On Energy Efficiency of Hybrid NOMA

This paper aims to prove the significant superiority of hybrid non-orthogonal multiple access (NOMA) over orthog onal multiple access (OMA) in terms of energy efficiency. In particular, a novel hybrid NOMA scheme is proposed in which a user can transmit signals not only by using its own time slot but also by using the time slots of other users. The data rate maximization problem is studied by optimizing the power allocation, where closed-form solutions are obtained. Further more, the conditions under which hybrid NOMA can achieve a higher instantaneous data rate with less power consumption than OMA are obtained. It is proved that the probability that hybrid NOMA can achieve a higher instantaneous data rate with less power consumption than OMA approaches one in the high SNR regime, indicating the superiority of hybrid NOMA in terms of power efficiency. Numerical results are also provided to verify the developed analysis and also to demonstrate the superior performance of hybrid NOMA.

Dcl-Net: Dual Contrastive Learning Network for Semi-Supervised Multi-Organ Segmentation

Semi-supervised learning is a sound measure to relieve the strict demand of abundant annotated datasets, especially for challenging multi-organ segmentation . However, most existing SSL methods predict pixels in a single image independently, ignoring the relations among images and categories. In this paper, we propose a two-stage Dual Contrastive Learning Network for semi-supervised MoS, which utilizes global and local contrastive learning to strengthen the relations among images and classes. Concretely, in Stage 1, we develop a similarity-guided global contrastive learning to explore the implicit continuity and similarity among images and learn global context. Then, in Stage 2, we present an organ-aware local contrastive learning to further attract the class representations. To ease the computation burden, we introduce a mask center computation algorithm to compress the category representations for local contrastive learning. Experiments conducted on the public 2017 ACDC dataset and an in-house RC-OARs dataset has demonstrated the superior performance of our method.