Power amplifiers are important blocks in communication systems and their efficiency and linearity are critical performance parameters. In this paper, we present a preliminary investigation into RF feedback systems designed to improve the efficiency-linearity trade-off for power amplifiers. We demonstrate very large potential improvement in trade-off due to feedback and significant actual improvement. In order to fully realise the potential improvement, further work needed is outlined.

Nuclei segmentation and classification are two basic and essential tasks in computer-aided diagnosis of digital pathology images, and those deep-learning-based methods have achieved significant success. Unfortunately, most of the existing studies accomplish the two tasks by splicing two related neural networks directly, resulting in repetitive computation efforts and a redundant-and-large neural network. Thus, this paper proposes a lightweight deep learning framework (GSN-HVNET) with an encoder–decoder structure for simultaneous segmentation and classification of nuclei. The decoder consists of three branches outputting the semantic segmentation of nuclei, the horizontal and vertical (HV) distances of nuclei pixels to their mass centers, and the class of each nucleus, respectively. The instance segmentation results are obtained by combing the outputs of the first and second branches. To reduce the computational cost and improve the network stability under small batch sizes, we propose two newly designed blocks, Residual-Ghost-SN (RGS) and Dense-Ghost-SN (DGS). Furthermore, considering the practical usage in pathological diagnosis, we redefine the classification principle of the CoNSeP dataset. Experimental results demonstrate that the proposed model outperforms other state-of-the-art models in terms of segmentation and classification accuracy by a significant margin while maintaining high computational efficiency.

With the surge in the amount of data transmitted on the network, intelligent learning and other technologies have emerged to solve the problem of anomaly detection of streaming data in large data. For network security issues, based on the extraction of network traffic characteristics, network traffic classification or clustering is an important technical means to discover network faults and network attacks. In this paper, a distributed detection framework for detecting anomalous behaviors of encrypted network traffic is proposed. The intelligent router is adopted to obtain the encrypted network traffic monitoring stub, and then the neural network codec is used to adaptively learn the characteristics of the encrypted traffic and identify the abnormal behavior of the encrypted protocol traffic. A wide coverage traffic pattern extraction algorithm based on the network state sequence is designed to obtain the traffic patterns that represent the network conditions of the data center. Finally, the simulation test verifies that the model has frequent traffic patterns with priority. The performance of the network anomaly detection model is better than other detection methods, which improves the accuracy of detection and has a better recognition effect.

No abstract