No abstract

Abstract-A modified differential evolution algorithm (MDE) for pattern synthesis of antenna arrays is proposed in this paper. By employing the novel strategies of best of random mutation and randomized local search, the convergence of standard differential evolution algorithm (SDE) is significantly accelerated. Five standard benchmark functions are optimized to testify the proposed algorithm by comparison with several other optimization algorithms. The numerical results verify the superior performance of the proposed MDE. Furthermore, the MDE is applied to two pattern synthesis examples, including a linear array and a cylindrical conformal array. Experiment results demonstrate that the proposed MDE has better performance than the other optimization methods in both of these two examples, which indicate the proposed algorithm is a competitive optimization algorithm in pattern synthesis.

Abstract-In this paper, a new approach using quasi-self-complementary antenna (QSCA) to reduce the wideband mutual coupling is proposed and discussed. QSCA element proposed in this paper is composed of a semi-circular radiation patch and a complementary-cut ground plane, which is easy to achieve ultra-wideband operation because its impedance is frequency independent. The proposed compact four-element ultra-wideband (UWB) multiple-input multiple-output (MIMO) array consists of four QSCA elements, and by arranging them anticlockwise, a good impedance matching and high port-to-port isolation (|S 11 | covers 2.95-12.1 GHz with |S 21 | = |S 31 | ≤ −15 dB, |S 41 | ≤ −17.8 dB) can be achieved. Notably, the isolation is obtained without using any other decoupling methods and totally benefits from the asymmetrical radiation property of QSCA. As an example, the proposed four-element UWB MIMO array is fabricated and tested. And the measured radiation pattern, gains and total efficiencies are displayed and show good performances which make it a nice candidate for future UWB diversity applications.

No abstract

End-to-end task-oriented dialog systems have attracted vast amounts of attention in recent years, mainly because of their ease of training. However, such an end-to-end model requires a large number of labeled dialogs to train. Labeled dialogs are always difficult to obtain in real-world settings. We propose a domain adaptive end-to-end task-oriented dialog model that transfers knowledge in source domains to a target domain with limited training samples. Specifically, we design a domain adaptive filter in the encoder-decoder model to reduce useless features in source domains and preserve common features. A domain adaptive amplifier is designed to enhance the target domain impact. We evaluate our method on both synthetic dialog and human-human dialog datasets and achieve state-of-the-art results. INDEX TERMS Neural network, dialogue system, domain adaptation, encoder-decoder model.