Search

As one of the representative models in the field of image generation, generative adversarial networks (GANs) face a significant challenge: how to make the best trade-off between the quality of generated images and training stability. The U-Net based GAN (U-Net GAN), a recently developed approach, can generate high-quality synthetic images by using a U-Net architecture for the discriminator. However, this model may suffer from severe mode collapse. In this study, a stable U-Net GAN (SUGAN) is proposed to mainly solve this problem. First, a gradient normalization module is introduced to the discriminator of U-Net GAN. This module effectively reduces gradient magnitudes, thereby greatly alleviating the problems of gradient instability and overfitting. As a result, the training stability of the GAN model is improved. Additionally, in order to solve the problem of blurred edges of the generated images, a modified residual network is used in the generator. This modification enhances its ability to capture image details, leading to higher-definition generated images. Extensive experiments conducted on several datasets show that the proposed SUGAN significantly improves over the Inception Score (IS) and Fréchet Inception Distance (FID) metrics compared with several state-of-the-art and classic GANs. The training process of our SUGAN is stable, and the quality and diversity of the generated samples are higher. This clearly demonstrates the effectiveness of our approach for image generation tasks. The source code and trained model of our SUGAN have been publicly released. [ABSTRACT FROM AUTHOR]

The remanufacturing industry can realize sustainability by saving resources and protecting the environment. The remanufactured products have a high similarity with the new products, and how they affect consumers' evaluation of the new products is an important issue. This paper discusses the relationship between the similarity of remanufactured products and consumers' willingness to purchase new products, with a mediating effect of perceived quality and a moderating effect of brand reputation. The empirical analysis based on 498 valid samples shows our conclusions below. First, the similarity of remanufactured products has a significant negative impact on the purchase intention of new products. Second, the perceived quality of new products plays a partial mediating role between similarity and purchase intention. Third, brand reputation negatively moderates the relationship between the similarity of remanufactured products and the perceived quality of new products. Furthermore, brand reputation also moderates the mediating effect of the similarity of remanufactured products on the purchase intention of new products through the perceived quality of new products. This paper suggests a new perspective for the research of remanufactured products and provides an important reference for the healthy and sustainable development of recycling and remanufacturing industry. [ABSTRACT FROM AUTHOR]

Maleic anhydride (MAH)–divinyl benzene (DVB) multi-monomer melt-grafting onto isotactic polybutene-1 (iPB-1) was carried out in a torque rheometer. The effects of dicumyl peroxide (DCP), MAH, and DVB concentrations, and temperature, on the reaction, were investigated. The optimized conditions were 170 °C, DVB/MAH = 4:6 (mass ratio). DVB as a comonomer enhanced the grafting degree (Gd) and grafting efficiency (Ge) of iPB-g-MAH better than styrene. The initiator DCP had little effect on Gd as its concentration over 0.2 phr, but the grafts' melt flow rate (MFR) increased significantly, and relative molecular weight decreased remarkably with increased DCP concentration. With increasing Gd, the contact angle of grafts with water decreased, and there was a larger crystallization rate. The study of iPB-1 and iPB-g-MAH (Gd = 1.5%)-modified polyamide 6 (PA6) showed that iPB-g-MAH had an obviously toughening effect on PA6. With increasing iPB-g-MAH concentration, the blends of impact strength and elongation at break increased obviously, tensile strength decreased slightly, and MFR decreased prominently, which greatly slowed the processing degradation of PA6. The properties of iPB-1/PA6 blends deteriorated. Both DSC curves and SEM micrographs confirmed that PA6/iPB-g-MAH blends had much better compatibility than PA6/iPB. The reason was that the anhydride group in iPB-g-MAH reacted with amide group in PA6 to improve the compatibility between two phases, and iPB-g-MAH is an excellent modifier for PA6. [ABSTRACT FROM AUTHOR]