Search

Wheat variety identification from hyperspectral images holds significant importance in both fine breeding and intelligent agriculture. However, the discriminatory accuracy of some techniques is limited due to insufficient datasets, data redundancy, and noise interference. To address these issues, we propose a wheat variety identification framework called generate adversarial-driven cross-aware network (GACNet), comprising a semi-supervised generative adversarial network (GAN) for data augmentation and a cross-aware attention network (CAANet) for variety identification. First, the semi-supervised GAN (SSGAN) alleviates data scarcity by generating fake hyperspectral images as realistically as possible through learning the distribution hypothesis of real hyperspectral images, while the discriminator distinguishes between real and fake hyperspectral images. Subsequently, the CAANet is employed for wheat variety identification, which leverages a cascading cross-learning of 3-D and 2-D convolutions to fully exploit spectral, spatial, and texture features and refines the features through an embedded attention mechanism in the cross-convolutional module. Additionally, we constructed a hyperspectral wheat variety dataset (HWVD) comprising 4560 samples of 19 categories. Extensive experiments on our dataset demonstrate that our GACNet outperforms state-of-the-art methods for wheat variety identification. The HWVD will be made available.

Flexible adaptation to differentiated quality of service (QoS) is quite important for future 6G network with a variety of services. Mobile ad hoc networks (MANETs) are able to provide flexible communication services to users through self-configuration and rapid deployment. However, the dynamic wireless environment, the limited resources, and complex QoS requirements have presented great challenges for network routing problems. Motivated by the development of artificial intelligence, a deep reinforcement learning-based collaborative routing (DRLCR) algorithm is proposed. Both routing policy and subchannel allocation are considered jointly, aiming at minimizing the end-to-end (E2E) delay and improving the network capacity. After sufficient training by the cluster head node, the Q-network can be synchronized to each member node to select the next hop based on local observation. Moreover, we improve the performance of training by considering historical observations, which can improve the adaptability of routing policies to dynamic environments. Simulation results show that the proposed DRLCR algorithm outperforms other algorithms in terms of resource utilization and E2E delay by optimizing network load to avoid congestion. In addition, the effectiveness of the routing policy in a dynamic environment is verified.

Nephron progenitor cells (NPCs) self-renew and differentiate into nephrons, the functional units of the kidney. Here, manipulation of p38 and YAP activity allowed for long-term clonal expansion of primary mouse and human NPCs and induced NPCs (iNPCs) from human pluripotent stem cells (hPSCs). Molecular analyses demonstrated that cultured iNPCs closely resemble primary human NPCs. iNPCs generated nephron organoids with minimal off-target cell types and enhanced maturation of podocytes relative to published human kidney organoid protocols. Surprisingly, the NPC culture medium uncovered plasticity in human podocyte programs, enabling podocyte reprogramming to an NPC-like state. Scalability and ease of genome editing facilitated genome-wide CRISPR screening in NPC culture, uncovering genes associated with kidney development and disease. Further, NPC-directed modeling of autosomal-dominant polycystic kidney disease (ADPKD) identified a small-molecule inhibitor of cystogenesis. These findings highlight a broad application for the reported iNPC platform in the study of kidney development, disease, plasticity, and regeneration.

A distributed reinforcement learning (RL) based resource management framework is proposed for a mobile edge computing (MEC) system with both latency-sensitive and latency-insensitive services. We investigate joint optimization of both computing and radio resources to achieve efficient on-demand matches of multi-dimensional resources and diverse requirements of users. A multi-objective integer programming problem is formulated by two subproblems, i.e., access point (AP) selection and subcarrier allocation, which can be solved jointly by our proposed distributed RL-based approach with a heuristic iteration algorithm. The proposed algorithm allows for the reduction in complexity since each user needs to consider only its own selection of AP without knowing full global information. Simulation results show that our algorithm can achieve near-optimal performance while reducing computational complexity significantly. Compared with other algorithms that only optimize either of the two sub-problems, the proposed algorithm can serve more users with much less power consumption and content delivery latency.

The concept of edge network caching has been proposed to alleviate the excessive pressure on the core networks. Furthermore, video segment caching technology, a method to cut the whole video into segments and cache them separately, has brought a novel idea to solve the caching problem in the smaller space for massive data. The adoption of segment caching in edge networks will divide the simple video transmission process into two coupling stages because of separate data caching, which leads to more complicated resource allocation. In this paper, this problem is discussed, and its mathematical model is established to minimize the energy consumption of video transmissions. By introducing an efficient prediction window of channel fading, an optimal dynamic scheduling algorithm based on Q-learning is proposed to minimize power consumption while ensuring smooth video streaming. The proposed Q-learning algorithm is simulated and the impacts of channel state, target video bit rate and large-scale channel parameter are evaluated. Simulation results show that the proposed method can effectively reduce the total power consumption while ensuring the smooth playback of video service, thanks to the fact that the proposed method is intelligent which can effectively utilize idle resources in favorable channel states.

Drug repositioning represents a cost- and time-efficient strategy for drug development. Artificial intelligence-based algorithms have been applied in drug repositioning by predicting drug-target interactions in an efficient and high throughput manner. Here, we present a workflow of in silicodrug repositioning for host-based antivirals using specially defined targets, a refined list of drug candidates, and an easily implemented computational framework. The workflow described here can also apply to more general purposes, especially when given a user-defined druggable target gene set.