skip to main content


Search for: All records

Creators/Authors contains: "Wang, Hui"

Note: When clicking on a Digital Object Identifier (DOI) number, you will be taken to an external site maintained by the publisher. Some full text articles may not yet be available without a charge during the embargo (administrative interval).
What is a DOI Number?

Some links on this page may take you to non-federal websites. Their policies may differ from this site.

  1. Wang, Yan ; Yang, Hui (Ed.)
    Abstract

    The scarcity of measured data for defect identification often challenges the development and certification of additive manufacturing processes. Knowledge transfer and sharing have become emerging solutions to small-data challenges in quality control to improve machine learning with limited data, but this strategy raises concerns regarding privacy protection. Existing zero-shot learning and federated learning methods are insufficient to represent, select, and mask data to share and control privacy loss quantification. This study integrates differential privacy in cybersecurity with federated learning to investigate sharing strategies of manufacturing defect ontology. The method first proposes using multilevel attributes masked by noise in defect ontology as the sharing data structure to characterize manufacturing defects. Information leaks due to the sharing of ontology branches and data are estimated by epsilon differential privacy (DP). Under federated learning, the proposed method optimizes sharing defect ontology and image data strategies to improve zero-shot defect classification given privacy budget limits. The proposed framework includes (1) developing a sharing strategy based on multilevel attributes in defect ontology with controllable privacy leaks, (2) optimizing joint decisions in differential privacy, zero-shot defect classification, and federated learning, and (3) developing a two-stage algorithm to solve the joint optimization, combining stochastic gradient descent search for classification models and an evolutionary algorithm for exploring data-sharing strategies. A case study on zero-shot learning of additive manufacturing defects demonstrated the effectiveness of the proposed method in data-sharing strategies, such as ontology sharing, defect classification, and cloud information use.

     
    more » « less
    Free, publicly-accessible full text available November 7, 2025
  2. Abstract

    The Pacific–South American (PSA) pattern is a key mode of climate variability in the mid-to-high latitudes of the Southern Hemisphere, impacting circulation and rainfall anomalies over South America. However, the effect of South American rainfall on the PSA has not been previously explored. This study focuses on the impact of rainfall over southeastern South American (SESA) during the austral summer (December–February). Observational analyses reveal that the PSA pattern remains confined to higher southern latitudes when SESA rainfall anomalies are weak. In contrast, strong SESA rainfall anomalies can generate a quasi-stationary Rossby wave train, which represents a cross-equatorial extension of the PSA. This wave train propagates along a southwest–northeast great circle path from higher latitudes, crosses the equator, and reaches the tropical Atlantic, southern Europe, and northern Africa, inducing a wet and cool weather condition over western and southern Europe. The observed wave train can be reproduced by the linear baroclinic model (LBM) simulations. Given the PSA’s connection to tropical forcing over the central Pacific, we examine differences in the wave response to central Pacific forcing alone versus combined central Pacific and SESA forcings. By incorporating SESA forcing, the wave train originally triggered by central Pacific forcing is amplified and extended. Our findings confirm the significant role of SESA rainfall anomalies in extending the PSA pattern to the Northern Hemisphere and highlight the South American continent as a land bridge that links circulation anomalies across the Pacific and Atlantic Oceans and the Southern and Northern Hemispheres.

     
    more » « less
  3. Biogenic isoprene emissions from herbaceous plants are generally lower than those from trees. However, our study finds widespread isoprene emission in herbaceous sedge plants, with a stronger temperature response surpassing current tree-derived models. We measured and compared isoprene emissions from sedges grown in different climatic zones, all showing an exponential temperature response with a Q10 range of 7.2 to 12, significantly higher than the Q10 of about 3 for other common isoprene emitters. The distinct temperature sensitivity of sedges makes them a hidden isoprene source, significant during heat waves but not easily detected in mild weather. For instance, isoprene emissions fromCarex praegraciliscan increase by 320% with a peak emission of over 100 nmol m−2s−1compared to preheat wave emissions. During heat waves, the peak isoprene emissions fromC. praegraciliscan match those fromLophostemon confertus, a commonly used street tree species which is considered the dominant urban isoprene source due to higher biomass and emission capacities. This surge in isoprene from globally distributed sedges, including those in urban landscapes, could contribute to peak ozone and aerosol pollutants during heat waves.

     
    more » « less
    Free, publicly-accessible full text available November 5, 2025
  4. Free, publicly-accessible full text available June 24, 2025
  5. Free, publicly-accessible full text available August 20, 2025
  6. In physics, it is crucial to identify operational measurement procedures to give physical meaning to abstract quantities. There has been significant effort to define time operationally using quantum systems, but the same has not been achieved for space. Developing an operational procedure to obtain information about the location of a quantum system is particularly important for a theory combining general relativity and quantum theory, which cannot rest on the classical notion of spacetime. Here, we take a first step towards this goal, and introduce a model to describe an extended material quantum system working as a position measurement device. Such a quantum ruler is composed ofNharmonically interacting dipoles and serves as a (quantum) reference system for the position of another quantum system. We show that we can define a quantum measurement procedure corresponding to the superposition of positions, and that by performing this measurement we can distinguish when the quantum system is in a coherent or incoherent superposition in the position basis. The model is fully relational, because the only meaningful variables are the relative positions between the ruler and the system, and the measurement is expressed in terms of an interaction between the measurement device and the measured system.

     
    more » « less
    Free, publicly-accessible full text available May 6, 2025
  7. Graph Neural Networks (GNNs) have emerged as powerful tools for processing graph-structured data, enabling applications in various domains. Yet, GNNs are vulnerable to model extraction attacks, imposing risks to intellectual property. To mitigate model extraction attacks, model ownership verification is considered an effective method. However, throughout a series of empirical studies, we found that the existing GNN ownership verification methods either mandate unrealistic conditions or present unsatisfactory accuracy under the most practical settings—the black-box setting where the verifier only requires access to the final output (e.g., posterior probability) of the target model and the suspect model. Inspired by the studies, we propose a new, black-box GNN ownership verification method that involves local independent models and shadow surrogate models to train a classifier for performing ownership verification. Our method boosts the verification accuracy by exploiting two insights: (1) We consider the overall behaviors of the target model for decision-making, better utilizing its holistic fingerprinting; (2) We enrich the fingerprinting of the target model by masking a subset of features of its training data, injecting extra information to facilitate ownership verification. To assess the effectiveness of our proposed method, we perform an intensive series of evaluations with 5 popular datasets, 5 mainstream GNN architectures, and 16 different settings. Our method achieves nearly perfect accuracy with a marginal impact on the target model in all cases, significantly outperforming the existing methods and enlarging their practicality. We also demonstrate that our method maintains robustness against adversarial attempts to evade the verification. 
    more » « less
    Free, publicly-accessible full text available May 19, 2025
  8. Free, publicly-accessible full text available August 1, 2025