Recently, significant progress has been made in sequential recommendation with deep learning. Existing neural sequential recommendation models usually rely on the item prediction loss to learn model parameters or data representations. However, the model trained with this loss is prone to suffer from data sparsity problem. Since it overemphasizes the final performance, the association or fusion between context data and sequence data has not been well captured and utilized for sequential recommendation. To tackle this problem, we propose the model S 3-Rec, which stands for Self-Supervised learning for Sequential Recommendation, based on the self-attentive neural architecture. The main idea of our approach is to utilize the intrinsic data correlation to derive self-supervision signals and enhance the data representations via pre-training methods for improving sequential recommendation. For our task, we devise four auxiliary self-supervised objectives to learn the correlations among attribute, item, subsequence, and sequence by utilizing the mutual information maximization (MIM) principle. MIM provides a unified way to characterize the correlation between different types of data, which is particularly suitable in our scenario. Extensive experiments conducted on six real-world datasets demonstrate the superiority of our proposed method over existing state-of-the-art methods, especially when only limited training data is available. Besides, we extend our self-supervised learning method to other recommendation models, which also improve their performance. CCS CONCEPTS • Information systems → Recommender systems. † Equal contribution.
As an important and challenging problem in machine learning and computer vision, neural network acceleration essentially aims to enhance the computational efficiency without sacrificing the model accuracy too much. In this paper, we propose a progressive blockwise learning scheme for teacher-student model distillation at the subnetwork block level. The proposed scheme is able to distill the knowledge of the entire teacher network by locally extracting the knowledge of each block in terms of progressive blockwise function approximation. Furthermore, we propose a structure design criterion for the student subnetwork block, which is able to effectively preserve the original receptive field from the teacher network. Experimental results demonstrate the effectiveness of the proposed scheme against the state-of-the-art approaches.
We quantify the impact of residual hardware impairments (RHI) on a non-orthogonal multiple access (NOMA)-based relaying network, where a source communications simultaneously with multiple users via an amplify-and-forward relay. Specifically, exact and asymptotic expressions for the outage probability are first derived in closed-form over Nakagami-m fading channels, accounting for RHI at the source, relay, and all users. Our results show that the outage performance loss induced by RHI is small in the low signal-to-noise ratio (SNR) regime or/and at low target rates, however it is significant at high SNRs or/and target rates. Furthermore, we present new tight analytical approximated and asymptotic expressions for the system ergodic sum rate (ESR). For comparison, we also discuss the ESR of the conventional hardware-impaired relaying system employing orthogonal multiple access (OMA) transmission and obtain the asymptotic high-SNR expression of the ESR. The provided numerical results demonstrate that in the absence of RHI, the ESR in either the NOMA or the OMA system increases monotonically with the increase of SNR, whereas an unavoidable ESR ceiling is introduced in the hardware-impaired scenario for both systems. Notably, since the ESR ceiling value is only depended on the RHI levels, the NOMA-based and OMA-based systems achieve the same ESR performance in high SNRs. INDEX TERMS Non-orthogonal multiple access, residual hardware impairments, outage probability, ergodic sum rate.
Ecological environment issues put forward higher requirements for enterprises to assume environmental responsibilities, and stimulating employee green behavior (EGB) to practice the concept of green development is of great significance. EGB has become the focus of academic attention. EGB is divided into voluntary green behavior (VGB) and task-related green behavior (TGB). However, existing studies have not distinguished the impact mechanism of green human resource management (GHRM) on employee VGB and TGB. Based on self-determination theory and social identity theory, this study discusses how GHRM affects VGB and TGB. This study used a questionnaire survey and collected valid data of 228 employees from manufacturing enterprises in China for empirical analysis. Results show that GHRM positively affects VGB and TGB, environmental belief (EB) mediates the positive relationship between GHRM and VGB, and green organizational identity (GOI) mediates the positive relationship between GHRM and TGB. Theoretical contributions, practical implications, and future research are also discussed.
A novel method for computer-generated rainbow holograms (CGRHs) of full-color objects is proposed. First, a new algorithm for fabricating full-color CGRHs of real-existing objects is proposed based on the interrelationship between coding of a CGRH and reconstruction of the hologram. Second, a color rainbow hologram for a real-existing object is generated by combining the proposed algorithm and computer-generated hologram generating system. Finally, the hologram is outputted by an auto-microfilming system. The principle of the algorithm, the process of hologram calculation, and the hologram generating system for real-existing objects and experimental results are presented. The experimental results demonstrate that the new method is feasible.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.