Insect identification is crucial for agriculture, entomology, and ecological monitoring, where accurate pest detection can avoid crop damage and reduce pesticide use. To assure model transparency and dependability, this work suggests an improved method for automated insect categorization that combines explainable artificial intelligence (XAI) techniques with transfer learning. The main goal is to create a high-accuracy, easily deployable classification system with excellent interpretability by utilizing the ResNet152v2 architecture. Nine different classes of insects totaling 4509 photos were gathered and pre-processed for noise reduction, resolution standardization, and data normalization. The ResNet152v2 model was trained, and Grad-CAM (gradient-weighted class activation mapping) was applied to illustrate significant characteristics driving model decisions. With a 96% classification accuracy, the model proved useful in practical applications, minimizing the need for big labeled datasets while preserving transparency. Using this model in agricultural contexts can help farmers protect crops from pests, use less pesticides, and improve farmland fertility, among other practical uses. This system, which uses a web-based application, is accessible, inexpensive, and simple to use. It provides farmers with timely information so they may take preventive measures against pest infestations. This work contributes significantly to the field by providing a scalable and trustworthy solution for real-time insect identification, laying the groundwork for future AI-driven innovations in sustainable agriculture and ecological monitoring.

The growth of mobile financial services over the last decade is well accepted by the people of Bangladesh. However, although reported here and there, a complete understanding of illegal payments and fraudulent transactions is still missing. This paper studies one of the popular and experienced financial transaction services of Bangladesh, Bkash, to identify fraudulent text messages through the service. We propose an improved classification method for fraudulent messages through feature extraction and a hybrid deep learning (DL) approach with ML, which combines a convolutional neural network with sequential learning, referred to as the customized Bkash fraud semantic sequential learning (CBF-SSL) model. Seven machine-learning classifiers and five deep-learning sequential models are experimentally evaluated within the hybrid classifier. The performance of the fraud detection system is assessed using the loss function and confusion matrix on a dataset of 500 Bengali text messages. The performance evaluation demonstrates that the hybrid model of convolution neural network (CNN) using long short-term memory (LSTM) outperforms all other classifiers with the lowest error 1 rate of 0.2532%, the highest F1-measure of 75.00% with the highest test accuracy. The proposed text features in combination with CNN and LSTM prove to be highly effective in detecting fraudulent messages.

This paper suggests an approach to insect identification using transfer learning combined with an explainable artificial intelligence (XAI) technique. Thisstudy represents the importance of incorporating XAI to ensure the reliabilityand trustworthiness of automated insect identification systems. The proposedresearch influences the expanding body of knowledge on advanced AI techniquesfor biological classification, paving the way for future innovations in entomology, agriculture, and ecological monitoring. Transparency and interoperabilitymade possible by the integration of XAI enable a thorough comprehension ofthe decision-making process underlying the model’s predictions. The gradient-weighted class activation mapping, or Grad-CAM, approach is what we employ. Establishing transparency is crucial in fostering confidence in automated systemsand guaranteeing their pragmatic implementation in real-life situations. We proposed the ResNet152v2 model achieving a classification accuracy of 96% on acomprehensive dataset of 4,509 insect images, spanning 9 distinct classes. Thisapproach boosts model performance and reduces the dependency on extensivelabeled datasets.