Explore the vast range of titles available.

Route choice modelling is a critical aspect of analysing urban rail transit (URT) networks and provides a foundation for URT planning and operation. Unlike in a free-flow road network, the consideration set for route choice decisions in a URT network does not depend purely on the physical connectivity of the network and decision makers’characteristics. Instead, it is also contingent on the train schedules. This paper delves into the evolution of research on route choices in URT networks, encompassing both probabilistic route choice modelling derived from utility maximisation theory and logit curve with physical connectivity, and retrospective route choice modelling based on travel time chaining along with comprehensive transport data. The former is noted for its conciseness, simplicity, and interpretability in real-world applications, even though the methodologies may not be cutting-edge. The latter incorporates dynamic temporal information to understand activities of passengers in URT networks. Enhancements of each genres are also examined. However, these improvements might not fully address the inherent limitations of models relating to a dependency on the quality of parameters, experience of experts, and calculation efficiency. In addition, novel research adopting contemporary data mining techniques instead of classical models are introduced. The historical development of research on URT network route choices underscores the importance of amalgamating independent information networks such as surveillance networks and social networks to establish a comprehensive multi-dimensional network. Such an approach integrates passenger attributes across networks, offering a multi-dimensional understanding of passengers’ route choice behaviours. Our review work aims to present not only a systematic conceptual framework for route choices in URT networks but also a novel path for transport researchers and practitioners to decipher the travel behaviours of passengers.

With the rapid advancement of intelligent technologies and network environments, the efficient and accurate handling of uncertain decision-making information has become an urgent challenge. Traditional methods often struggle to process complex and incomplete information, especially in cyber defense. To address this, we introduce discrete multi-dimensional Z-numbers (MZs) as a mathematical tool for modeling uncertainty and reliability in network defense decisions. This paper proposes a synthesis method for MZs, enabling the integration of multi-source information while considering both uncertainty and reliability. By leveraging a hidden probability model, we extend MZs into multi-dimensional Z + -numbers, enhancing their expressiveness in handling uncertainty. Furthermore, we define utility functions based on MZs and develop a multi-attribute group decision-making framework tailored for network defense. This approach offers a novel perspective for designing strategies against highly adaptive and covert cyberattacks. The proposed method is validated through a case study on the network security assessment of an intelligent logistics company. Results demonstrate significant improvements in the accuracy and efficiency of decision-making, highlighting the method’s advantages and broad potential in cyber defense. Beyond logistics, this integrated MZ -based decision framework provides an adaptable and intelligent tool for strengthening network security defenses.

The pervasiveness of Web 2.0 and social networking sites has enabled people to interact with each other easily through various social media. For instance, popular sites like Del.icio.us, Flickr, and YouTube allow users to comment on shared content (bookmarks, photos, videos), and users can tag their favorite content. Users can also connect with one another, and subscribe to or become a fan or a follower of others. These diverse activities result in a multi-dimensional network among actors, forming group structures with group members sharing similar interests or affiliations. This work systematically addresses two challenges. First, it is challenging to effectively integrate interactions over multiple dimensions to discover hidden community structures shared by heterogeneous interactions. We show that representative community detection methods for single-dimensional networks can be presented in a unified view. Based on this unified view, we present and analyze four possible integration strategies to extend community detection from single-dimensional to multi-dimensional networks. In particular, we propose a novel integration scheme based on structural features. Another challenge is the evaluation of different methods without ground truth information about community membership. We employ a novel cross-dimension network validation (CDNV) procedure to compare the performance of different methods. We use synthetic data to deepen our understanding, and real-world data to compare integration strategies as well as baseline methods in a large scale. We study further the computational time of different methods, normalization effect during integration, sensitivity to related parameters, and alternative community detection methods for integration.

This paper proposes a novel secure and fast multiple image encryption technique to encrypt multiple images of arbitrary sizes. In the proposed technique, a group of images is divided into non-overlapping blocks of size 2 × 2 pixels. For odd numbered image size, separate 2 × 2 sized blocks are formed from the last row and/or column pixels. The generated blocks and the remaining pixels (if any) are then arranged in separate arrays. Finally, the array of blocks and remaining pixels are separately permuted and diffused using different piece-wise linear chaotic map (PWLCM) systems. The significance of this algorithm is the use of arbitrary sized multiple images to perform multiple image encryptions. Another significance of this algorithm is the use of only PWLCM systems in permutation and diffusion operations to make the algorithm secure and efficient in both software and hardware platforms. The computer simulation reveals the good encryption results of the proposed cryptosystem. The security analysis shows that the proposed method performs better against widely known security attacks.

Seed quality is of great importance for agricultural cultivation. High-throughput phenotyping techniques can collect magnificent seed information in a rapid and non-destructive manner. Emerging deep learning technology brings new opportunities for effectively processing massive and diverse data from seeds and evaluating their quality. This article comprehensively reviews the principle of several high-throughput phenotyping techniques for non-destructively collection of seed information. In addition, recent research studies on the application of deep learning-based approaches for seed quality inspection are reviewed and summarized, including variety classification and grading, seed damage detection, components prediction, seed cleanliness, vitality assessment, etc. This review illustrates that the combination of deep learning and high-throughput phenotyping techniques can be a promising tool for collection of various phenotype information of seeds, which can be used for effective evaluation of seed quality in industrial practical applications, such as seed breeding, seed quality inspection and management, and seed selection as a food source.

Deep learning models have been widely used in hyperspectral images classification. However, the classification results are not satisfactory when the number of training samples is small. Focused on above-mentioned problem, a novel Two-stage Multi-dimensional Convolutional Stacked Autoencoder (TMC-SAE) model is proposed for hyperspectral images classification. The proposed model is composed of two sub-models SAE-1 and SAE-2. The SAE-1 is a 1D autoencoder with asymmetric structre based on full connection layers and 1D convolution layers to reduce spectral dimensionality. The SAE-2 is a hybrid autoencoder composed of 2D and 3D convolution operations to extract spectral-spatial features from the reduced dimensionality data by SAE-1. The SAE-1 is trained with raw data by unsupervised learning and the encoder of SAE-1 is employed to reduce spectral dimensionality of raw data. The data after dimension reduction is used to train the SAE-2 by unsupervised learning. The fine-tuning of SAE-2 encoder and the training of classifier are implemented simultaneously with small number of samples by supervised learning. Comparative experiments are performed on three widely used hyperspectral remote sensing data. The extensive comparative experiments demonstrate that the proposed architecture can effectively extract deep features and maintain high classification accuracy with small number of training samples.

Based on chaotic encryption technology and DNA cryptography, a multidimensional chaotic image encryption algorithm based on DNA coding is proposed in this paper. Firstly, the MD5 algorithm is used to extract the features of the image, generate a new key in association with the user key, and then encode the original image in DNA. The traditional three-dimensional Lorenz system is improved to form a four-dimensional hyperchaotic Lorenz system. According to the principle of DNA cryptography, a series of operations such as scrambling, mutation, DNA addition and DNA XOR are performed on the DNA coding sequence, and finally decoded to obtain a password image. Key space analysis, statistical analysis, known plaintext attack analysis and experimental results show that the algorithm has better image encryption performance and ability to resist various common attacks.

Session-based recommendation (SBR) aims to recommend the next clicked item to users by mining the user’s interaction sequences in the current session. It has received widespread attention recently due to its excellent privacy protection capabilities. However, existing SBR methods have the following limitations: (1) there exists noisy information in session sequences; (2) it is a challenge to simultaneously model both the long-term stable and dynamic changing interests of users; (3) the internal relationships between different interest representations are often neglected. To address the above issues, we propose an E xploring M ulti- D imensional I nterests for session-based recommendation model, termed EMDI, which attempts to predict more accurate and complete user intentions from multiple dimensions of user interests. Specifically, the EMDI contains the following three aspects: (1) the interest enhancement module aims to filter noise and enhance the interest expressions in the user’s behavior sequences, providing high-quality item embeddings; (2) the interest mining module separately mines users’ multi-dimensional interests, including static interests, local dynamic interests, and global dynamic interests, to capture users’ tendencies in different dimensions of interest; (3) the interest fusion module is designed to dynamically aggregate users’ interest representations from different dimensions through a novel multi-layer gated fusion network so that the implicit association between interest representations can be captured. Extensive experimental results show that the EMDI performs significantly better than other state-of-the-art methods.

Purpose Community question answering (CQA) platforms play a significant role in knowledge dissemination and information retrieval. Expert recommendation can assist users by helping them find valuable answers efficiently. Existing works mainly use content and user behavioural features for expert recommendation, and fail to effectively leverage the correlation across multi-dimensional features. Design/methodology/approach To address the above issue, this work proposes a multi-dimensional feature fusion-based method for expert recommendation, aiming to integrate features of question–answerer pairs from three dimensions, including network features, content features and user behaviour features. Specifically, network features are extracted by first learning user and tag representations using network representation learning methods and then calculating questioner–answerer similarities and answerer–tag similarities. Secondly, content features are extracted from textual contents of questions and answerer generated contents using text representation models. Thirdly, user behaviour features are extracted from user actions observed in CQA platforms, such as following and likes. Finally, given a question–answerer pair, the three dimensional features are fused and used to predict the probability of the candidate expert answering the given question. Findings The proposed method is evaluated on a data set collected from a publicly available CQA platform. Results show that the proposed method is effective compared with baseline methods. Ablation study shows that network features is the most important dimensional features among all three dimensional features. Practical implications This work identifies three dimensional features for expert recommendation in CQA platforms and conducts a comprehensive investigation into the importance of features for the performance of expert recommendation. The results suggest that network features are the most important features among three-dimensional features, which indicates that the performance of expert recommendation in CQA platforms is likely to get improved by further mining network features using advanced techniques, such as graph neural networks. One broader implication is that it is always important to include multi-dimensional features for expert recommendation and conduct systematic investigation to identify the most important features for finding directions for improvement. Originality/value This work proposes three-dimensional features given that existing works mostly focus on one or two-dimensional features and demonstrate the effectiveness of the newly proposed features.

Throughout the building’s entire lifetime, the construction industry faces significant challenges in the areas of real-time monitoring, energy efficiency, and intelligent decision-making. Traditional approaches lack the responsiveness and integration necessary for smart building operations. As a consequence, they frequently result in delays, inefficiencies in resource utilization, and unsatisfactory interior environments. This research presents a unique framework for enhancing building operations and maintenance, based on Multi-Dimensional Digital Twin Technology-assisted Building Information Modeling (MD-DTT-BIM). The purpose of this framework is to address the challenges that have been identified. The proposed system utilizes real-time data integration, simulation, and predictive analytics to enhance the planning and management of construction projects and facilities simultaneously. The MD-DTT-BIM model achieves substantial performance improvements over existing models, as demonstrated by experimental validation. These improvements include a 97.6% increase in operational efficiency, a 96.7% improvement in real-time monitoring accuracy, a 95.3% reduction in energy consumption, a 94.3% rise in occupant satisfaction, and a 98.3% accuracy in predicting the quality of the indoor environment. Based on these findings, it is evident that the model has the potential to facilitate a transition toward more intelligent and environmentally responsible building methods.