The semantic understanding of numbers requires association with context. However, powerful neural networks overfit spurious correlations between context and numbers in training corpus can lead to the occurrence of contextual bias, which may affect the network's accurate estimation of number magnitude when making inferences in real-world data. To investigate the resilience of current methodologies against contextual bias, we introduce a novel out-of-distribution (OOD) numerical question-answering (QA) dataset that features specific correlations between context and numbers in the training data, which are not present in the OOD test data. We evaluate the robustness of different numerical encoding and decoding methods when confronted with contextual bias on this dataset. Our findings indicate that encoding methods incorporating more detailed digit information exhibit greater resilience against contextual bias. Inspired by this finding, we propose a digit-aware position embedding strategy, and the experimental results demonstrate that this strategy is highly effective in improving the robustness of neural networks against contextual bias.

Instance segmentation is a challenging research in the field of computer vision, which combines the prediction results of object detection and semantic segmentation to provide richer image feature information. Focusing on the instance segmentation in the street scene, the real-time instance segmentation method based on SOLOv2 is proposed in this paper. First, a cross-stage fusion backbone network based on position attention is designed to increase the model accuracy and reduce the computational effort. Then, the loss of shallow location information is decreased by integrating two-way feature pyramid networks. Meanwhile, cross-stage mask feature fusion is designed to resolve the small objects missed segmentation. Finally, the adaptive minimum loss matching method is proposed to decrease the loss of segmentation accuracy due to object occlusion in the image. Compared with other mainstream methods, our method meets the real-time segmentation requirements and achieves competitive performance in segmentation accuracy.

State of the art machine learning methods can enhance the analysis of clinical data and improve the ability to predict patient outcomes because data collected from clinical records, such as heart failure mortality studies, are often high dimensional, heterogeneous and give challenges to traditional statistical analysis techniques. To address this challenge, this study conducted a survival analysis based on a dataset of 299 patients with heart failure, using Python libraries. Cox regression was used to model and analyse mortality, and to find which features are strongly associated with this outcome. The Kaplan-Meier survival curve approach was used to show the patterns of patient survival over time. The analysis showed that age, ejection fraction, and serum creatinine level were significantly (p≤0.001) associated with mortality. Anaemia and creatinine phosphokinase also reached statistical significance (p-values 0.026 and 0.007, respectively). The Cox model showed good concordance (0.77) with the data, suggesting that the identified variables are useful for predicting mortality in patients with heart failure.

One of the biggest challenges for the medical professionals is spotting cardiovascular issues in the earliest stages. Around the world, Cardiovascular Diseases (CVD) are a major cause of death for almost 18 million people each year. Heart disease is therefore a serious concern that needs to be treated. The numerous elements that affect health, such as excessive blood pressure, elevated cholesterol, aberrant pulse rate, and many other factors, might make it challenging to detect heart disease. Consequently, early disease detection and the development of effective treatments can benefit greatly from the field of artificial intelligence. The purpose of this work is to develop a new IoT based healthcare monitoring framework for the prediction of CVD using machine learning algorithm. Here, the data preprocessing has been performed to create the normalized dataset for improving classification. Then, an Artificial Gorilla Troop Optimization (AGTO) algorithm is deployed to choose the most pertinent features from the normalized dataset. Moreover, the Multi-Linear Regression Classification (MLRC) model is also implemented for accurately categorizing the medical information as whether healthy or CVD affected. The results of the proposed AGTO-MLRC mechanism is validated and compared using the popular benchmarking datasets.

The Internet of Things (IoT) is facing growing security challenges due to its vulnerability. It is imperative to address the security issues using lightweight and efficient encryption schemes in resource-limited IoT. In this paper, we propose an enhanced message authentication encryption (MAE) scheme based on physical-layer key generation (PKG), which uses the random nature of wireless channels to generate and negotiate keys, and simultaneously encrypts the messages and authenticates the source. The proposed enhanced MAE scheme can greatly improve the security performance via dynamic keyed primitives construction while consuming very few resources. The enhanced MAE scheme is an efficient and lightweight secure communication solution, which is very suitable for resource-limited IoT. Theoretical analysis and simulations are carried out to confirm the security of the enhanced MAE scheme and evaluate its performance. A one-bit flipping in the session key or plain texts will result in a 50%-bit change in the ciphertext or message authentication code. The numerical results demonstrate the good performance of the proposed scheme in terms of diffusion and confusion. With respect to the typical advanced encryption standard (AES)-based scheme, the performance of the proposed scheme improves by 80.5% in terms of algorithm execution efficiency.

Due to COVID19 pandemic, most of educational institutions and schools changed the traditional way of teaching to online teaching and learning using well-known Learning Management Systems (LMS) such as Moodle, Canvas, Blackboard, etc. Accordingly, LMS started to generate a large data related to students' characteristics and achievements and other course-related information. This makes it difficult to teachers to monitor students' behaviour and performance. Therefore, a need to support teachers with a tool alerting student who might be in risk based on their recorded activities and achievements in adopted LMS in the school. This paper focuses on the benefits of using recorded data in LMS platforms, specifically Moodle, to predict students' performance by analysing their behavioural data and engagement activities using data mining techniques. As part of the overall process, this study encountered the task of extracting and selecting relevant data features for predicting performance, along with designing the framework and choosing appropriate machine learning techniques. The collected data underwent pre-processing operations to remove random partitions, empty values, duplicates, and code the data. Different machine learning techniques, including k-NN, TREE, Ensembled Tree, SVM, and MLPNNs were applied to the processed data. The results showed that the MLPNNs technique outperformed other classification techniques, achieving a classification accuracy of 93%, while SVM and k-NN achieved 90% and 87% respectively. This indicates the possibility for future research to investigate incorporating other neural network methods for categorizing students using data from LMS.

Brain tumors, characterized by uncontrollable cellular growths, are a significant global health challenge. Navigating the complexities of tumor identification due to their varied dimensions and positions, our research introduces enhanced methods for precise detection. Utilizing advanced learning techniques, we've improved early identification by preprocessing clinical dataset-derived images, augmenting them via a Generative Adversarial Network, and applying an Improved Privacy-Preserving Collaborative Convolutional Neural Network (IPC-CNN) for segmentation. Recognizing the critical importance of data security in today's digital era, our framework emphasizes the preservation of patient privacy. We evaluated the performance of our proposed model on the Figshare and BRATS 2018 datasets. By facilitating a collaborative model training environment across multiple healthcare institutions, we harness the power of distributed computing to securely aggregate model updates, ensuring individual data protection while leveraging collective expertise. Our IPC-CNN model achieved an accuracy of 99.40%, marking a notable advancement in brain tumor classification and offering invaluable insights for both the medical imaging and machine learning communities.

Siamese-based segmentation and tracking algorithms improve accuracy and stability for video object segmentation and tracking tasks simultaneously. Although effective, variability in target appearance and background clutter can still affect segmentation accuracy and further influence the performance of tracking. In this paper, we present a memory propagation-based target-aware and mask-attention decision network for robust object segmentation and tracking. Firstly, a mask propagation-based attention module (MPAM) is constructed to explore the inherent correlation among image frames, which can mine mask information of the historical frames. By retrieving a memory bank (MB) that stores features and binary masks of historical frames, target attention maps are generated to highlight the target region on backbone features, thus suppressing the adverse effects of background clutter. Secondly, an attention refinement pathway (ARP) is designed to further refine the segmentation profile in the process of mask generation. A lightweight attention mechanism is introduced to calculate the weight of low-level features, paying more attention to low-level features sensitive to edge detail so as to obtain segmentation results. Finally, a mask fusion mechanism (MFM) is proposed to enhance the accuracy of the mask. By utilizing a mask quality assessment decision network, the corresponding quality scores of the "initial mask" and the "previous mask" can be obtained adaptively, thus achieving the assignment of weights and the fusion of masks. Therefore, the final mask enjoys higher accuracy and stability. Experimental results on multiple benchmarks demonstrate that our algorithm performs outstanding performance in a variety of challenging tracking tasks.

Images taken in low lamination condition suffer from low contrast and loss of information. Low lumination image enhancement algorithms are required to improve the quality and broaden the applications of such images. In this study, we proposed a new Low lumination image enhancement architecture consisting of a transformer-based curve learning and an encoder-decoder-based texture enhancer. Considering the high effectiveness of curve matching, we constructed a transformer-based network to estimate the learnable curve for pixel mapping. Curve estimation requires global relationships that can be extracted through the transformer framework. To further improve the texture detail, we introduced an encoder-decoder network to extract local features and suppress the noise. Experiments on LOL and SID datasets showed that the proposed method not only has competitive performance compared to state-of-the-art techniques but also has great efficiency.

In the process of modern traffic management, information technology has become an important part of intelligent traffic governance. Real-time monitoring can accurately and effectively track and record vehicles, which is of great significance to modern urban traffic management. Existing tracking algorithms are affected by the environment, viewpoint, etc., and often have problems such as false detection, imprecise anchor boxes, and ID switch. Based on the YOLOv5 algorithm, we improve the loss function, propose a new feature extraction module to obtain the receptive field at different scales, and do adaptive fusion with the SGE attention mechanism, so that it can effectively suppress the noise information during feature extraction. The trained model improves the mAP value by 5.7% on the public dataset UA-DETRAC without increasing the amount of calculations. Meanwhile, for vehicle feature recognition, we adaptively adjust the network structure of the DeepSort tracking algorithm. Finally, we tested the tracking algorithm on the public dataset and in a realistic scenario. The results show that the improved algorithm has an increase in the values of MOTA and MT etc., which generally improves the reliability of vehicle tracking.

The philosophy of Zero Trust in cybersecurity lies in the notion that nothing assumes to be trustworthy by default. This drives defense organizations to modernize their cybersecurity architecture through integrating with the zero-trust principles. The enhanced architecture is expected to shift protection strategy from static and perimeter-centric protection to dynamic and proactive measures depending on the logical contexts of users, assets, and infrastructure. Given the domain context of defense environment, we aim three challenge problems to tackle and identify four technical approaches by the security capabilities defined in the Zero Trust Architecture. First approach, dynamic access control manages visibility and accessibility to resources or services with Multi Factor Authentication and Software Defined Perimeter. Logical network separation approach divides networks on a functional basis by using Software Defined Network and Micro-segmentation. Data-driven analysis approach enables machine-aided judgement by utilizing Artificial Intelligence, User and Entity Behavior Analytics. Lastly, Security Awareness approach observes fluid security context of all resources through Continuous Monitoring and Visualization. Based on these approaches, a comprehensive study of modern technologies is presented to materialize the concept that each approach intends to achieve. We expect this study to provide a guidance for defense organizations to take a step on the implementation of their own zero-trust architecture.

Over the past decade, law enforcement organizations have been dealing with the development of cybercrime. To address this growing problem, law enforcement organizations apply various digital forensic (DF) tools and techniques to investigate crimes involving digital devices. This ensures that evidence is admissible in legal proceedings. Consequently, DF analysts may need to invest more in proprietary DF hardware and software to maintain the viability of the DF lab, which will burden budget-constrained organizations. As an alternative, the open source DF tool is considered a cost-saving option. However, the admissibility of digital evidence obtained from these tools has yet to be tested in courts, especially in Malaysia. Therefore, this study aimed to explore the admissibility of digital evidence obtained through open source DF tools. By reviewing the existing literature, the factors that affect the admissibility of the evidence produced by these tools in courts were identified. Further, based on the findings, a conceptual framework was developed to ensure the admissibility of the evidence so that it will be accepted in the court of law. This conceptual framework was formed to outline the factors affecting the admissibility of digital evidence from open source DF tools, which include; 1) The Availability and Capability of open source DF tools, 2) the Reliability and Integrity of the digital evidence obtained from open source DF tools, 3) the Transparency of the open source DF tools, and 4) the Lack of Reference and Standard of open source DF tools. This study provides valuable insights into the digital forensic field, and the conceptual framework can be used to integrate open source DF tools into digital forensic investigations.

Cloud computing has extensively grown in recent years. A large amount of data is stored in cloud servers. To ensure confidentiality, these data is often encrypted and then stored in cloud servers. However, encryption makes range queries difficult to perform. To solve this issue, we present a scheme that facilitates fast range queries on encrypted multi-dimensional data in scenarios involving multiple users. In our scheme, we construct a tree index on encrypted multi-dimensional data, and each node is linked to a secure enhanced multi-dimensional range (MDR). To support efficient range query on the tree index, we adopt bloom filter technique. Additionally, users' privileges are designed in a one-way calculation manner to support that different users can only perform range queries within their own privileges. Finally, we conduct extensive experiments which show the efficiency of our scheme, and also conduct a thorough analysis of its security.

In wide area SDP networks, the failure of SDP components caused by malicious attacks will be accompanied by different deployment locations, profoundly affecting network service latency. However, traditional deployment methods based on prior knowledge are no longer applicable to dynamic SDP networks. This article proposes a dynamic and dimensionally variable deployment mechanism DDPG-SDPCR for SDP components based on DDPG, which enhances the network's endogenous security capability and improves attack tolerance. Based on this, we constructed corresponding mathematical models for latency, load balancing, and attack tolerance. The DDPG-SDPCR mechanism dynamically deploys new SDP nodes to replace faulty nodes based on the real-time status of the network, thereby achieving imperceptible attack tolerance for users. We have implemented a wide area SDP prototype with endogenous security capabilities and evaluated it under different network topologies, traffic sizes, and network attacks. The evaluation results indicate that under high traffic conditions, our proposed redeployment mechanism outperforms the baseline by 36.42% in latency, and only increases by 19.24% compared to the non attacked situation.

Federated learning (FL) has been proposed as an emerging distributed machine learning framework, which lowers the risk of privacy leakage by training models without uploading original data. Therefore, it has been widely utilized in the Industrial Internet of Things (IIoT). Despite this, FL still faces challenges including the non-independent identically distributed (Non-IID) data and heterogeneity of devices, which may cause difficulties in model convergence. To address these issues, a local surrogate function is initially constructed for each device to ensure a smooth decline in global loss. Subsequently, aiming to minimize the system energy consumption, an FL approach for joint CPU frequency control and bandwidth allocation, called FCBAFL is proposed. Specifically, the maximum delay of a single round is first treated as a uniform delay constraint, and a limited-memory Broyden-Fletcher-Goldfarb-Shanno bounded (L-BFGS-B) algorithm is employed to find the optimal bandwidth allocation with a fixed CPU frequency. Following that, the result is utilized to derive the optimal CPU frequency. Numerical simulation results show that the proposed FCBAFL algorithm exhibits more excellent convergence compared with baseline algorithm, and outperforms other schemes in declining the energy consumption.

The integration of artificial intelligence technology with medicine has rapidly evolved, with increasing demands for quality of life. However, falls remain a significant risk leading to severe injuries and fatalities, especially among the elderly. Therefore, the development and application of computer vision-based fall detection technologies have become increasingly important. In this paper, firstly, the keypoint detection algorithm ViTPose++ is used to obtain the coordinates of human body keypoints from the camera images. Human skeletal feature maps are generated from this keypoint coordinate information. Meanwhile, human dense feature maps are produced based on the DensePose algorithm. Then, these two types of feature maps are confused as dual-channel inputs for the model. The convolutional gated recurrent unit is introduced to extract the frame-to-frame relevance in the process of falling. To further integrate features across three dimensions (spatio-temporal-channel), a dual-channel fall detection algorithm based on video streams is proposed by combining the Convolutional Block Attention Module (CBAM) with the ConvGRU. Finally, experiments on the public UR Fall Detection Dataset demonstrate that the improved ConvGRU-CBAM achieves an F1 score of 92.86% and an AUC of 95.34%.

This study focuses on the performance of device-to-device (D2D) communications in underlay cellular networks by analyzing key metrics such as successful transmission probability, coverage probability, and throughput. Under the homogeneous Poisson point process (PPP) spatial distribution of full-duplex (FD)-D2D users in cellular networks, stochastic geometry tools are used to derive approximate expressions for D2D users' coverage probability and throughput. In comparison to the conventional half-duplex (HD) communication mode, when the self-interference cancellation factor β reaches -95 dB, there is a substantial improvement in the throughput of FD-D2D users, nearly doubling their gain. Additionally, experimental results demonstrate that the Newton iterative algorithm can be used to optimize the targeted signal-to-interference-plus-noise-ratio (SINR) threshold of users within the range of (10, 20) dB.