Synthetic data generation plays a crucial role in many areas where data is scarce and privacy/confidentiality is a significant concern. Generative Adversarial Networks (GANs), arguably one of the most widely used data synthesis techniques, allow for the training of a model (i.e., generator) that can generate real-looking data by playing a min-max game with a discriminator model. When multiple organizations are reluctant to share their sensitive data, GANs models can be trained in a federated manner, commonly with the use of differential privacy (DP). In order to achieve a reasonable level of model utility, DP trades privacy exhibiting vulnerability to various attacks (e.g., membership inference attack). In this paper, we propose a hybrid solution, PP-FedGAN, to the asynchronous federated, privacy-preserving training of GANs models by combining the CKKS homomorphic encryption (HE) scheme with differential privacy. The addition of HE results in around 10 seconds of overhead on the client side per round and 115 seconds on the entire training procedure. We also analyze the security of PP-FedGAN under the honest-but-curious security model. Where stronger security guarantees are required, our proposal presents a better alternative to solutions that only employ DP.

This paper presents an experimental investigation on the dynamic mechanical performance of S30408 austenitic stainless steel (ASS) under elevated temperatures, which is essential for determining the behaviour of structures made with this type of steel subjected to the coupled fire and impact/explosion. For this purpose, the quasi-static and dynamic compression tests using Split Hopkinson Pressure Bar (SHPB) were conducted under temperatures of 20–600 °C and strain rates from 0.001 to 3000 s⁻¹. In addition, the corresponding microstructures of tested samples were observed. The stress–strain responses, strain rate and temperature effects as well as the microstructural evolutions were analyzed. Test results show that the stress–strain responses are sensitive to the strain rate and temperature. The strain-rate sensitivity coefficient increases as the strain rate and temperature rise. The microstructural observation reveals that the grain dimension declines with an increment of strain rate or a decreasing temperature. Finally, the dynamic compressive stress–strain models for S30408 ASS under 20–600 °C were suggested on the basis of the Johnson-Cook (J-C) model and have been proved to give a reasonable prediction.

The global spread of COVID-19 has caused a huge number of confirmed cases and deaths, which in return leads to a plethora of mental disorders across the world. In order to address citizens’ psychological problems, government agencies in many countries have employed AI-based chatbots to provide mental health services. However, there is a limited understanding of the determinants affecting citizens’ user experience and user satisfaction when mental health services supported by chatbots are provided. Thus, based on the Theory of Consumption Values (TCV), this study proposes an analytical framework to investigate the factors that are important to citizens’ user experience and user satisfaction when they interact with mental health chatbots. Analysis of data collected from 295 chatbot users in Wuhan and Chongqing reveals that personalization, enjoyment, learning, and condition are positively related to user experience and user satisfaction. However, voice interaction fails to devote to citizens’ user experience and user satisfaction. Thus, government agencies and their AI service contractors should enhance the functions and systems of mental health chatbots to ensure citizens’ user experience and user satisfaction. Also, they should more positively promote the use of mental health chatbots during the public health emergency.