Personalized learning environments requiring the elicitation of a student’s knowledge state have ... more Personalized learning environments requiring the elicitation of a student’s knowledge state have inspired researchers to propose distinct models to understand that knowledge state. Recently, the spotlight has shone on comparisons between traditional, interpretable models such as Bayesian Knowledge Tracing (BKT) and complex, opaque neural network models such as Deep Knowledge Tracing (DKT). Although DKT appears to be a powerful predictive model, little effort has been expended to dissect the source of its strength. We begin with the observation that DKT differs from BKT along three dimensions: (1) DKT is a neural network with many free parameters, whereas BKT is a probabilistic model with few free parameters; (2) a single instance of DKT is used to model all skills in a domain, whereas a separate instance of BKT is constructed for each skill; and (3) the input to DKT interlaces practice from multiple skills, whereas the input to BKT is separated by skill. We tease apart these three d...
Advances in Intelligent Systems and Computing, 2016
Understanding landslide risks is important for people living in hilly areas in India. A promising... more Understanding landslide risks is important for people living in hilly areas in India. A promising way of communicating landslide risks is via simulation tools, where these tools integrate both human factors (e.g., public investments to mitigate landslides) and environmental factors (e.g., spatial geology and rainfall). In this paper, we develop an interactive simulation model on landslide risks and use it to design a web-based Interactive Landslide Simulator (ILS) microworld. The ILS microworld is based on the assumption that landslides occur due to both environmental factors (spatial geology and rainfall) as well as human factors (lack of monetary investments to mitigate landslides). We run a lab-based experiment involving human participants performing in ILS and we show that the ILS performance helps improve public understanding of landslide risks. Overall, we propose ILS to be an effective tool for doing what-if analyses by policymakers and for educating public about landslide risks.
To investigate how differing amounts of experiential feedback and feedback s availability in an i... more To investigate how differing amounts of experiential feedback and feedback s availability in an interactive simulation tool influences people s decision-making against landslide risks. Feedback via simulation tools is likely to help people improve their decisions against disasters; however, currently little is known on how differing amounts of experiential feedback and feedback's availability in simulation tools influences people's decisions against landslides. We tested the influence of differing amounts of experiential feedback and feedback's availability on people's decisions against landslide risks in an Interactive Landslide Simulation (ILS) tool. In an experiment, in high-damage conditions, the probabilities of damages to life and property due to landslides were 10-times higher than those in the low-damage conditions. In feedback-present condition, experiential feedback was provided in numeric, text, and graphical formats in ILS. In feedback-absent conditions, the probabilities of damages were described; however, there was no experiential feedback present. Investments were greater in conditions where experiential feedback was present and damages were high compared to conditions where experiential feedback was absent and damages were low. Furthermore, only high-damage feedback produced learning in ILS. Experience gained in ILS enables people to improve their decision-making against landslide risks. Simulation tools seem appropriate for landslide risk communication and for performing what-if analyses. 1 Introduction Landslides cause massive damages to life and property worldwide (Chaturvedi and Dutt, 2015; Margottini et al., 2011). Knowledge about causes-and-consequences of landslides and awareness about landslide disaster mitigation are likely to help people take good mitigation actions that prevent landslides from occurring (Becker et al., 2013; Osuret et al., 2016; Webb and Ronan, 2014). However, to educate people about cause-and-effect relationships concerning landslides, effective landslide risk communication systems (RCSs) are needed (Glade et al., 2005). To be effective, these RCSs should possess five main components (Rogers and Tsirkunov, 2011): monitoring; analyzing; risk communication; warning dissemination; and, capacity building. Among these components, prior research has focused on monitoring and analyzing the occurrence of landslide events (Dai et al., 2002; Montrasio et al., 2011). For example, there exists various statistical and processbased models for predicting landslides (
Personalized learning environments requiring the elicitation of a student’s knowledge state have ... more Personalized learning environments requiring the elicitation of a student’s knowledge state have inspired researchers to propose distinct models to understand that knowledge state. Recently, the spotlight has shone on comparisons between traditional, interpretable models such as Bayesian Knowledge Tracing (BKT) and complex, opaque neural network models such as Deep Knowledge Tracing (DKT). Although DKT appears to be a powerful predictive model, little effort has been expended to dissect the source of its strength. We begin with the observation that DKT differs from BKT along three dimensions: (1) DKT is a neural network with many free parameters, whereas BKT is a probabilistic model with few free parameters; (2) a single instance of DKT is used to model all skills in a domain, whereas a separate instance of BKT is constructed for each skill; and (3) the input to DKT interlaces practice from multiple skills, whereas the input to BKT is separated by skill. We tease apart these three d...
Advances in Intelligent Systems and Computing, 2016
Understanding landslide risks is important for people living in hilly areas in India. A promising... more Understanding landslide risks is important for people living in hilly areas in India. A promising way of communicating landslide risks is via simulation tools, where these tools integrate both human factors (e.g., public investments to mitigate landslides) and environmental factors (e.g., spatial geology and rainfall). In this paper, we develop an interactive simulation model on landslide risks and use it to design a web-based Interactive Landslide Simulator (ILS) microworld. The ILS microworld is based on the assumption that landslides occur due to both environmental factors (spatial geology and rainfall) as well as human factors (lack of monetary investments to mitigate landslides). We run a lab-based experiment involving human participants performing in ILS and we show that the ILS performance helps improve public understanding of landslide risks. Overall, we propose ILS to be an effective tool for doing what-if analyses by policymakers and for educating public about landslide risks.
To investigate how differing amounts of experiential feedback and feedback s availability in an i... more To investigate how differing amounts of experiential feedback and feedback s availability in an interactive simulation tool influences people s decision-making against landslide risks. Feedback via simulation tools is likely to help people improve their decisions against disasters; however, currently little is known on how differing amounts of experiential feedback and feedback's availability in simulation tools influences people's decisions against landslides. We tested the influence of differing amounts of experiential feedback and feedback's availability on people's decisions against landslide risks in an Interactive Landslide Simulation (ILS) tool. In an experiment, in high-damage conditions, the probabilities of damages to life and property due to landslides were 10-times higher than those in the low-damage conditions. In feedback-present condition, experiential feedback was provided in numeric, text, and graphical formats in ILS. In feedback-absent conditions, the probabilities of damages were described; however, there was no experiential feedback present. Investments were greater in conditions where experiential feedback was present and damages were high compared to conditions where experiential feedback was absent and damages were low. Furthermore, only high-damage feedback produced learning in ILS. Experience gained in ILS enables people to improve their decision-making against landslide risks. Simulation tools seem appropriate for landslide risk communication and for performing what-if analyses. 1 Introduction Landslides cause massive damages to life and property worldwide (Chaturvedi and Dutt, 2015; Margottini et al., 2011). Knowledge about causes-and-consequences of landslides and awareness about landslide disaster mitigation are likely to help people take good mitigation actions that prevent landslides from occurring (Becker et al., 2013; Osuret et al., 2016; Webb and Ronan, 2014). However, to educate people about cause-and-effect relationships concerning landslides, effective landslide risk communication systems (RCSs) are needed (Glade et al., 2005). To be effective, these RCSs should possess five main components (Rogers and Tsirkunov, 2011): monitoring; analyzing; risk communication; warning dissemination; and, capacity building. Among these components, prior research has focused on monitoring and analyzing the occurrence of landslide events (Dai et al., 2002; Montrasio et al., 2011). For example, there exists various statistical and processbased models for predicting landslides (
Uploads
Papers by Akshit Arora