Falls are one of the main causes of elderly injuries. If the faller can be found in time, further injury can be effectively avoided. In order to protect personal privacy and improve the accuracy of fall detection, this paper proposes a fall detection algorithm using the CNN-Casual LSTM network based on three-axis acceleration and three-axis rotation angular velocity sensors. The neural network in this system includes an encoding layer, a decoding layer, and a ResNet18 classifier. Furthermore, the encoding layer includes three layers of CNN and three layers of Casual LSTM. The decoding layer includes three layers of deconvolution and three layers of Casual LSTM. The decoding layer maps spatio-temporal information to a hidden variable output that is more conducive relative to the work of the classification network, which is classified by ResNet18. Moreover, we used the public data set SisFall to evaluate the performance of the algorithm. The results of the experiments show that the algorithm has high accuracy up to 99.79%.

Falls can easily cause major harm to the health of the elderly, and timely detection can avoid further injuries. To detect the occurrence of falls in time, we propose a new method called Patch-Transformer Network (PTN) wearable-sensor-based fall detection algorithm. The neural network includes a convolution layer, a Transformer encoding layer, and a linear classification layer. The convolution layer is used to extract local features and project them into feature matrices. After adding positional coding information, the global features of falls are learned through the multi-head self-attention mechanism in the Transformer encoding layer. Global average pooling (GAP) is used to strengthen the correlation between features and categories. The final classification results are provided by the linear layer. The accuracy of the model obtained on the public available datasets SisFall and UnMib SHAR is 99.86% and 99.14%, respectively. The network model has fewer parameters and lower complexity, with detection times of 0.004 s and 0.001 s on the two datasets. Therefore, our proposed method can timely and accurately detect the occurrence of falls, which is important for protecting the lives of the elderly.

Since the actual factors in the instant distribution service scenario are not considered enough in the existing distribution route optimization, a route optimization model of the instant distribution system based on customer time satisfaction is proposed. The actual factors in instant distribution, such as the soft time window, the pay-to-order mechanism, the time for the merchant to prepare goods before delivery, and the deliveryman’s order combining, were incorporated in the model. A multi-objective optimization framework based on the total cost function and time satisfaction of the customer was established. Dual-layer chromosome coding based on the deliveryman-to-node mapping and the access order was conducted, and the nondominated sorting genetic algorithm version II (NSGA-II) was used to solve the problem. According to the numerical results, when time satisfaction of the customer was considered in the instant distribution routing problem, the customer satisfaction increased effectively and the balance between customer satisfaction and delivery cost in the means of Pareto optimization were obtained, with a minor increase in the delivery cost, while the number of deliverymen slightly increased to meet the on-time delivery needs of customers.

Network Intrusion Detection Systems (NIDS) are fundamental to network security. However, traditional approaches often struggle with high-dimensional and imbalanced network traffic data. To address these challenges, we propose LRT-DDPM, a novel network traffic data generation model based on the Denoising Diffusion Probabilistic Model (DDPM). Our model employs a stepwise denoising mechanism to generate high-fidelity synthetic data, enhances feature extraction through a convolutional neural network with residual connections and introduces the Time-Adaptive Control Block (TACBlock) strategy for dynamic time step adjustment during diffusion. LRT-DDPM demonstrates superior performance on two widely-used benchmark datasets. It achieves an accuracy of 99.40% on NSL-KDD and 99.53% on CICIDS2017. Furthermore, the model attains high precision, recall, and F1-score metrics, consistently outperforms existing methods by enhancing detection accuracy and reducing false positive rates.

Malaria is one of the major global health threats. Microscopic examination has been designated as the “gold standard” for malaria detection by the World Health Organization. However, it heavily relies on the experience of doctors, resulting in long diagnosis time, low efficiency, and a high risk of missed or misdiagnosed cases. To alleviate the pressure on healthcare workers and achieve automated malaria detection, numerous target detection models have been applied to the blood smear examination for malaria cells. This paper introduces the multi-level attention split network (MAS-Net) that improves the overall detection performance by addressing the issues of information loss for small targets and mismatch between the detection receptive field and target size. Therefore, we propose the split contextual attention structure (SPCot), which fully utilizes contextual information and avoids excessive channel compression operations, reducing information loss and improving the overall detection performance of malaria cells. In the shallow detection layer, we introduce the multi-scale receptive field detection head (MRFH), which better matches targets of different scales and provides a better detection receptive field, thus enhancing the performance of malaria cell detection. On the NLM—Malaria Dataset provided by the National Institutes of Health, the improved model achieves an average accuracy of 75.9% in the public dataset of Plasmodium vivax (malaria)-infected human blood smear. Considering the practical application of the model, we introduce the Performance-aware Approximation of Global Channel Pruning (PAGCP) to compress the model size while sacrificing a small amount of accuracy. Compared to other state-of-the-art (SOTA) methods, the proposed MAS-Net achieves competitive results.

Network intrusion detection is a vital component in contemporary cybersecurity defense frameworks. As network traffic data experiences exponential growth, deep learning-based intrusion detection algorithms have become a hot research topic. However, existing intrusion detection methods commonly exhibit low accuracy, high false positive rates, and poor performance in detecting minority class attacks. To address these challenges, this paper presents an efficient two-stage intrusion detection model, MARnet. MARNet consists of two stages: the first stage is multi-scale feature classification extraction, where numerical and categorical features in the dataset are processed separately and then fused. The second stage is the classification phase, which employs residual networks and channel attention to automatically and effectively extract features from network traffic, while identifying crucial features. Furthermore, we implemented EQL v2 loss to tackle the problem of class imbalance within the dataset. Experiments were conducted using the UNSW-NB15 and NSL-KDD datasets,After a thorough and comprehensive evaluation on two widely used datasets, the model achieved a multi-class accuracy of 90.55% and a detection rate of 98.14% with a false positive rate of 0.61% on the UNSW-NB15 dataset. On the NSL-KDD dataset, the model’s multi-class accuracy was 99.96%, detection rate was 99.96%, and the false positive rate was 0.01%. The experimental results show that the model surpasses existing ones in accuracy, detection rate, false alarm rate, and detection of minority class attacks, effectively meeting the requirements of modern large-scale intrusion detection systems.