A novel approach based on recurrent neural networks applied to nonlinear systems optimization

This paper presents an efficient approach based on recurrent neural network for solving nonlinear optimization. More specifically, a modified Hopfield network is developed and its internal parameters are computed using the valid subspace technique. These parameters guarantee the convergence of the network to the equilibrium points that represent an optimal feasible solution. The main advantage of the developed network is that it treats optimization and constraint terms in different stages with no interference with each other. Moreover, the proposed approach does not require specification of penalty and weighting parameters for its initialization. A study of the modified Hopfield model is also developed to analyze its stability and convergence. Simulation results are provided to demonstrate the performance of the proposed neural network.     

微量元素制剂辅助治疗复发性口疮疗效观察

报道了１０３例ＲＡＵ患者使用微量元素制剂辅助治疗的临床疗效观察。将两个实验组分别与对照组比较。结果显示：应用锌、铁制剂辅助雷公藤、复方丹参等治疗的实验组Ｉ效果优于雷公藤、复方丹参对照组（Ｐ〈０．０１），而仅用锌、铁制剂治疗的实验组Ⅱ临床效果比对照组差（Ｐ〈０．０１），但锌、铁制剂有改善ＲＡＵ临床症状作用，而不能较好地控制其复发。     

中西医结合治疗复发性口疮102例疗效观察

用参麦汤合葡萄糖酸锌片治疗复发性口疮１０２例，经过３个月治疗，观察到该药能有效地改善临床症状，降低复发率，明显地提高体内锌的水平，与维生素类对照组比较，疗效有显著性差异（Ｐ＜０．０１）。     

Unilateral recurrent laryngeal nerve paralysis:The importance of “preoperative” voice therapy

This preliminary retrospective study of 19 female patients and 22male patients with unilateral recurrent nerve lesions demonstrated the promise of objective measurements in predicting the need for surgery, the efficacy of voice therapy in ameliorating vocal symptoms, and the effects of therapy in conjunction with surgery. Sixty-eight percent (68%) of the female patients and 64% of the male patients did not elect to have surgery. Outcome satisfaction of nonsurgical and surgical patients appeared to be similar. The data from this study support the importance of preoperative therapy for patients with unilateral vocal fold paralysis.     

Recent Results on Recurrent Solutions and Limit Distributions of SDEs

The limit distribution for homogeneous Markov processes is studied extensively and well understood, but it is not the case for inhomogeneous Markov processes. In this paper, we review some recent results on inhomogeneous Markov processes generated by non-autonomous stochastic (partial) differential equations (SDE in short). Under some suitable conditions, we show that the distribution of recurrent solutions of SDEs constitutes the limit distribution of the corresponding inhomogeneous Markov processes.     

Structure activity relationship of bergenin,p-hydroxybenzoyl bergenin, 11-O-galloylbergenin as potent antioxidant and urease inhibitor isolated from Bergenia ligulata

Ethanol extract of the aerial parts of Bergenia ligulata was subjected to solvent–solvent separation followed by various chromatographic techniques that lead to isolation of bergenine (1), p-hydroxybenzoyl bergenin (2), 11-O-galloylbergenin (3) and methyl gallate (4) as major constituents. Ethyl acetate fraction showed a dose-dependent urease inhibitory pattern with IC₅₀ value of 54μg/mL. Structures of compounds 1 and 3 were established by XRD and 2, 4 by NMR. All these compounds were subjected to DPPH scavenging activity, reducing power assay and urease inhibitory activity. The EC₅₀ 7.45 ± 0.2 μg/mL and 5.39 ± 0.28 μg/mL values in terms of antioxidant and reducing power, respectively, were less for 3. Compounds 1–3 showed moderate to significant urease inhibitory potential with IC₅₀ 57.1 ± 0.7, IC₅₀ 48.4 ± 0.3 and 38.6 ± 1.5. Antioxidant activities and urease inhibitory potential were investigated and compound 3 was found to be the most active.     

中医药膳防治复发性阿弗他溃疡

通过近年来有关中医药膳防治复发性阿弗他溃疡的研究,从中西医角度论述了复发性阿弗他溃疡的发病机理,挖掘出有效防治复发性阿弗他溃疡的中医药膳,总结了中医药膳的功能成分如何作用于复发性阿弗他溃疡,归纳了可行的中医药膳调理的方法和应用原则,为开展药膳干预复发性阿弗他溃疡发病提供理论基础.     

Recurrent points and non-wandering points of graph maps

Let G be a graph and f:G→G be a continuous map. Denote by P(f), R(f) and Ω(f) the sets of periodic points, recurrent points and non-wandering points of f, respectively. In this paper we show that: (1) If L=(x,y) is an open arc contained in an edge of G such that {fm(x),fk(y)}⊂(x,y) for some m,k∈N, then R(f)∩(x,y)≠∅; (2) Any isolated point of P(f) is also an isolated point of Ω(f); (3) If x∈Ω(f)−Ω(fn) for some n∈N, then x is an eventually periodic point. These generalize the corresponding results in W. Huang and X. Ye (2001) [9] and J. Xiong (1983, 1986) [17] and [19] on interval maps or tree maps.     

基于深度学习的太赫兹时域光谱识别研究

太赫兹时域光谱技术,由于其具有物质“指纹谱”特性,是一种可以快速无损地鉴别物质的重要手段,在毒品和爆炸物的无损检测等方面有广阔的应用前景。其中,光谱识别是太赫兹时域光谱技术应用研究的重要方向之一。现有的光谱识别方法多是依靠手工选取特征后进行机器学习分类,或是通过设置吸收峰阈值门限进行判断。由于一些物质在太赫兹波段内并没有明显的吸收峰特征,同时样品浓度、空气湿度、各类噪声等会对太赫兹时域光谱造成干扰从而使信噪比下降,这些方法并不能很好地适应,并且物质类别和数量的增加也会导致计算量不断增加。近年来,随着深度学习技术兴起,以卷积神经网络（CNN）和循环神经网络（RNN)为代表的方法在计算机视觉和自然语言处理等领域得到广泛应用,相比于传统的机器学习方法其效果有了很大的提升。由于深度学习技术强大的非线性分类能力,基于RNN和CNN设计了两个网络用于光谱识别：基于RNN的一维谱线分类网络和基于CNN的二维谱图分类网络。模拟实际应用场景,在非真空环境下采集了12种物质的两万多个光谱数据作为训练集和测试集。在分析了样品浓度、空气湿度对光谱特征的影响后,使用S-G（Savitzky-Golay）滤波对光谱进行降噪。实验结果表明,对比未处理和经过S-G预处理的数据,处理后的光谱特征更加明显,识别准确率更高;与传统的机器学习算法k最近邻（k-NN）方法相比,RNN和CNN方法在测试集上有更好的准确率,且算法速度更快;对于光谱识别,CNN方法比RNN方法能够更好地克服噪声的影响。因此,深度学习技术可以对太赫兹时域光谱进行快速有效的识别,能够为新型无损安全检查技术提供理论和实验基础。