Title: Design of a bioinspired augmented model for prediction of LD students via online behaviour analysis

Authors: Masooda Modak; Prachi Gharpure; Sasikumar Mukundan

Addresses: S.P.I.T., Bhavan's Campus, Andheri, Mumbai, India; Information Technology Department, Mukesh Patel School of Technology Management and Engineering, SVKM's NMIMS, Mumbai, India ' SVKM's NMIMS University, Indore Campus, Gandhi Nagar, Indore, India ' Centre for Development of Advanced Computing (CDAC), Juhu, Mumbai, India

Abstract: Identification of students with learning disabilities (LD) requires analysis of various parameters, including student's analytical quotient, logical reasoning, mathematical analysis and Language processing capabilities, etc. Models like convolutional neural networks (CNN), recurrent NNs (RNNs), etc., are unable to identify micro patterns from the input dataset, which limits their context-specific performance and applicability. To overcome this limitation, a gamification-based analysis model using augmented bioinspired computing is proposed. The proposed model initially collects a large dataset from both LD and non-LD students to train a genetic algorithm (GA)-based feature selection model. These features were evaluated for multiple logical and analytical categories. Based on this analysis, a particle swarm optimiser (PSO) was deployed, assisting in the selection of classifier configurations to identify LD students. The model was tested on a custom manual dataset of over 1,000 students in grades 6th, 7th, 8th. Its accuracy performance was observed to be 6.5% better, while precision and recall were observed to be 5.4% and 3.9% better when compared with various state-of-the-art methods respectively.

Keywords: learning disability; genetic algorithm; particle swarm optimiser; PSO; identification; research design and utilisation; convolutional neural networks; CNN.

DOI: 10.1504/IJBIC.2024.142554

International Journal of Bio-Inspired Computation, 2024 Vol.24 No.4, pp.240 - 251

Received: 21 Feb 2023
Accepted: 25 Mar 2024
Published online: 08 Nov 2024 *

Full-text access for editors Full-text access for subscribers Purchase this article Comment on this article