Title: Hybridised pre-trained deep network with Aspen-Lupus bidirectional long short-term memory classifier for image-based event classification

Authors: Shrikant P. Sanas; Tanuja Sarode

Addresses: Computer Engineering, Thadomal Shahani Engineering College, Mumbai, Maharashtra, 400050, India ' Computer Engineering, Thadomal Shahani Engineering College, Mumbai, Maharashtra, 400050, India

Abstract: The proposed Aspen-Lupus optimisation-based BiLSTM classifier (ALO opt BiLSTM) is employed in this research to develop an event classification model that accurately identifies the events. The pre-trained hybridised model, which is proposed for feature extraction, is developed via a conventional hybridisation of the VGG-16 and ResNet-101 models. The deep BiLSTM classifier gathers the collected features and utilises them to effectively increase prediction accuracy. The development of the proposed ALO algorithm resulted from the typical hybridisation of the Aspen and Lupus optimisation. Based on the achievements, at training percentage 90, the accuracy of 95.65%, sensitivity of 94.27%, specificity of 96.63% in database-1 respectively is attained and for database-2, achievements of 94.22% in accuracy, 92.86% insensitivity and 95.18% in specificity is acquired.

Keywords: event classification model; Aspen-Lupus optimisation; BiLSTM classifier; border collie; grey wolf; hybrid pre-trained model.

DOI: 10.1504/IJNVO.2024.138482

International Journal of Networking and Virtual Organisations, 2024 Vol.30 No.3, pp.282 - 307

Received: 06 Jul 2023
Accepted: 17 Sep 2023
Published online: 07 May 2024 *

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