High-Density Parking (HDP) offers an alternative solution to enhance urban space utilization by increasing parking density through modifications to existing parking facilities. Current HDP dispatch strategies primarily rely on heuristic methods, the adaptability of which to diverse parking scenarios has not been adequately evaluated. 7 distinct parking demand patterns are identified with different entry/exit distribution from real-world parking data to create a comprehensive evaluation framework. A novel Dueling Deep Q-Network (Dueling DQN) approach is proposed, which effectively captures the complex spatial-temporal relationships in HDP environments through specialized state representation. Extensive experiments evaluated the adaptability of existing heuristic and the proposed approaches under diverse parking scenarios with different parking patterns, stack depths, supply-demand ratios, and departure time estimation errors. The Dueling DQN model performs superiorly and exhibits exceptional reusability, maintaining superior performance without retraining when deployed across different parking scenarios. This research provides practical insights for implementing efficient HDP dispatch strategies in real-world settings.

In recent years, Transformer-based Change Detection (CD) in Remote Sensing Images has achieved significant advances, making it an emerging hot research topic. However, the current CD methods suffer from some problems, such as incomplete detection of change regions and missed detection of small change regions. In this paper, a Global Context-aware Transformer is proposed for CD tasks, named GCFormer, to address above issues by efficiently enhancing the global context information. It is fulfilled from two aspects based on the hybrid Convolutional Neural Network (CNN)+Transformer framework. Firstly, a Multi-Receptive Field Conv-Attention (MRFCA) mechanism is designed, which combines dilated convolutions with multiple rates and Conv-Attention, fully leveraging the advantages of convolution operation and self-attention mechanism. It is embedded at the highest layer of CNN to extract multi-receptive-field global context information. Secondly, a Context-aware Relative Position Encoding (CRPE) mode is proposed to replace the Absolute Position Encoding (APE) mode of Transformer. As a result, it can capture long-range dependency more efficiently and further enhance the global context information extraction and representation ability of the network. Experimental results on three public benchmark datasets of LEVIR-CD, WHU-CD and DSIFN-CD show that, the proposed GCFormer achieves superior detection performance with lower model complexity than the state-of-the-art Transformer-based CD methods. The source code is available at: https://github.com/yuwanting828/yuwanting828.github.io.

Automatic source code summarization system aims to generate a valuable natural language description for a program, which can facilitate software development and maintenance, code categorization, and retrieval. However, previous sequence-based research did not consider the long-distance dependence and highly structured characteristics of source code simultaneously. In this article, we present a Transformer-based Graph-Augmented Source Code Summarization (GA-SCS), which can effectively incorporate inherent structural and textual features of source code to generate an effective code description. Specifically, we develop a graph-based structure feature extraction scheme leveraging abstract syntax tree and graph attention networks to mine global syntactic information. And then, to take full advantage of the lexical and syntactic information of code snippets, we extend the original attention to a syntax-informed self-attention mechanism in our encoder. In the training process, we also adopt a reinforcement learning strategy to enhance the readability and informativity of generated code summaries. We utilize the Java dataset and Python dataset to evaluate the performance of different models. Experimental results demonstrate that our GA-SCS model outperforms all competitive methods on BLEU, METEOR, ROUGE, and human evaluations.

Assessments of postural control provide important insights into health. We created a smartphone App-based assessment of standing posture—completed with the phone placed in the user’s pocket—to enable remote monitoring of the function while minimizing the need for in-person assessment or contact. We tested the reliability of App-derived postural sway metrics, as well as their sensitivity to age and task conditions. Fifteen older and 15 younger adults completed two separate laboratory visits. They followed multimedia instructions provided by the App to complete three 30-s trials of eyes-open (EO), eyes-closed (EC), and dual-task (DT) standing. Sway data were recorded by the App and a force plate. Participants also used the App to complete the assessment in their homes on three separate days. Sway path length and root-mean square were derived from the angular velocity and acceleration acquired from phone’s internal motion sensor, and from the center-of-pressure signal from force plate. App-derived path length and root-mean-square of acceleration and angular velocity across conditions demonstrated moderate-to-excellent test–retest reliability (intraclass correlation coefficients (ICCs) = 0.69 − 0.97) for both younger and older adults. Reliability was comparable to metrics derived from the force plate (ICCs = 0.44 − 0.93). As expected, App-derived sway outcomes exhibited somewhat greater variability when tested across days at home (ICCs = 0.14 − 0.79). All App-derived metrics were sensitive to age ($F>43.8$,$p < 0.001$) and testing condition ($F>31.8$,$p < 0.001$) in both laboratory and home settings, revealing that the smartphone App enabled reliable and sensitive assessment of standing posture in both healthy younger and older adults.

Laboratory-based assessments of standing posture provide important insights into balance, mobility, and many other factors. We created a smartphone-based assessment of standing posture-completed with the phone placed in the user's pocket-for use in remote, non-laboratory settings. We here tested the reliability and sensitivity of App-derived metrics of postural sway to participant age and standing condition. Fifteen healthy younger and 15 older adults completed two laboratory visits separated by one week. On each visit, they followed multi-media instructions provided by the App to complete three 30-second trials each of standing with eyes open (EO), eyes closed (EC), and eyes open while performing a serial subtraction dual task (DT). Sway data were simultaneously collected with the App and a gold-standard force plate. Participants also completed App-based tests within their own homes on three separate days. To characterize sway, path length and root-mean-square (RMS) were derived from the acceleration and angular velocity signal acquired from the phone's internal motion sensor, and from the center of pressure (COP) signal acquired by the force plate. Across repeated trials conducted on the same day within the laboratory, App-derived path length and RMS of acceleration and angular velocity across conditions demonstrated moderate to excellent test-retest reliability (Intraclass Correlation Coefficients (ICCs)=0.69-0.97) for both the younger and older adults. Force plate-derived metrics demonstrated low to excellent test-retest reliability (ICCs=0.44-0.93). Compared to within-day laboratory testing, App-derived sway outcomes generally exhibited greater variability when tested across days within the home (ICCs=0.14-0.79). All sway metrics derived from the App were sensitive to age group (F>43.8, p31.8, p<; 0.001) in both laboratory and home settings. The smartphone App we created enabled reliable and sensitive assessment of standing postural sway within different task conditions, in both relatively healthy younger and older adults.

In image classification of deep learning, adversarial examples where input is intended to add small magnitude perturbations may mislead deep neural networks (DNNs) to incorrect results, which means DNNs are vulnerable to them. Different attack and defense strategies have been proposed to better research the mechanism of deep learning. However, those researches in these networks are only for one aspect, either an attack or a defense. There is in the improvement of offensive and defensive performance, and it is difficult to promote each other in the same framework. In this paper, we propose Cycle-Consistent Adversarial GAN (CycleAdvGAN) to generate adversarial examples, which can learn and approximate the distribution of the original instances and adversarial examples, especially promoting attackers and defenders to confront each other and improve their ability. For CycleAdvGAN, once the Generator A and D are trained, GA can generate adversarial perturbations efficiently for any instance, improving the performance of the existing attack methods, and GD can generate recovery adversarial examples to clean instances, defending against existing attack methods. We apply CycleAdvGAN under semiwhite-box and black-box settings on two public datasets MNIST and CIFAR10. Using the extensive experiments, we show that our method has achieved the state-of-the-art adversarial attack method and also has efficiently improved the defense ability, which made the integration of adversarial attack and defense come true. In addition, it has improved the attack effect only trained on the adversarial dataset generated by any kind of adversarial attack.

Human Activity Recognition (HAR) automatically recognizes human activities such as daily life and work based on digital records, which is of great significance to medical and health fields. Egocentric video and human acceleration data comprehensively describe human activity patterns from different aspects, which have laid a foundation for activity recognition based on multimodal behavior data. However, on the one hand, the low-level multimodal signal structures differ greatly and the mapping to high-level activities is complicated. On the other hand, the activity labeling based on multimodal behavior data has high cost and limited data amount, which limits the technical development in this field. In this paper, an activity recognition model MAFE based on multimodal attribute feature embedding is proposed. Before the activity recognition, the middle-level attribute features are extracted from the low-level signals of different modes. On the one hand, the mapping complexity from the low-level signals to the high-level activities is reduced, and on the other hand, a large number of middle-level attribute labeling data can be used to reduce the dependency on the activity labeling data. We conducted experiments on Stanford-ECM datasets to verify the effectiveness of the proposed MAFE method.