  
Learning Performance ◽  
Phasianus Colchicus ◽  
Dyadic Relationships ◽  
Social Hierarchies ◽  
The Social ◽  
Motivation To Participate

Dominant individuals differ from subordinates in their performances on cognitive tasks across a suite of taxa. Previous studies often only consider dyadic relationships, rather than the more ecologically relevant social hierarchies or networks, hence failing to account for how dyadic relationships may be adjusted within larger social groups. We used a novel statistical method: randomized Elo-ratings, to infer the social hierarchy of 18 male pheasants, Phasianus colchicus , while in a captive, mixed-sex group with a linear hierarchy. We assayed individual learning performance of these males on a binary spatial discrimination task to investigate whether inter-individual variation in performance is associated with group social rank. Task performance improved with increasing trial number and was positively related to social rank, with higher ranking males showing greater levels of success. Motivation to participate in the task was not related to social rank or task performance, thus indicating that these rank-related differences are not a consequence of differences in motivation to complete the task. Our results provide important information about how variation in cognitive performance relates to an individual's social rank within a group. Whether the social environment causes differences in learning performance or instead, inherent differences in learning ability predetermine rank remains to be tested.

Effects of alcohol on attention orienting and dual-task performance during simulated driving: An event-related potential study

2010 ◽  
Vol 24 (9) ◽  
pp. 1333-1348 ◽  
Author(s):  
Anne E Wester ◽  
Joris C Verster ◽  
Edmund R Volkerts ◽  
Koen BE Böcker ◽  
J. Leon Kenemans
Keyword(s):  
Task Performance ◽  
Dual Task ◽  
Event Related Potential ◽  
Simulated Driving ◽  
Attention Orienting ◽  
Potential Study ◽  
Dual Task Performance

Perception in Consequences of Free Riding

2002 ◽  
Vol 90 (1) ◽  
pp. 215-225 ◽  
Author(s):  
Jeffrey A. Miles ◽  
Howard J. Klein
Keyword(s):  
College Students ◽  
Task Performance ◽  
Ad Hoc ◽  
Group Performance ◽  
Individual Performance ◽  
Free Riding ◽  
The Other ◽  
Group Outcomes ◽  
Group Members ◽  
Future Task

This study examined the relationships between perceptions of group members' free riding and group outcomes using Mulvey and Klein's 1998 perceived free riding scale. In a laboratory study, three free riding conditions were created (no free riding, free riding, free riding with justification) in which 97 college students performed two short number-finding tasks as members of temporary ad hoc three-person groups. 55% of the students were male and the average age was 22.9 yr. ( SD = 3.0). Participants' perceptions of free riding were negatively related to commitment to the assigned group goal, task performance, and goals for group performance and individual performance. In the condition wherein free riding was justified by low ability, participants set lower goals for their future task performance than did those in the other two conditions.

scholarly journals Dual-task Interference in a Simulated Driving Environment: Serial or Parallel Processing?

2019 ◽  
Author(s):  
Mojtaba Abbas-Zadeh ◽  
Gholam-Ali Hossein-Zadeh ◽  
Maryam Vaziri-Pashkam
Keyword(s):  
Task Performance ◽  
Dual Task ◽  
Task Condition ◽  
Desktop Computer ◽  
Task Interference ◽  
Dual Task Condition ◽  
Simulated Driving ◽  
Evidence Accumulation ◽  
Dual Task Performance ◽  
Dual Task Interference

AbstractWhen humans are required to perform two tasks concurrently, their performances decrease as the two tasks get closer together in time. This effect is known as dual-task interference. This limitation of the human brain could have lethal effects during demanding everyday tasks such as driving. Are the two tasks processed serially or in parallel during dual-task performance in naturalistic settings? Here, we investigated dual-task interference in a simulated driving environment and investigated the serial/parallel nature of processing during dual-task performance. Participants performed a lane change task on a desktop computer, along with an image discrimination task. We systematically varied the time difference between the onset of the two tasks (Stimulus Onset Asynchrony, SOA) and measured its effect on the amount of dual-task interference. Results showed that the reaction times (RTs) of two tasks in the dual-task condition were higher than those in the single-task condition. SOA influenced RTs of both tasks when they were presented second and the RTs of the image task when it was presented first. Manipulating the predictability of the order of the two tasks, we showed that unpredictability attenuated the effect of SOA by changing the order of the response to the two tasks. Next, using drift-diffusion modeling, we modeled the reaction time and choice of the subjects during dual-task performance in both predictable and unpredictable task order conditions. The modeling results indicated that performing two tasks concurrently, affects both the rate of evidence accumulation and the delays outside the evidence accumulation period, suggesting that the two tasks are performed in a partial-parallel manner. These results extend the findings of previous dual-task experiments to more naturalistic settings and deepen our understanding of the mechanisms of dual-task interference.

Event-related potentials and secondary task performance during simulated driving

2008 ◽  
Vol 40 (1) ◽  
pp. 1-7 ◽  
Author(s):  
A.E. Wester ◽  
K.B.E. Böcker ◽  
E.R. Volkerts ◽  
J.C. Verster ◽  
J.L. Kenemans
Keyword(s):  
Task Performance ◽  
Secondary Task ◽  
Event Related Potentials ◽  
Simulated Driving ◽  
Related Potentials

Coas: Combined Object-Action Selection: A Human Factors Experiment

1992 ◽  
Vol 36 (4) ◽  
pp. 316-320 ◽  
Author(s):  
J. F. Kelley ◽  
J. Ukelson
Keyword(s):  
Task Performance ◽  
Action Selection ◽  
Complex Task ◽  
Interaction Techniques ◽  
Simple Task ◽  
File Management ◽  
Standard Object ◽  
Management Application ◽  
Action Model ◽  
High Level

12 participants with a high level of domain experience used two different, mouse-based, interaction techniques to carry out three workstation file management tasks of varying complexity. One technique followed a standard Object-Action model; the other was a newly developed technique called COAS (Combined Object-Action Selection). There was little difference in performance on a simple task; performance for participants using the new technique was 38% faster on a moderately complex task and was 21% faster on a complex task. The file management application, interaction techniques and experiment were implemented in an OS/2 Presentation Manager style using ITS (Interactive Transaction System).

Sign in / Sign up

Export Citation Format

Share Document