Food production is a growing challenge with the increasing global population. To increase the yield of food production, we need to adopt new biotechnology-based fertilization techniques. Furthermore, we need to improve early prevention steps against plant disease. Guava is an essential fruit in Asian countries such as Pakistan, which is fourth in its production. Several pathological and fungal diseases attack guava plants. Furthermore, postharvest infections might result in significant output losses. A professional opinion is essential for disease analysis due to minor variances in various guava disease symptoms. Farmers’ poor usage of pesticides may result in financial losses due to incorrect diagnosis. Computer-vision-based monitoring is required with developing field guava plants. This research uses a deep convolutional neural network (DCNN)-based data enhancement using color-histogram equalization and the unsharp masking technique to identify different guava plant species. Nine angles from 360${}^{\circ }$ were applied to increase the number of transformed plant images. These augmented data were then fed as input into state-of-the-art classification networks. The proposed method was first normalized and preprocessed. A locally collected guava disease dataset from Pakistan was used for the experimental evaluation. The proposed study uses five neural network structures, AlexNet, SqueezeNet, GoogLeNet, ResNet-50, and ResNet-101, to identify different guava plant species. The experimental results proved that ResNet-101 obtained the highest classification results, with 97.74% accuracy.

The rapid expansion of the Internet of Things (IoT) has led to the widespread adoption of sensor networks, with Long-Range Wide-Area Networks (LoRaWANs) emerging as a key technology due to their ability to support long-range communication while minimizing power consumption. However, optimizing network performance and energy efficiency in dynamic, large-scale IoT environments remains a significant challenge. Traditional methods, such as the Adaptive Data Rate (ADR) algorithm, often fail to adapt effectively to rapidly changing network conditions and environmental factors. This study introduces a hybrid approach that leverages Deep Learning (DL) techniques, namely Long Short-Term Memory (LSTM) networks, and Machine Learning (ML) techniques, namely Artificial Neural Networks (ANNs), to optimize key network parameters such as Signal-to-Noise Ratio (SNR) and Received Signal Strength Indicator (RSSI). LSTM-ANN model trained on the “LoRaWAN Path Loss Dataset including Environmental Variables” from Medellín, Colombia, and the model demonstrated exceptional predictive accuracy, achieving an R2 score of 0.999, Mean Squared Error (MSE) of 0.041, Root Mean Squared Error (RMSE) of 0.203, and Mean Absolute Error (MAE) of 0.167, significantly outperforming traditional regression-based approaches. These findings highlight the potential of combining advanced ML and DL techniques to address the limitations of traditional optimization strategies in LoRaWAN. By providing a scalable and adaptive solution for large-scale IoT deployments, this work lays the foundation for real-world implementation, emphasizing the need for continuous learning frameworks to further enhance energy efficiency and network resilience in dynamic environments.

Accurate path loss prediction is essential for optimizing Long-Range Wide-Area Network (LoRaWAN) performance. Previous studies have employed various Machine Learning (ML) models for path loss prediction. However, environmental factors such as temperature, humidity, barometric pressure, and particulate matter have been largely neglected. This study bridges this gap by evaluating the performance of five boosting ML models—AdaBoost, XGBoost, LightGBM, GentleBoost, and LogitBoost—under dynamic environmental conditions. The models were compared with theoretical models (Log-Distance and Okumura-Hata) and existing studies that employed the same dataset based on metrics such as RMSE, MAE, and R². Furthermore, a detailed performance vs. complexity analysis was conducted using metrics such as training time, inference latency, model size, and energy consumption. Notably, barometric pressure emerged as the most influential environmental factor affecting path loss across all models. Bayesian Optimization was applied to fine-tune hyperparameters to improve model accuracy. Results showed that LightGBM outperformed other models with the lowest RMSE of 0.5166 and the highest R² of 0.7151. LightGBM also offered the best trade-off between accuracy and computational efficiency. The findings show that boosting algorithms, particularly LightGBM, are highly effective for path loss prediction in LoRaWANs.

Businesses are overgrowing worldwide; people struggle for their businesses and startups in almost every field of life, whether industrial or academic. The businesses or services have multiple income streams with which they generate revenue. Most companies use different marketing and advertisement strategies to engage their customers and spread their services worldwide. Service recommendation systems are gaining popularity to recommend the best services and products to customers. In recent years, the development of service-oriented computing has had a significant impact on the growth of businesses. Knowledge graphs are commonly used data structures to describe the relations among data entities in recommendation systems. Domain-oriented user and service interaction knowledge graph (DUSKG) is a framework for keyword extraction in recommendation systems. This paper proposes a novel method of chunking-based keyword extractions for hybrid recommendations to extract domain-specific keywords in DUSKG. We further show that the performance of the hybrid approach is better than other techniques. The proposed chunking method for keyword extraction outperforms the existing value feature entity extraction (VF2E) by extracting fewer keywords.