Microwave Humidity and Temperature Sounder (MWHTS) is one of the most important payloads on the Fengyun-3 (FY-3) satellite, and it can realize simultaneous detection of atmospheric temperature and humidity profiles. At present, the measurement data from MWHTS is not yet assimilated in the Arctic region. It is helpful to improve the initial field of the assimilation system by evaluating the accuracy of MWHTS atmospheric temperature and humidity profiles retrieval parameters in the Arctic region. In this paper, we propose a deep learning method for atmospheric temperature and humidity profiles retrieval to evaluate the accuracy of MWHTS atmospheric parameters. The results show that the RMSE of temperature profile is less than 2K, and the RMSE of humidity profile is between 10-20%. And the retrieval accuracy of LSTM is slightly better than that of DNN.

Graph classification problem is becoming one of research hotspots in the realm of graph mining, which has been widely used in cheminformatics, bioinformatics and social network analytics. Existing approaches, such as graph kernel methods and graph Convolutional Neural Network, are facing the challenges of non-interpretability and high dimensionality. To address the problems, we propose a novel recurrent attention model, called g-Inspector, which applies the attention mechanism to investigate the significance of each region to make the results interpretable. It also takes a shift operation to guide the inspector agent to discover the next relevant region, so that the model sequentially loads small regions instead of the entire large graph, to solve the high dimensionality problem. The experiments conducted on standard graph datasets show the effectiveness of our g-Inspector in graph classification problems.

Reading efficiency is one of the key factors to evaluate Radio Frequency Identification (RFID) systems. For the system using multi-branch protocols, the performance would be better if the tags are properly divided into multiple groups. This paper firstly gives the closed-form of system efficiency for binary tree algorithm. Based on the theoretical analysis, the optimal branches number is derived. An efficient multi-branch tree (EMBT) algorithm is proposed subsequently, along with a tag number estimation algorithm and performance evaluation. Both theoretical analysis and simulation results indicate that multi-branch tree algorithm has better performance than the conventional binary tree algorithm. System identification efficiency of the proposed method can achieve above 45%, while that of the binary tree algorithm is only 34.8%.

The system efficiency is one of the key factors for RFID (Radio Frequency Identification) systems. This paper gives theoretical study on binary tree algorithm that is adopted by the international standard ISO 18000-6B. The close form for calculating system efficiency is derived. Simulation is performed and results agree with theoretical one. In addition, the simplified expression for very large tag number is also provided.