Generating Questions Based on Semi-Automated and End-to-End Neural Network

With the emergence of large-scale knowledge base, how to use triple information to generate natural questions is a key technology in question answering systems. The traditional way of generating questions require a lot of manual intervention and produce lots of noise. To solve these problems, we propose a joint model based on semi-automated model and End-to-End neural network to automatically generate questions. The semi-automated model can generate question templates and real questions combining the knowledge base and center graph. The End-to-End neural network directly sends the knowledge base and real questions to BiLSTM network. Meanwhile, the attention mechanism is utilized in the decoding layer, which makes the triples and generated questions more relevant. Finally, the experimental results on SimpleQuestions demonstrate the effectiveness of the proposed approach.     

Iterative equalization combining algorithm in intersymbol interference channel

To realize joint optimization of spatial diversity and equalization combining in the intersymbol interference (ISI) channel, an iterative equalization combining algorithm is proposed. The proposed algorithm uses the coefficients of Turbo equalization to calculate the combination weights without estimating the signal to noise ratio in each diversity branch. The equalized symbols from different diversity branches are combined, and the extrinsic information output from the decoder is fed back to the equalizers, so as to exchange soft information between the equalizers and the decoder. The performance of the proposed algorithm is analyzed using the extrinsic information transfer (EXIT) chart and verified by simulations. Results show that our approach fully exploits time domain information from the multipath channel and spatial domain information from multi receiving antennas, which efficiently improve the performance of the receiver in the severe ISI channel.     

Enhance cycling performance of LiMn2O4 cathode by Sr2+ and Cr3+ doping

Spinel LiSr_0·1Cr_0·1Mn_1·8O₄ was synthesised by high temperature solid state method in order to enhance the electrochemical performance. The LiSr_0·1Cr_0·1Mn_1·8O₄ (LSCMO) materials were characterised by X-ray diffraction (XRD), scanning electron microscopy (SEM) and electrochemical tests. The XRD and SEM studies confirm that LSCMO had spinel crystal structure with a space group of Fd3m, and the particle of LSCMO shows irregular shape. The cyclic voltammetry data illustrated that the heavy current charge–discharge performance of LMO was improved by Sr²⁺ and Cr³⁺ doping. The galvanostatic charge–discharge of LSCMO cathode materials was measured at 1, 5, 10 and 20 C. The results indicated that LSCMO improved the capacity retention.