劉新 陳麗麗 黃帥
摘 要:討論了一類具有傳遞時(shí)滯和分布時(shí)滯以及區(qū)間不確定性的脈沖隨機(jī)反應(yīng)-擴(kuò)散細(xì)胞神經(jīng)網(wǎng)絡(luò)(CNNs)的均方指數(shù)穩(wěn)定性問題。利用Hlder不等式,It等距性質(zhì)和壓縮映射原理,提出了保證上述神經(jīng)網(wǎng)絡(luò)均方指數(shù)穩(wěn)定的充分條件。此外,給出一個(gè)具體例子來驗(yàn)證所獲得的結(jié)果是有效的。
關(guān)鍵詞:指數(shù)穩(wěn)定性;細(xì)胞神經(jīng)網(wǎng)絡(luò);壓縮映像原理
DOI:10.15938/j.jhust.2020.05.021
中圖分類號(hào): O177.2
文獻(xiàn)標(biāo)志碼: A
文章編號(hào): 1007-2683(2020)05-0149-09
Abstract:In this paper, the problem of the mean square exponential stability of a class of impulsive stochastic reaction-diffusion cellular neural networksCNNs) with transmission delay and distributed delay, and parameter uncertainties is discussed. By using Hlder inequality, It isometric nature and Contraction Mapping Principle, a sufficient condition to guarantee the mean square exponential stability of the above CNNs is proposed. Moreover, an example is given to demonstrate that the obtained result is effective.
Keywords:exponential stability; cellular neural networks; contraction mapping principle
0 引 言
近幾十年,神經(jīng)網(wǎng)絡(luò)因其在模式識(shí)別、聯(lián)想記憶、信號(hào)處理、圖像處理、組合優(yōu)化等領(lǐng)域的廣泛應(yīng)用而得到了廣泛的研究[1]。 然而,在神經(jīng)網(wǎng)絡(luò)的實(shí)現(xiàn)過程中,不可避免地遇到時(shí)間延遲。 已經(jīng)發(fā)現(xiàn),時(shí)間延遲的存在可能導(dǎo)致神經(jīng)網(wǎng)絡(luò)中的不穩(wěn)定性和振蕩。 因此,具有時(shí)滯的神經(jīng)網(wǎng)絡(luò)的穩(wěn)定性分析受到了廣泛關(guān)注[2-8]。
最近,國內(nèi)外學(xué)者對(duì)各種CNNs的穩(wěn)定性進(jìn)行了一系列的研究。在文[9]中,作者探討了脈沖擾動(dòng)下具有泄漏延遲的模糊細(xì)胞神經(jīng)網(wǎng)絡(luò)解的存在性和全局穩(wěn)定性。在文[10]中,當(dāng)激活函數(shù)滿足Lipschitz連續(xù)性條件時(shí),研究了具有恒定時(shí)滯的CNNs的漸近穩(wěn)定性。在文[11]中,作者考慮了離散時(shí)間CNNs,并獲得了幾個(gè)充分條件來檢驗(yàn)唯一均衡的全局指數(shù)穩(wěn)定性。 Lyapunov泛函方法是近幾十年來解
決神經(jīng)網(wǎng)絡(luò)穩(wěn)定性的常用技術(shù)之一,該方法往往需要構(gòu)建一個(gè)復(fù)雜的Lyapunov函數(shù),從而檢查更高維的LMI。此外,計(jì)算的復(fù)雜性增加,Lyapunov函數(shù)的構(gòu)造需要強(qiáng)大的數(shù)學(xué)技能。因此,國內(nèi)外學(xué)者開始探索解決這些問題的新方法。2001 年,Burton 首次將不動(dòng)點(diǎn)理論方法引入到研究神經(jīng)網(wǎng)絡(luò)穩(wěn)定性以來,該方法受到眾多學(xué)者的青睞并得到了快速的發(fā)展。2010 年,Luo基于不動(dòng)點(diǎn)理論,研究了隨機(jī) Volterra-Levin方程的指數(shù)穩(wěn)定性問題,給出了該類方程在均方意義下的指數(shù)穩(wěn)定性判據(jù)[12-15]。2013年, Guo C,ORegand 等[16]利用 Krasnoselskii 不動(dòng)點(diǎn)定理以及分析技巧,給出了均方意義下隨機(jī)中立型細(xì)胞神經(jīng)網(wǎng)絡(luò)具有指數(shù)穩(wěn)定性的判據(jù)。2015 年,Zhou 利用Brouwer不動(dòng)點(diǎn)定理證明了具有比例時(shí)滯混合BAM神經(jīng)網(wǎng)絡(luò)的平衡點(diǎn)的存在性和唯一性[17]。2017年,Rao 等[18-20]利用壓縮映象原理對(duì)脈沖隨機(jī)反應(yīng)擴(kuò)散細(xì)胞神經(jīng)網(wǎng)絡(luò)進(jìn)行穩(wěn)定性分析,并給出不確定參數(shù)脈沖積分微分方程的魯棒指數(shù)穩(wěn)定性判據(jù)。
本文的目的是研究一類具有傳輸延遲和分布延遲以及參數(shù)不確定性的脈沖隨機(jī)反應(yīng)-擴(kuò)散細(xì)胞神經(jīng)網(wǎng)絡(luò)(CNNs)的均方指數(shù)穩(wěn)定性。通過使用Hlder不等式,It等距性質(zhì)和壓縮映射原理,得到了一個(gè)充分條件來保證所考慮的CNN的均方指數(shù)穩(wěn)定性。此外,還給出了一個(gè)有效性的例子驗(yàn)證理論結(jié)果。
1 模型描述
4 結(jié) 論
對(duì)于神經(jīng)網(wǎng)絡(luò)而言,由于網(wǎng)絡(luò)的輸出是一個(gè)時(shí)間的函數(shù),對(duì)于給定的輸入,網(wǎng)絡(luò)的響應(yīng)可能收斂到一個(gè)穩(wěn)定的輸出,也可能出現(xiàn)振蕩等不穩(wěn)定的模式。因此,在神經(jīng)網(wǎng)絡(luò)的設(shè)計(jì)和分析中,穩(wěn)定性的分析是至關(guān)重要的。目前,針對(duì)具有混合時(shí)滯和脈沖神經(jīng)網(wǎng)絡(luò)穩(wěn)定特性的研究方法中被廣泛使用的當(dāng)屬 Lyapunov 泛函方法,但使用該方法有時(shí)需要構(gòu)造復(fù)雜的 Lyapunov-Krasovskii 泛函,這導(dǎo)致需要檢驗(yàn)一個(gè)更高維的LMI,增加了計(jì)算的復(fù)雜性和技巧性。將不動(dòng)點(diǎn)理論與神經(jīng)網(wǎng)絡(luò)的穩(wěn)定性問題相結(jié)合,利用Hlder不等式,It等距性質(zhì)和壓縮映射原理,得到了一個(gè)充分條件來保證所考慮的CNN的均方指數(shù)穩(wěn)定性。并給出一個(gè)具體例子來驗(yàn)證所獲得的結(jié)果是有效的。
參 考 文 獻(xiàn):
[1] JIANG F, YANG H, SHEN Y. Stability of Econd-order Stochastic Neutral Partial Functional Differential Equations Driven by Impulsive Noises[J]. Science China, 2016, 59(11): 117.
[2] TRCALA M. Spectral Stochastic Modeling of Uncertainties in Nonlinear Diffusion Problems of Moisture Transfer in Wood[J]. Applied Mathematical Modelling, 2015, 39(5/6): 1740.
[3] LUO J, TANIGUCHI T. Fixed Points and Stability of Stochastic Neutral Partial Differential Eequations with Infinite Delays[J]. Stochastic Analysis & Applications, 2009,27(6): 1163.
[4] SUTTISAWAT Y, SAKAIH H, ABE M. Microwave Effect in the Dehydrogenation of Tetralin and Decalin with a Fixed-bed Reactor[J]. International Journal of Hydrogen Energy, 2012, 37(4): 3242.
[5] KANDI H, MISHRA D, GORTHI S. Exploring the Learning Capabilities of Convolutional Neural Networks for Robust Image Watermarking[J]. Computers & Security, 2017, 65: 247.
[6] XIN X, WANG C, YING X, et al. Deep Community Detection in Topologically Incomplete Networks[J]. Physica A: Statistical Mechanics and its Applications, 2017, 469: 342.
[7] TANG P, WANG H, KWONG S. GoogLeNet Based Multi-stage Feature Fusion of Deep CNN for Scene Recognition[J]. Neurocomputing, 2017, 225: 188.
[8] YUDISTIRA N, KURITA T. Gated Spatio and Temporal Convolutional Neural Network for Activity Recognition: Towards Gated Multimodal Deep Learning[J]. Eurasip Journal on Image & Video Processing, 2017, 2017(1): 1.
[9] BRZDEK J, CIEPLINSKI K. On a Fixed Point Theorem in 2-Banach Spaces and Some of Its Applications[J]. Acta. Math. Sci., 2018, 38(2): 377.
[10]RAO R, ZHOUG S. Stability Analysis of Impulsive Stochastic Reaction Diffusion Cellular Neural Network with Distributed Delay Via Fixed Point Theory[J]. Complexity, 2017, 2017: 1.
[11]SAKTHIVEL R, LUO J. Asymptotic Stability of Impulsive Stochastic Partial Differential Equations with Infinite Delays[J]. Journal of Mathematical Analysis and Applications, 2009, 356(1): 1.
[12]LUO J. Fixed Points and Exponential Stability of Mild Solutions of Stochastic Partial Differential Equations with Delays[J]. Journal of Mathematical Analysis and Applications, 2008, 3422): 753.
[13]SAKTHIVEL R, LUO J. Asymptotic Stability of Nonlinear Impulsive Stochastic Differential Equations[J]. Statistics & Probability Letters, 2009, 79(9): 1219.
[14]SAKTHIVEL R, LUO J. Asymptotic Stability of Impulsive Stochastic Partial Differential Equations with Infinite Delays[J]. Journal of Mathematical Analysis and Applications, 2009, 356(1): 1.
[15]LUO J, TANIGUCHI T. Fixed Points and Stability of Stochastic Neutral Partial Differential Equations with Infinite Delays[J].
[16]GUO C, O′REGAN D, DENG F, et al. Fixed Points and Exponential Stability for a Stochastic Neutral Cellular Neural Network[J]. Applied Mathematics Letters, 2013, 26(8): 849.
[17]ZHOU L. Novel Global Exponential Stability Criteria for Hybrid BAM Neural Network with Proprtional[J]. Neurocomputing, 2015, 161: 99.
[18]RAO R, ZHOU G S. LMI-based Robust Exponential Stability Criterion of Impulsive Integro Differential Equations with Uncertain Parameters Via Contraction Mapping Theory[J]. Advances in Difference Equations, 2017, 2017: 1.
[19]HUDSON R L, PARTHASARATHY K R. Quantum Ito′s Formula and Stochastic Evolutions[J]. Commun. Mathphys, 1984, 93(3): 301.
[20]CARABALLO, TOMS, LIU K . Exponential Stability of Mild Solutions of Stochastic Partial Differential Equations with Delays[J]. Stochastic Analysis and Applications, 1999, 17(5):743.
(編輯:王 萍)