1、<p><b>　　英文文獻(xiàn)</b></p><p><b>　　英文資料：</b></p><p>　　Artificial neural networks (ANNs) to ArtificialNeuralNetworks, abbreviations also referred to as the neural network

2、(NNs) or called connection model (ConnectionistModel), it is a kind of model animals neural network behavior characteristic, distributed parallel information processing algorithm mathematical model. This network rely on

3、the complexity of the system, through the adjustment of mutual connection between nodes internal relations, so as to achieve the purpose of processing information. </p><p>　　(1) the nonlinear relationship is

4、 the nature of the nonlinear common characteristics. The wisdom of the brain is a kind of non-linear phenomena. Artificial neurons in the activation or inhibit the two different state, this kind of behavior in mathematic

5、s performance for a nonlinear relationship. Has the threshold of neurons in the network formed by the has better properties, can improve the fault tolerance and storage capacity. </p><p>　　(2) the limitation

6、s a neural network by DuoGe neurons widely usually connected to. A system of the overall behavior depends not only on the characteristics of single neurons, and may mainly by the unit the interaction between the, connect

7、ed to the. Through a large number of connection between units simulation of the brain limitations. Associative memory is a typical example of limitations. </p><p>　　(3) very qualitative artificial neural net

8、work is adaptive, self-organizing, learning ability. Neural network not only handling information can have all sorts of change, and in the treatment of the information at the same time, the nonlinear dynamic system itsel

9、f is changing. Often by iterative process description of the power system evolution. </p><p>　　(4) the convexity a system evolution direction, in certain conditions will depend on a particular state function

10、. For example energy function, it is corresponding to the extreme value of the system stable state. The convexity refers to the function extreme value, it has DuoGe DuoGe system has a stable equilibrium state, this will

11、cause the system to the diversity of evolution. </p><p>　　Artificial neural network, the unit can mean different neurons process of the object, such as characteristics, letters, concept, or some meaningful a

12、bstract model. The type of network processing unit is divided into three categories: input unit, output unit and hidden units. Input unit accept outside the world of signal and data; Output unit of output system processi

13、ng results; Hidden unit is in input and output unit, not between by external observation unit. The system The connections between n</p><p>　　Artificial neural network is used the parallel distributed system,

14、 with the traditional artificial intelligence and information processing technology completely different mechanism, overcome traditional based on logic of the symbols of the artificial intelligence in the processing of i

15、ntuition and unstructured information of defects, with the adaptive, self-organization and real-time characteristic of the study. </p><p>　　Development history </p><p>　　In 1943, psychologists W

16、.S.M cCulloch and mathematical logic W.P home its established the neural network and the math model, called MP model. They put forward by MP model of the neuron network structure and formal mathematical description metho

17、d, and prove the individual neurons can perform the logic function, so as to create artificial neural network research era. In 1949, the psychologist put forward the idea of synaptic contact strength variable. In the s,

18、the artificial neural network to fur</p><p>　　Network model </p><p>　　Artificial neural network model of the main consideration network connection topological structure, the characteristics, the

19、 learning rule neurons. At present, nearly 40 kinds of neural network model, with back propagation network, sensor, self-organizing mapping, the Hopfield network.the computer, wave boltzmann machine, adapt to the ear res

20、onance theory. According to the topology of the connection, the neural network model can be divided into: </p><p>　　(1) prior to the network before each neuron accept input and output level to the next level

21、, the network without feedback, can use a loop to no graph. This network realization from the input space to the output signal of the space transformation, it information processing power comes from simple nonlinear func

22、tion of DuoCi compound. The network structure is simple, easy to realize. Against the network is a kind of typical prior to the network. </p><p>　　(2) the feedback network between neurons in the network has

23、feedback, can use a no to complete the graph. This neural network information processing is state of transformations, can use the dynamics system theory processing. The stability of the system with associative memory fun

24、ction has close relationship. The Hopfield network.the computer, wave ear boltzmann machine all belong to this type. </p><p>　　Learning type </p><p>　　Neural network learning is an important con

25、tent, it is through the adaptability of the realization of learning. According to the change of environment, adjust to weights, improve the behavior of the system. The proposed by the Hebb Hebb learning rules for neural

26、network learning algorithm to lay the foundation. Hebb rules say that learning process finally happened between neurons in the synapse, the contact strength synapses parts with before and after the activity and synaptic

27、neuron changes. B</p><p>　　The network can through the weights between adjustment, the structure of the objective world, said the formation of inner characteristics of information processing method, informat

28、ion storage and processing reflected in the network connection. According to the learning environment is different, the study method of the neural network can be divided into learning supervision and unsupervised learnin

29、g. In the supervision and study, will the training sample data added to the network input, and the c</p><p>　　Analysis method </p><p>　　Study of the neural network nonlinear dynamic properties,

30、mainly USES the dynamics system theory and nonlinear programming theory and statistical theory to analysis of the evolution process of the neural network and the nature of the attractor, explore the synergy of neural net

31、work behavior and collective computing functions, understand neural information processing mechanism. In order to discuss the neural network and fuzzy comprehensive deal of information may, the concept of chaos theory an

32、d </p><p>　　superiority </p><p>　　The artificial neural network of characteristics and advantages, mainly in three aspects: first, self-learning. For example, only to realize image recognition t

33、hat the many different image model and the corresponding should be the result of identification input artificial neural network, the network will through the self-learning function, slowly to learn to distinguish similar

34、 images. The self-learning function for the forecast has special meaning. The prospect of artificial neural network comp</p><p>　　Research direction </p><p>　　The research into the neural networ

35、k can be divided into the theory research and application of the two aspects of research. Theory study can be divided into the following two categories: </p><p>　　1, neural physiological and cognitive scienc

36、e research on human thinking and intelligent mechanism. </p><p>　　2, by using the neural basis theory of research results, with mathematical method to explore more functional perfect, performance more superi

37、or neural network model, the thorough research network algorithm and performance, such as: stability and convergence, fault tolerance, robustness, etc.; The development of new network mathematical theory, such as: neural

38、 network dynamics, nonlinear neural field, etc. </p><p>　　Application study can be divided into the following two categories: </p><p>　　1, neural network software simulation and hardware realiza

39、tion of research. </p><p>　　2, the neural network in various applications in the field of research. These areas include: pattern recognition, signal processing, knowledge engineering, expert system, optimize

40、 the combination, robot control, etc. Along with the neural network theory itself and related theory, related to the development of technology, the application of neural network will further. </p><p>　　Devel

41、opment trend and research hot spot </p><p>　　Artificial neural network characteristic of nonlinear adaptive information processing power, overcome traditional artificial intelligence method for intuitive, su

42、ch as mode, speech recognition, unstructured information processing of the defects in the nerve of expert system, pattern recognition and intelligent control, combinatorial optimization, and forecast areas to be successf

43、ul application. Artificial neural network and other traditional method unifies, will promote the artificial intelligen</p><p>　　Neural network in many fields has got a very good application, but the need to

44、research is a lot. Among them, are distributed storage, parallel processing, since learning, the organization and nonlinear mapping the advantages of neural network and other technology and the integration of it follows

45、that the hybrid method and hybrid systems, has become a hotspot. Since the other way have their respective advantages, so will the neural network with other method, and the combination of strong points</p><p&g

46、t;　　漢語翻譯意見與反饋 選擇人工翻譯服務(wù)，獲得更專業(yè)的翻譯結(jié)果。 </p><p>　　人工神經(jīng)網(wǎng)絡(luò)（ArtificialNeuralNetworks，簡(jiǎn)寫為ANNs）也簡(jiǎn)稱為神經(jīng)網(wǎng)絡(luò)（NNs）或稱作連接模型（ConnectionistModel），它是一種模范動(dòng)物神經(jīng)網(wǎng)絡(luò)行為特征，進(jìn)行分布式并行信息處理的算法數(shù)學(xué)模型。這種網(wǎng)絡(luò)依靠系統(tǒng)的復(fù)雜程度，通過調(diào)整內(nèi)部大量節(jié)點(diǎn)之間相互連接的關(guān)系，從而達(dá)到處理信息的目的

47、。人工神經(jīng)網(wǎng)絡(luò)具有自學(xué)習(xí)和自適應(yīng)的能力，可以通過預(yù)先提供的一批相互對(duì)應(yīng)的輸入－輸出數(shù)據(jù)，分析掌握兩者之間潛在的規(guī)律，最終根據(jù)這些規(guī)律，用新的輸入數(shù)據(jù)來推算輸出結(jié)果，這種學(xué)習(xí)分析的過程被稱為“訓(xùn)練”。人工神經(jīng)網(wǎng)絡(luò)是由大量處理單元互聯(lián)組成的非線性、自適應(yīng)信息處理系統(tǒng)。它是在現(xiàn)代神經(jīng)科學(xué)研究成果的基礎(chǔ)上提出的，試圖通過模擬大腦神經(jīng)網(wǎng)絡(luò)處理、記憶信息的方式進(jìn)行信息處理。人工神經(jīng)網(wǎng)絡(luò)具有四個(gè)基本特征:</p><p>　　

48、（1）非線性 非線性關(guān)系是自然界的普遍特性。大腦的智慧就是一種非線性現(xiàn)象。人工神經(jīng)元處于激活或抑制二種不同的狀態(tài)，這種行為在數(shù)學(xué)上表現(xiàn)為一種非線性關(guān)系。具有閾值的神經(jīng)元構(gòu)成的網(wǎng)絡(luò)具有更好的性能，可以提高容錯(cuò)性和存儲(chǔ)容量。 </p><p>　?。?）非局限性 一個(gè)神經(jīng)網(wǎng)絡(luò)通常由多個(gè)神經(jīng)元廣泛連接而成。一個(gè)系統(tǒng)的整體行為不僅取決于單個(gè)神經(jīng)元的特征，而且可能主要由單元之間的相互作用、相互連接所決定。通過單元之間的大量

49、連接模擬大腦的非局限性。聯(lián)想記憶是非局限性的典型例子。 </p><p>　?。?）非常定性 人工神經(jīng)網(wǎng)絡(luò)具有自適應(yīng)、自組織、自學(xué)習(xí)能力。神經(jīng)網(wǎng)絡(luò)不但處理的信息可以有各種變化，而且在處理信息的同時(shí)，非線性動(dòng)力系統(tǒng)本身也在不斷變化。經(jīng)常采用迭代過程描寫動(dòng)力系統(tǒng)的演化過程。 </p><p>　?。?）非凸性 一個(gè)系統(tǒng)的演化方向，在一定條件下將取決于某個(gè)特定的狀態(tài)函數(shù)。例如能量函數(shù)，它的極值相

50、應(yīng)于系統(tǒng)比較穩(wěn)定的狀態(tài)。非凸性是指這種函數(shù)有多個(gè)極值，故系統(tǒng)具有多個(gè)較穩(wěn)定的平衡態(tài)，這將導(dǎo)致系統(tǒng)演化的多樣性。 </p><p>　　人工神經(jīng)網(wǎng)絡(luò)中，神經(jīng)元處理單元可表示不同的對(duì)象，例如特征、字母、概念，或者一些有意義的抽象模式。網(wǎng)絡(luò)中處理單元的類型分為三類：輸入單元、輸出單元和隱單元。輸入單元接受外部世界的信號(hào)與數(shù)據(jù)；輸出單元實(shí)現(xiàn)系統(tǒng)處理結(jié)果的輸出；隱單元是處在輸入和輸出單元之間，不能 由系統(tǒng)外部觀察的單元。神

51、經(jīng)元間的連接權(quán)值反映了單元間的連接強(qiáng)度，信息的表示和處理體現(xiàn)在網(wǎng)絡(luò)處理單元的連接關(guān)系中。人工神經(jīng)網(wǎng)絡(luò)是一種非程序化、適應(yīng)性、大腦風(fēng)格的信息處理 ，其本質(zhì)是通過網(wǎng)絡(luò)的變換和動(dòng)力學(xué)行為得到一種并行分布式的信息處理功能，并在不同程度和層次上模仿人腦神經(jīng)系統(tǒng)的信息處理功能。它是涉及神經(jīng)科學(xué)、思維科學(xué)、人工智能、計(jì)算機(jī)科學(xué)等多個(gè)領(lǐng)域的交叉學(xué)科。 </p><p>　　人工神經(jīng)網(wǎng)絡(luò)是并行分布式系統(tǒng)，采用了與傳統(tǒng)人工智能和信息

52、處理技術(shù)完全不同的機(jī)理，克服了傳統(tǒng)的基于邏輯符號(hào)的人工智能在處理直覺、非結(jié)構(gòu)化信息方面的缺陷，具有自適應(yīng)、自組織和實(shí)時(shí)學(xué)習(xí)的特點(diǎn)。 </p><p><b>　　發(fā)展歷史</b></p><p>　　1943年，心理學(xué)家W.S.McCulloch和數(shù)理邏輯學(xué)家W.Pitts建立了神經(jīng)網(wǎng)絡(luò)和數(shù)學(xué)模型，稱為MP模型。他們通過MP模型提出了神經(jīng)元的形式化數(shù)學(xué)描述和網(wǎng)絡(luò)結(jié)構(gòu)方

53、法，證明了單個(gè)神經(jīng)元能執(zhí)行邏輯功能，從而開創(chuàng)了人工神經(jīng)網(wǎng)絡(luò)研究的時(shí)代。1949年，心理學(xué)家提出了突觸聯(lián)系強(qiáng)度可變的設(shè)想。60年代，人工神經(jīng)網(wǎng)絡(luò)的到了進(jìn)一步發(fā)展，更完善的神經(jīng)網(wǎng)絡(luò)模型被提出 ，其中包括感知器和自適應(yīng)線性元件等。M.Minsky等仔細(xì)分析了以感知器為代表的神經(jīng)網(wǎng)絡(luò)系統(tǒng)的功能及局限后，于1969年出版了《Perceptron》一書，指出感知器不能解決高階謂詞問題。他們的論點(diǎn)極大地影響了神經(jīng)網(wǎng)絡(luò)的研究，加之當(dāng)時(shí)串行計(jì)算機(jī)和人工智

54、能所取得的成就，掩蓋了發(fā)展新型計(jì)算機(jī)和人工智能新途徑的必要性和迫切性，使人工神經(jīng)網(wǎng)絡(luò)的研究處于低潮。在此期間，一些人工神經(jīng)網(wǎng)絡(luò)的研究者仍然致力于這一研究，提出了適應(yīng)諧振理論（ART網(wǎng)）、自組織映射、認(rèn)知機(jī)網(wǎng)絡(luò)，同時(shí)進(jìn)行了神經(jīng)網(wǎng)絡(luò)數(shù)學(xué)理論的研究。以上研究為神經(jīng)網(wǎng)絡(luò)的研究和發(fā)展奠定了基礎(chǔ)。1982年，美國加州工學(xué)院物理學(xué)家J.J.Hopfield提出了Hopfield神經(jīng)網(wǎng)格模型，引入了“計(jì)</p><p><

55、b>　　網(wǎng)絡(luò)模型</b></p><p>　　人工神經(jīng)網(wǎng)絡(luò)模型主要考慮網(wǎng)絡(luò)連接的拓?fù)浣Y(jié)構(gòu)、神經(jīng)元的特征、學(xué)習(xí)規(guī)則等。目前，已有近40種神經(jīng)網(wǎng)絡(luò)模型，其中有反傳網(wǎng)絡(luò)、感知器、自組織映射、Hopfield網(wǎng)絡(luò)、波耳茲曼機(jī)、適應(yīng)諧振理論等。根據(jù)連接的拓?fù)浣Y(jié)構(gòu)，神經(jīng)網(wǎng)絡(luò)模型可以分為： </p><p>　?。?）前向網(wǎng)絡(luò) 網(wǎng)絡(luò)中各個(gè)神經(jīng)元接受前一級(jí)的輸入，并輸出到下一級(jí)，網(wǎng)絡(luò)中沒

56、有反饋，可以用一個(gè)有向無環(huán)路圖表示。這種網(wǎng)絡(luò)實(shí)現(xiàn)信號(hào)從輸入空間到輸出空間的變換，它的信息處理能力來自于簡(jiǎn)單非線性函數(shù)的多次復(fù)合。網(wǎng)絡(luò)結(jié)構(gòu)簡(jiǎn)單，易于實(shí)現(xiàn)。反傳網(wǎng)絡(luò)是一種典型的前向網(wǎng)絡(luò)。 </p><p>　　（2）反饋網(wǎng)絡(luò) 網(wǎng)絡(luò)內(nèi)神經(jīng)元間有反饋，可以用一個(gè)無向的完備圖表示。這種神經(jīng)網(wǎng)絡(luò)的信息處理是狀態(tài)的變換，可以用動(dòng)力學(xué)系統(tǒng)理論處理。系統(tǒng)的穩(wěn)定性與聯(lián)想記憶功能有密切關(guān)系。Hopfield網(wǎng)絡(luò)、波耳茲曼機(jī)均屬于這種類

57、型。 </p><p><b>　　學(xué)習(xí)類型</b></p><p>　　學(xué)習(xí)是神經(jīng)網(wǎng)絡(luò)研究的一個(gè)重要內(nèi)容，它的適應(yīng)性是通過學(xué)習(xí)實(shí)現(xiàn)的。根據(jù)環(huán)境的變化，對(duì)權(quán)值進(jìn)行調(diào)整，改善系統(tǒng)的行為。由Hebb提出的Hebb學(xué)習(xí)規(guī)則為神經(jīng)網(wǎng)絡(luò)的學(xué)習(xí)算法奠定了基礎(chǔ)。Hebb規(guī)則認(rèn)為學(xué)習(xí)過程最終發(fā)生在神經(jīng)元之間的突觸部位，突觸的聯(lián)系強(qiáng)度隨著突觸前后神經(jīng)元的活動(dòng)而變化。在此基礎(chǔ)上，人們提出

58、了各種學(xué)習(xí)規(guī)則和算法，以適應(yīng)不同網(wǎng)絡(luò)模型的需要。有效的學(xué)習(xí)算法，使得神 </p><p>　　經(jīng)網(wǎng)絡(luò)能夠通過連接權(quán)值的調(diào)整，構(gòu)造客觀世界的內(nèi)在表示，形成具有特色的信息處理方法，信息存儲(chǔ)和處理體現(xiàn)在網(wǎng)絡(luò)的連接中。根據(jù)學(xué)習(xí)環(huán)境不同，神經(jīng)網(wǎng)絡(luò)的學(xué)習(xí)方式可分為監(jiān)督學(xué)習(xí)和 非監(jiān)督學(xué)習(xí)。在監(jiān)督學(xué)習(xí)中，將訓(xùn)練樣本的數(shù)據(jù)加到網(wǎng)絡(luò)輸入端，同時(shí)將相應(yīng)的期望輸出與網(wǎng)絡(luò)輸出相比較，得到誤差信號(hào)，以此控制權(quán)值連接強(qiáng)度的調(diào)整，經(jīng)多次訓(xùn)練后收

59、斂到一個(gè)確定的權(quán)值。當(dāng)樣本情況發(fā)生變化時(shí)，經(jīng)學(xué)習(xí)可以修改權(quán)值以適應(yīng)新的環(huán)境。使用監(jiān)督學(xué)習(xí)的神經(jīng)網(wǎng)絡(luò)模型有反傳網(wǎng)絡(luò)、感知器等。非監(jiān)督學(xué)習(xí)時(shí)，事先不給定標(biāo)準(zhǔn)樣本，直接將網(wǎng)絡(luò)置于環(huán)境之中，學(xué)習(xí)階段與工作階段成為一體。此時(shí)，學(xué)習(xí)規(guī)律的變化服從連接權(quán)值的演變方程。非監(jiān)督學(xué)習(xí)最簡(jiǎn)單的例子是Hebb學(xué)習(xí)規(guī)則。競(jìng)爭(zhēng)學(xué)習(xí)規(guī)則是一個(gè)更復(fù)雜的非監(jiān)督學(xué)習(xí)的例子，它是根據(jù)已建立的聚類進(jìn)行權(quán)值調(diào)整。自組織映射、適應(yīng)諧振理論網(wǎng)絡(luò)等都是與競(jìng)爭(zhēng)學(xué)習(xí)有關(guān)的典型模型。 &l

60、t;/p><p><b>　　分析方法</b></p><p>　　研究神經(jīng)網(wǎng)絡(luò)的非線性動(dòng)力學(xué)性質(zhì)，主要采用動(dòng)力學(xué)系統(tǒng)理論、非線性規(guī)劃理論和統(tǒng)計(jì)理論，來分析神經(jīng)網(wǎng)絡(luò)的演化過程和吸引子的性質(zhì)，探索神經(jīng)網(wǎng)絡(luò)的協(xié)同行為和集體計(jì)算功能，了解神經(jīng)信息處理機(jī)制。為了探討神經(jīng)網(wǎng)絡(luò)在整體性和模糊性方面處理信息的可能，混沌理論的概念和方法將會(huì)發(fā)揮作用?；煦缡且粋€(gè)相當(dāng)難以精確定義的數(shù)學(xué)概念。

61、一般而言，“混沌”是指由確定性方程描述的動(dòng)力學(xué)系統(tǒng)中表現(xiàn)出的非確定性行為，或稱之為確定的隨機(jī)性?！按_定性”是因?yàn)樗蓛?nèi)在的原因而不是外來的噪聲或干擾所產(chǎn)生，而“隨機(jī)性”是指其不規(guī)則的、不能預(yù)測(cè)的行為，只可能用統(tǒng)計(jì)的方法描述?；煦鐒?dòng)力學(xué)系統(tǒng)的主要特征是其狀態(tài)對(duì)初始條件的靈敏依賴性，混沌反映其內(nèi)在的隨機(jī)性。混沌理論是指描述具有混沌行為的非線性動(dòng)力學(xué)系統(tǒng)的基本理論、概念、方法，它把動(dòng)力學(xué)系統(tǒng)的復(fù)雜行為理解為其自身與其在同外界進(jìn)行物質(zhì)、能量和信

62、息交換過程中內(nèi)在的有結(jié)構(gòu)的行為，而不是外來的和偶然的行為，混沌狀態(tài)是一種定態(tài)?；煦鐒?dòng)力學(xué)系統(tǒng)的定態(tài)包括：靜止、平穩(wěn)量、周期性、準(zhǔn)同 期性和混沌解?；煦畿壘€是整體上穩(wěn)定與局部不穩(wěn)定相結(jié)合的結(jié)果，稱之為奇異吸引子。一個(gè)奇異吸引子有如下一些特征：（1）奇異吸引</p><p><b>　　優(yōu)越性</b></p><p>　　人工神經(jīng)網(wǎng)絡(luò)的特點(diǎn)和優(yōu)越性，主要表現(xiàn)在三個(gè)方面：

63、第一，具有自學(xué)習(xí)功能。例如實(shí)現(xiàn)圖像識(shí)別時(shí)，只在先把許多不同的圖像樣板和對(duì)應(yīng)的應(yīng)識(shí)別的結(jié)果輸入人工神經(jīng)網(wǎng)絡(luò)，網(wǎng)絡(luò)就會(huì)通過自學(xué)習(xí)功能，慢慢學(xué)會(huì)識(shí)別類似的圖像。自學(xué)習(xí)功能對(duì)于預(yù)測(cè)有特別重要的意義。預(yù)期未來的人工神經(jīng)網(wǎng)絡(luò)計(jì)算機(jī)將為人類提供經(jīng)濟(jì)預(yù)測(cè)、市場(chǎng)預(yù)測(cè)、效益預(yù)測(cè)，其應(yīng)用前途是很遠(yuǎn)大的。第二，具有聯(lián)想存儲(chǔ)功能。用人工神經(jīng)網(wǎng)絡(luò)的反饋網(wǎng)絡(luò)就可以實(shí)現(xiàn)這種聯(lián)想。第三，具有高速尋找優(yōu)化解的能力。尋找一個(gè)復(fù)雜問題的優(yōu)化解，往往需要很大的計(jì)算量，利用一個(gè)針

64、對(duì)某問題而設(shè)計(jì)的反饋型人工神經(jīng)網(wǎng)絡(luò)，發(fā)揮計(jì)算機(jī)的高速運(yùn)算能力，可能很快找到優(yōu)化解。 </p><p><b>　　研究方向</b></p><p>　　神經(jīng)網(wǎng)絡(luò)的研究可以分為理論研究和應(yīng)用研究?jī)纱蠓矫妗?理論研究可分為以下兩類： </p><p>　　1、利用神經(jīng)生理與認(rèn)知科學(xué)研究人類思維以及智能機(jī)理。 </p><p>

65、;　　2、利用神經(jīng)基礎(chǔ)理論的研究成果，用數(shù)理方法探索功能更加完善、性能更加優(yōu)越的神經(jīng)網(wǎng)絡(luò)模型，深入研究網(wǎng)絡(luò)算法和性能，如：穩(wěn)定性、收斂性、容錯(cuò)性、魯棒性等；開發(fā)新的網(wǎng)絡(luò)數(shù)理理論，如：神經(jīng)網(wǎng)絡(luò)動(dòng)力學(xué)、非線性神經(jīng)場(chǎng)等。 </p><p>　　應(yīng)用研究可分為以下兩類： </p><p>　　1、神經(jīng)網(wǎng)絡(luò)的軟件模擬和硬件實(shí)現(xiàn)的研究。 </p><p>　　2、神經(jīng)網(wǎng)絡(luò)在各個(gè)

66、領(lǐng)域中應(yīng)用的研究。這些領(lǐng)域主要包括：模式識(shí)別、信號(hào)處理、知識(shí)工程、專家系統(tǒng)、優(yōu)化組合、機(jī)器人控制等。隨著神經(jīng)網(wǎng)絡(luò)理論本身以及相關(guān)理論、相關(guān)技術(shù)的不斷發(fā)展，神經(jīng)網(wǎng)絡(luò)的應(yīng)用定將更加深入。 </p><p><b>　　發(fā)展趨勢(shì)及研究熱點(diǎn)</b></p><p>　　人工神經(jīng)網(wǎng)絡(luò)特有的非線性適應(yīng)性信息處理能力，克服了傳統(tǒng)人工智能方法對(duì)于直覺，如模式、語音識(shí)別、非結(jié)構(gòu)化信息處

67、理方面的缺陷，使之在神經(jīng)專家系統(tǒng)、模式識(shí)別、智能控制、組合優(yōu)化、預(yù)測(cè)等領(lǐng)域得到成功應(yīng)用。人工神經(jīng)網(wǎng)絡(luò)與其它傳統(tǒng)方法相結(jié)合，將推動(dòng)人工智能和信息處理技術(shù)不斷發(fā)展。近年來，人工神經(jīng)網(wǎng)絡(luò)正向模擬人類認(rèn)知的道路上更加深入發(fā)展，與模糊系統(tǒng)、遺傳算法、進(jìn)化機(jī)制等結(jié)合，形成計(jì)算智能，成為人工智能的一個(gè)重要方向，將在實(shí)際應(yīng)用中得到發(fā)展。將信息幾何應(yīng)用于人工神經(jīng)網(wǎng)絡(luò)的研究，為人工神經(jīng)網(wǎng)絡(luò)的理論研究開辟了新的途徑。神經(jīng)計(jì)算機(jī)的研究發(fā)展很快，已有產(chǎn)品進(jìn)入市場(chǎng)

68、。光電結(jié)合的神經(jīng)計(jì)算機(jī)為人工神經(jīng)網(wǎng)絡(luò)的發(fā)展提供了良好條件。 </p><p>　　神經(jīng)網(wǎng)絡(luò)在很多領(lǐng)域已得到了很好的應(yīng)用，但其需要研究的方面還很多。其中，具有分布存儲(chǔ)、并行處理、自學(xué)習(xí)、自組織以及非線性映射等優(yōu)點(diǎn)的神經(jīng)網(wǎng)絡(luò)與其他技術(shù)的結(jié)合以及由此而來的混合方法和混合系統(tǒng)，已經(jīng)成為一大研究熱點(diǎn)。由于其他方法也有它們各自的優(yōu)點(diǎn)，所以將神經(jīng)網(wǎng)絡(luò)與其他方法相結(jié)合，取長(zhǎng)補(bǔ)短，繼而可以獲得更好的應(yīng)用效果。目前這方面工作有神經(jīng)網(wǎng)