Due to the inherent limitations of underground spaces, such as the lack of natural ventilation and sunlight, underground space users tend to face more health risks compared with their aboveground counterparts. However, little is known about how the underground environment, users’ health, and their associations were impacted by the outbreak of the pandemic. In this study, we investigated and compared the impacts of the general underground environment on regular users’ physical and psychological health before and after the pandemic. To achieve this aim, the data from 525 surveys were collected from eleven underground sites, followed by an objective field measurement study conducted at five underground sites in Hong Kong pre- and post-outbreak of the pandemic. The multigroup structural equation modelling results indicated that: (i) surprisingly, the users’ satisfaction towards almost all underground environment factors, including greenery, connectivity with the aboveground environment, thermal comfort, ventilation, indoor air quality, acoustic comfort, and lighting, excluding wayfinding, were significantly higher in the post-outbreak period; (ii) the users’ health, both physical and psychological, was significantly better in the post-outbreak period; (iii) the impacts of visual comfort on the users’ physical and psychological health were significantly greater in the post-outbreak period (critical difference ratio (|CDR|) > 1.96); (iv) the impacts of wayfinding, greenery, and acoustic and thermal comfort on the users’ physical or psychological health were significant only in the pre-outbreak period (|CDR| > 1.96); (v) the impacts of connectivity on the users’ physical and psychological health were significant in both the pre- and post-outbreak periods (|CDR| < 1.96). The findings were further cross-validated using the objective measurement results. With an increasing need to develop healthy underground spaces, the study contributes to the development, design, and management of the underground environment to enhance the users’ health in the post-outbreak era.
Constructing the volume data which can be roaming in it can be well observed from multi-orientation. A modeling and simulation method based on the texture mapping to the volume data which makes it be roaming in it is proposed. First, get of the model of volume data, that is to say, after define the viewpoint, map the texture onto the geometrical model which containing the volume data. Thus finish the modeling of the volume data which can be roaming in it. Finally, let the viewers navigate in the model so as to realize the simulation of the volume data. If there are “outside volume data” and “inside volume data”, with this method, the viewers can not only navigate the “outside volume data”, but also inside the model so that they can get the rules of the “inside volume data” which make this method have a better visualization. Besides, some methods to accelerate the visualization of the volume data are adopted, such as the bounding box method, Linear Octree Space Partitioning algorithm when processing irregular volume data. The results have proved that the method proposed is simple and efficient.
Jiaji Logistics is one of the biggest logistics enterprises in our country, but the survey found that there are still many problems in it, which limits the development of Jiaji logistics. One of the important factors is the location of the distribution center. In order to overcome this difficulty, the location of Jiaji logistics is used scientifically, and it can also help the development of logistics industry in China. First of all, in this paper, by analyzing distribution center location of the relevant literature both at home and abroad, according to oneself the circumstance of master data, choose the Baumov Model for the theme as a solution. Then, the factors affecting the distribution center of Jiaji logistics are selected, and the optimal solution is obtained by the successive approximation method. The final result will be the distribution center capacity, location, volume allocation and so on. The improved model of the Baumov Model was a great help to the location of the Jiaji logistics, which reduced the cost of the Jiaji logistics.
Data augmentation and adversarial perturbation approaches have recently achieved promising results in solving the over-fitting problem in many natural language processing (NLP) tasks including sentiment classification. However, existing studies aimed to improve the generalization ability by augmenting the training data with synonymous examples or adding random noises to word embeddings, which cannot address the spurious association problem. In this work, we propose an end-to-end reinforcement learning framework, which jointly performs counterfactual data generation and dual sentiment classification. Our approach has three characteristics:1) the generator automatically generates massive and diverse antonymous sentences; 2) the discriminator contains a original-side sentiment predictor and an antonymous-side sentiment predictor, which jointly evaluate the quality of the generated sample and help the generator iteratively generate higher-quality antonymous samples; 3) the discriminator is directly used as the final sentiment classifier without the need to build an extra one. Extensive experiments show that our approach outperforms strong data augmentation baselines on several benchmark sentiment classification datasets. Further analysis confirms our approach’s advantages in generating more diverse training samples and solving the spurious association problem in sentiment classification.
Remote sensing (RS) images change detection (CD) aims to obtain change information of the target area between multi-temporal RS images. With the modernization of cities, building change detection (BCD) plays a pivotal role in land resource planning, smart city construction and natural disaster assessment, and it is a typical application field of change detection task. Recently, deep learning based methods have shown their superiority in RS image change detection. However, the performance of the existing supervised change detection methods relies heavily on a large amount of high quality annotated bi-temporal RS image as training data, which is usually hard to obtain in practice. To address this issue, a semi-supervised BCD method using a pseudo bi-temporal data generator with consistency regularization was proposed. This method only needs a very small amount of single-temporal RS images with building extraction labels as labeled data. Firstly, with the help of the pseudo bi-temporal data generator, the model can generate a large number of pseudo bi-temporal images with CD labels from a small number of single-temporal images and corresponding building extraction labels automatically, which greatly augments the labeled data set for CD model training. Then, we proposed an error-prone data enhancement fine-tuning strategy to improve the learning effect of the proposed model to these synthesized training data. Finally, we enhance the robustness of the model by forcing the model to make consistent predictions on the images before and after perturbations. Extensive experimental results demonstrate that our method can effectively improve the BCD performance of the model even if labeled data are scare, and outperforms the state-of-the-art methods.
Membership inference attacks (MIAs) are critical tools for assessing privacy risks and ensuring compliance with regulations like the General Data Protection Regulation (GDPR). However, their potential for auditing unauthorized use of data remains under explored. To bridge this gap, we propose a novel clean-label backdoor-based approach for MIAs, designed specifically for robust and stealthy data auditing. Unlike conventional methods that rely on detectable poisoned samples with altered labels, our approach retains natural labels, enhancing stealthiness even at low poisoning rates. Our approach employs an optimal trigger generated by a shadow model that mimics the target model's behavior. This design minimizes the feature-space distance between triggered samples and the source class while preserving the original data labels. The result is a powerful and undetectable auditing mechanism that overcomes limitations of existing approaches, such as label inconsistencies and visual artifacts in poisoned samples. The proposed method enables robust data auditing through black-box access, achieving high attack success rates across diverse datasets and model architectures. Additionally, it addresses challenges related to trigger stealthiness and poisoning durability, establishing itself as a practical and effective solution for data auditing. Comprehensive experiments validate the efficacy and generalizability of our approach, outperforming several baseline methods in both stealth and attack success metrics.
'/%3e%3cuse xlink:href='%23a' transform='rotate(23.036 .093 25.536)'/%3e%3cuse xlink:href='%23a' transform='rotate(45.87 1.273 16.18)'/%3e%3cuse xlink:href='%23a' transform='rotate(69.945 .996 12.078)'/%3e%3cuse xlink:href='%23a' transform='rotate(20.66 -19.689 31.932)'/%3e%3c/svg%3e)