National Institute of Education
National Institute of Education, Singapore: NIE Digital RepositoryNot a member yet
19917 research outputs found
Sort by
Mas Gift Agreement Event 2025 (5 Aug 2025)
No full textGroup photo with all MAS Gift Agreement event's guests
Exploratory investigation of electrodermal activity in learning from a large language model versus from curated texts
No full textThis paper reports a recent iteration of investigation in the science of learning, arising from a trajectory of work by the authors dating from 2021 at the intersection of neuroergonomics and data science. The work applies a frame of making and citizen science to the design of learning environments in which students seek to understand their own physiological responses as they participate in activities of learning in contexts authentic to themselves, as opposed to lab-based studies. In the present study, seven high school students were invited to learn about various topics through the reading of curated texts, as well as through interactions arising from prompts to a Large Language Model (LLM). During the activity, the participants wore non-intrusive sensors of electrodermal activity (EDA), as proxy measures of arousal and engagement. TVSymp, calculated using spectral powers of the EDA signal, has been found to correlate highly to orthostatic, cognitive, and physical stress. However, the findings from post-learning quiz results and EDA features do not suggest there to be any significant differences in effectiveness between learning from LLMs and curated texts. The present study serves as a small part of the body of literature related to AI in education to inform policy and to suggest ways forward as school leaders and teachers seek to navigate this evolving landscape.Accepted versio
A near black box parameter optimizer for NDDO-descendant semiempirical methods: Demonstrations for MNDO and AM1
No full textThe open access publication is available at https://doi.org/10.1021/acs.jpca.5c06861We provide a detailed description of an enhanced version of our previous geometry-corrected parameter optimization algorithm capable of accounting for reference geometries including all pertinent equations necessary for its implementation. The algorithm directly incorporates the derivatives of key lengths and angles, unlike PARAM, which relies on derived geometric functions. As a demonstration of the utility of our novel algorithm, reparameterizations for MNDO and AM1 using 1187 CHNO molecules in the PM7 training set are reported and compared to analogous results obtained with PARAM program used in the development of the PMx models; additional reparameterizations employing a partition of the full CHNO data set into training and validation subsets were also examined. Our AM1 reparameterizations achieve substantial improvements in the predicted geometrical properties (bond lengths, bond angles, and dihedral angles), demonstrating that the derived geometrical reference functions are ill-suited for parameter optimization and may also inadvertently incentivize smaller force constants for chemical bonds; however, this arises at the cost of decreased accuracies for reaction barrier heights, which is viewed to be a consequence of the absence of nonequilibrium geometries from the training set. Together, our results suggest that judicious parameter refinement could substantially enhance the performance of NDDO-descendant semiempirical models
