Jialu Wang

Cross-domain recommendation (CDR) has emerged as a promising approach to improve click-through rate (CTR) in the target domain by effectively transferring user interests from the source domain. However, existing methods either use a uniform interest transfer function or focus on user-level personalized transfer functions, neglecting the fact that the transition of user states in the target domain also influence the interests in the source domain. To address this issue, we present User-State based Interest Transfer network (USIT), a novel method that takes into account the user state evolution. USIT contains two main components: a User-State Transition module (UST) and a State-Level Interests Transfer module (SLIT). UST models the evolution of user states by predicting the next state in the target domain. As the user's state evolves, SLIT adaptively weights the interests by interest-level mask attention in the source domain. Extensive offline experiments and online A/B tests demonstrate that our proposed USIT method significantly outperforms current state-of-the-art models in CDR scenarios. Currently, we have deployed it on NetEase Cloud Music, affecting millions of users.

With the help of deep neural networks, video super-resolution (VSR) has made a huge breakthrough. However, these deep learning-based methods are rarely used in specific situations. In addition, training sets may not be suitable because many methods only assume that under ideal circumstances, low-resolution (LR) datasets are downgraded from high-resolution (HR) datasets in a fixed manner. In this paper, we proposed a model based on Generative Adversarial Network (GAN) and edge enhancement to perform super-resolution (SR) reconstruction for LR and blur videos, such as closed-circuit television (CCTV). The adversarial loss allows discriminators to be trained to distinguish between SR frames and ground truth (GT) frames, which is helpful to produce realistic and highly detailed results. The edge enhancement function uses the Laplacian edge module to perform edge enhancement on the intermediate result, which helps further improve the final results. In addition, we add the perceptual loss to the loss function to obtain a higher visual experience. At the same time, we also tried training network on different datasets. A large number of experiments show that our method has advantages in the Vid4 dataset and other LR videos.