Location

Online

Event Website

https://hicss.hawaii.edu/

Start Date

3-1-2023 12:00 AM

End Date

7-1-2023 12:00 AM

Description

Models that capture the heterogeneous perspectives of individuals are essential to test tailored interventions, such as behavior change interventions. Although Fuzzy Cognitive Maps (FCMs) have a rich history in depicting systems, they were either developed at an individual level through facilitated sessions, or created for an entire population through machine learning. The need to automatically create individual FCMs from data has started to be addressed, but the proposed solution was computationally prohibitive and thus could not be deployed over a large population. In this work, we use a state-of-the-art evolutionary algorithm (CMA-ES) to create individual FCMs by leveraging the growing availability of longitudinal data. We demonstrate on a real-world case study that our solution is both accurate and fast to compute. Our experiments on synthetic data also show that our approach can scale to a large number of measurements, but it cannot currently be applied to highly noisy datasets.

Share

COinS
 
Jan 3rd, 12:00 AM Jan 7th, 12:00 AM

Fast Generation of Heterogeneous Mental Models from Longitudinal Data by Combining Genetic Algorithms and Fuzzy Cognitive Maps

Online

Models that capture the heterogeneous perspectives of individuals are essential to test tailored interventions, such as behavior change interventions. Although Fuzzy Cognitive Maps (FCMs) have a rich history in depicting systems, they were either developed at an individual level through facilitated sessions, or created for an entire population through machine learning. The need to automatically create individual FCMs from data has started to be addressed, but the proposed solution was computationally prohibitive and thus could not be deployed over a large population. In this work, we use a state-of-the-art evolutionary algorithm (CMA-ES) to create individual FCMs by leveraging the growing availability of longitudinal data. We demonstrate on a real-world case study that our solution is both accurate and fast to compute. Our experiments on synthetic data also show that our approach can scale to a large number of measurements, but it cannot currently be applied to highly noisy datasets.

https://aisel.aisnet.org/hicss-56/da/soft_computing/5