Few-shot learning (FSL) is suitable for plant-disease recognition due to the shortage of data. However, the limitations of feature representation and the demanding generalization requirements are still pressing issues that need to be addressed. The recent studies reveal that the frequency representation contains rich patterns for image understanding. Given that most existing studies based on image classification have been conducted in the spatial domain, we introduce frequency representation into the FSL paradigm for plant-disease recognition. A discrete cosine transform module is designed for converting RGB color images to the frequency domain, and a learning-based frequency selection method is proposed to select informative frequencies. As a post-processing of feature vectors, a Gaussian-like calibration module is proposed to improve the generalization by aligning a skewed distribution with a Gaussian-like distribution. The two modules can be independent components ported to other networks. Extensive experiments are carried out to explore the configurations of the two modules. Our results show that the performance is much better in the frequency domain than in the spatial domain, and the Gaussian-like calibrator further improves the performance. The disease identification of the same plant and the cross-domain problem, which are critical to bring FSL to agricultural industry, are the research directions in the future.

The rapid advancement of deep learning techniques has expedited the progress of data-driven forecasting methods for lithium-ion battery health. The conventional deep learning techniques for battery health forecasting require the training and refining of the predictive model in a centralized manner. However, centralized approaches face challenges related to data privacy and scalability. Therefore, it is essential to explore a decentralized methodology for the forecasting of battery health in order to safeguard privacy, utilize onboard computing resources, and facilitate the rapid integration of new data. This article proposes the utilization of federated learning to train a lithium-ion battery health forecasting model in a decentralized manner. All the experiments carried out in this study have been specifically customized to align with real-world conditions. A client selection strategy designed specifically for battery health forecasting is presented, which is demonstrated to increase accuracy throughout the training process. The evaluation results show that the predictive model trained in a decentralized manner exhibits comparable overall performance to the centralized counterpart.

Purpose: In recent years, studies have shown that machine learning significantly improves student performance and retention and reduces the risk of student dropout and withdrawal. However, there is a lack of empirical research reviews focusing on the application of machine learning to predict student performance in terms of learning engagement and self-efficacy and exploring their relationships. Hence, this paper conducts a systematic research review on the application of machine learning in higher education from an empirical research perspective.Design/methodology/approach: This systematic review examines the application of machine learning (ML) in higher education, focusing on predicting student performance, engagement and self-efficacy. The review covers empirical studies from 2016 to 2024, utilizing a PRISMA framework to select 67 relevant articles from major databases.Findings: The findings show that ML applications are widely researched and published in high-impact journals. The primary functions of ML in these studies include performance prediction, engagement analysis and self-efficacy assessment, employing various ML algorithms such as decision trees, random forests, support vector machines and neural networks. Ensemble learning algorithms generally outperform single algorithms regarding accuracy and other evaluation metrics. Common model evaluation metrics include accuracy, F1 score, recall and precision, with newer methods also being explored.Research limitations/implications: First, empirical research literature was selected from only four renowned electronic journal databases, and the literature was limited to journal articles, with the latest review literature and conference papers published in the form of conference papers also excluded, which led to empirical research not obtaining the latest views of researchers in interdisciplinary fields. Second, this review focused mainly on the analysis of student grade prediction, learning engagement and self-efficacy and did not study students’ risk, dropout rates, retention rates or learning behaviors, which limited the scope of the literature review and the application field of machine learning algorithms. Finally, this article only conducted a systematic review of the application of machine learning algorithms in higher education and did not establish a metadata list or carry out metadata analysis.Originality/value: The review highlights ML’s potential to enhance personalized education, early intervention and identifying at-risk students. Future research should improve prediction accuracy, explore new algorithms and address current study limitations, particularly the narrow focus on specific outcomes and lack of interdisciplinary perspectives.

Plant counting is an important part in precision agriculture (PA). The Unmanned Aerial Vehicle (UAV) becomes popular in agriculture because it can capture data with higher spatiotemporal resolution. When it is equipped with multispectral sensors, more meaningful multispectral data is obtained for plants’ analysis. After tobacco seedlings are raised, they are transplanted into the field. The counting of tobacco plant stands in the field is important for monitoring the transplant survival rate, growth situation, and yield estimation. In this work, we adopt the object detection (OD) method of deep learning to automatically count the plants with multispectral images. For utilizing the advanced YOLOv8 network, we modified the architecture of the network to adapt to the different band combinations and conducted extensive data pre-processing work. The Red + Green + NIR combination obtains the best detection results, which reveal that using a specific band or band combinations can obtain better results than using the traditional RGB images. For making our method more practical, we designed an algorithm that can handling the image of a whole plot, which is required to be watched. The counting accuracy is as high as 99.53%. The UAV, multispectral data combined with the powerful deep learning methods show promising prospective in PA.

A key challenge of visible-infrared person re-identification (VI-ReID) comes from the modality difference between visible and infrared images, which further causes large intra-person and small inter-person distances. Most existing methods design feature extractors and loss functions to bridge the modality gap. However, the unpaired-images constrain the VI-ReID model's ability to learn instance-level alignment features. Different from these methods, in this paper, we propose a pose attention-guided paired-images generation network (PAPG) from the standpoint of data augmentation. PAPG can generate cross-modality paired-images with shape and appearance consistency with the real image to perform instance-level feature alignment by minimizing the distances of every pair of images. Furthermore, our method alleviates data insufficient and reduces the risk of VI-ReID model overfitting. Comprehensive experiments conducted on two publicly available datasets validate the effectiveness and generalizability of PAPG. Especially, on the SYSU-MM01 dataset, our method accomplishes 7.76% and 5.87% gains in Rank-1 and mAP.

Accurate storm surge forecasting is vital for saving lives and avoiding economic and infrastructural damage. Failure to accurately predict storm surge can have catastrophic repercussions. Advances in machine learning models show the ability to improve accuracy of storm surge prediction by leveraging vast amounts of historical and realtime data such as weather and tide patterns. This paper proposes a bidirectional attention-based LSTM storm surge architecture (BALSSA) to improve prediction accuracy. Training and evaluation utilized extensive meteorological and tide level data from 77 typhoon incidents in Hong Kong and Macao between 2017 and 2022. The proposed methodology is able to model complex non-linearities between large amounts of data from different sources and identify complex relationships between variables that are typically not captured by traditional physical methods. BALSSA effectively resolves the problem of long-term dependencies in storm surge prediction by the incorporation of an attention mechanism. It enables selective emphasis on significant features and boosts the prediction accuracy. Evaluation has been conducted using real-world datasets from Macao to validate our storm surge prediction model. Results show that accuracy and robustness of predictions were significantly improved by the incorporation of attention mechanisms in our models. BALSSA captures temporal dynamics effectively, providing highly accurate storm surge forecasts (MAE: 0.0126, RMSE: 0.0003) up to 72 h in advance. These findings have practical significance for disaster risk reduction strategies, saving lives through timely evacuation and early warnings. Experiments comparing BALSSA variations with other machine learning algorithms consistently validate BALSSA’s superior predictive performance. It offers an additional risk management tool for civil-protection agencies and governments, as well as an ideal solution for enhancing storm surge prediction accuracy, benefiting coastal communities.