Weiping Zhu

Time Sensitive Network (TSN) provides strict low latency and bounded jitter requirements for applications such as industrial systems, autonomous driving, etc. One of the important problems in TSN is to achieve high reliability and low latency by effectively routing and scheduling time-sensitive data flows. Existing work applies heuristic or integer programming to address flow routing and scheduling, yet often fail to achieve optimal solutions quickly. In this paper, we propose a new Reinforcement Learning (RL) based approach for routing and scheduling of redundant data flows, aiming to achieve load balancing on the network links as well as meeting the reliability and delay constraints. Our approach first leverages a simple heuristic algorithm to decide the redundant path candidate set, and then incorporates Proximal Policy Optimization (PPO) method to choose the most suitable multi-routing flows from the candidates, which can be aware of the network status dynamically to reduce the load on the bottleneck link of the network. On this basis, we further retrain the RL model by fine-tuning to adapt to the online environment. The simulation results show that our proposed solution outperforms the benchmark algorithms in terms of the degree of network balance by 38.7% in offline network environments and in terms of average delay by 14.0% in online network environments.

Parameter tuning based adapter methods have achieved notable success in vision-language models (VLMs). However, they face challenges in scenarios with insufficient training samples or limited resources. While leveraging image modality caching and retrieval techniques can reduce resource requirements, these approaches often overlook the significance of textual modality and cross-modal cues in VLMs. To address this, we propose a Conditional Cross-Modal learning model, which is abbreviated as CoCM. CoCM builds separate cache models for both the text and image modalities and embedding textual knowledge conditioned on image information. It dynamically adjusts the cross-modal fusion affinity ratio and disentangles similarity measures across different modalities. Additionally, CoCM incorporates intra-batch image similarity loss as a regularization term to identify hard samples and enhance fine-grained classification performance. CoCM surpasses existing methods in terms of accuracy, generalization ability, and efficiency, achieving a 0.28% accuracy improvement over XMAdapter across 11 datasets and demonstrating 44.79% generalization performance on four cross-domain datasets.

Background: A variety of viruses can cause acute respiratory infections (ARIs), resulting in a high disease burden worldwide. To explore the dominant viruses and their prevalence characteristics in children with ARIs, comprehensive surveillance was carried out in the Pudong New Area of Shanghai.Methods: Between January 2013 and December 2022, the basic and clinical information, and respiratory tract specimens of 0–14 years old children with ARIs were collected in five sentinel hospitals in Shanghai Pudong. Each specimen was tested for eight respiratory viruses, and the positive rates of different age groups, case types (inpatient or outpatient) were analyzed.Results: In our study, 30.67% (1294/4219) children with ARIs were positive for at least one virus. Influenza virus (IFV) was the most commonly detected respiratory virus (349/4219, 8.27%), followed by respiratory syncytial virus (RSV) (217/4219, 5.14%), para‐influenza virus (PIV) (215/4219, 5.10%), and human coronavirus (HCoV, including 229E, OC43, NL63, and HKU1) (184/4219, 4.36%). IFV was the leading respiratory virus in outpatients aged 5–14 years (201/1673, 12.01%); RSV was the most prevalent respiratory virus in both inpatients (61/238, 25.63%) and outpatients (4/50, 8.00%) for ARI patients aged <6 months old. For PIV, HMPV, HCoV, and HRV, the risk of infection usually was higher among young children. Co‐infection with more than two viruses was seen in 3.25% (137/4219).Conclusions: IFV and RSV played important roles in ARIs among children, but the risk populations were different. There are needs for targeted diagnosis and treatment and necessary immunization and non‐pharmaceutical interventions.

The promising wireless powered mobile edge computing (MEC) can offer sustainable energy and fast network service response for nearby wireless terminals (WTs) to satisfy real-time and flexible requirements. Online task offloading and wireless power transfer (WPT) are critical for the wireless powered MEC system to realize powerful function. However, existing researches usually schedule the task offloading of WTs serially to prevent mutual signal interference, and suffer from high task offloading time. Hence, to lower the task offloading time, we adopt the successive interference cancellation (SIC) technology and realize task concurrent offloading of multiple WTs to the edge server (ES). Specifically, we study the SIC-enabled online task concurrent offloading problem with the aim of optimizing the total task completion time. We prove this optimization problem to be NP-hard, and decompose this problem to reduce the problem solving difficulty. With the support of the SIC and deep reinforcement learning (DRL) technology, we present an efficient and intelligent algorithm named SIOA. Our SIOA algorithm provides online offloading decision generating strategies for WTs through the idea of task concurrent offloading and a well-devised DRL structure. Moreover, our SIOA algorithm assigns the ES’s WPT time via a feasible area analysis approach. Our SIOA algorithm can offer demonstrable feasibility assurance, and requires lower task completion time than existing baseline algorithms with low program running time, which is verified by experiments on a real data set.

The vaccination rate for seasonal influenza remains low in most regions of China. It is essential to understand the factors that associated with the low influenza vaccination rate in various populations after the COVID-19 pandemic. A cross-sectional survey was conducted with residents in Pudong New Area, Shanghai, China. Respondents' vaccination condition during the 2021-2022 flu season and the reasons for receiving or not receiving influenza vaccine were investigated. Binary logistic regression was conducted to explore potential factors influencing vaccination uptake. 2,476 of 14,001 respondents received an influenza vaccine, with a total coverage of 17.68% (95% CI: 17.05%, 18.32%). Children had the highest vaccination coverage (35.68%; 95% CI: 34.02, 37.33), followed by adults (12.75%; 95% CI: 11.91%, 13.58%) and elderly individuals (11.70%, 95% CI: 10.78%, 12.62%). For children, lower household income was an significant promoting factor. For adults, factors significantly associated with vaccination were household income, sex, and education level. For elderly, factors significantly associated with vaccination were household income, education level, living state, and having underlying diseases. (P < .05)The main reason for vaccine hesitancy among children was worried about side effects (21.49%), for adults and elderly was self-rated good health (adults: 37.14%, elderly people: 30.66%). The overall influenza vaccination coverage rate in Shanghai, especially among elderly individuals, is lower than many developed countries. Appropriate strategies and programs targeting different populations need to be implemented to enhance influenza vaccine coverage.

One of the prominent applications of RFID technology is stocktaking. In many scenarios, the users concern about the stocktaking of categories of objects rather than their individual IDs. Existing approaches that utilize the inventory list for rapid identification can also be used for the stocktaking of categories. However, the number of missing categories of objects varies in different applications and affects the efficiency of stocktaking. Moreover, it is not uncommon that some unexpected tags that are not in the inventory list to appear in the target area and cause mistaken stocktaking. In this study, we investigate the category stocktaking problem with unexpected tags and varying category missing rate. We exploit the compressed Bloom filter to deactivate unexpected tags, dynamically adapt to different category missing rate by applying different protocols, and adjust the parameters of these protocols. We perform extensive simulations to validate the performance of our proposed approach. The results show that our approach significantly outperforms the state-of-the-art solutions.

Reinfection with severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) variants due to immune escape is challenging for the global response to the pandemic. We estimated the Omicron reinfection prevalence among people who had a previous SARS-CoV-2 infection in Shanghai, China. We conducted a telephone survey in December 2022 with those who had previously been infected with Omicron between March and May 2022. Information on their demographics, coronavirus disease 2019 (COVID-19) testing, and vaccination history was collected. The overall and subgroup reinfection rates were estimated and compared. Among the 1981 respondents who were infected between March and May 2022, 260 had positive nucleic acid or rapid antigen tests in December 2022, with an estimated reinfection rate of 13.1% (95% confidence interval [95% CI]: 11.6–14.6). The reinfection rate for those who had a booster vaccination was 11.4% (95% CI: 9.2–13.7), which was significantly lower than that for those with an incomplete vaccination series (15.2%, 95% CI: 12.3–18.1) (adjusted odds ratio [aOR]: 0.579; 95% CI: 0.412–0.813). Reinfection with the Omicron variant was lower among individuals with a previous SARS-CoV-2 infection and those who had a booster vaccination, suggesting that hybrid immunity may offer protection against reinfection with Omicron sublineages.