求翻译下面文字翻译成英文

当数字混入噪声时，通常会降低识别率和识别精度。为此，本文研究了基于BP神经网络和Hopfield神经网络的有噪数字识别问题。用神经网络设计分类器，研究在噪声环境下BP网路和Hopfield网络的设计，BP网络和Hopfield网络的实现方案等。
When sneaking into noise as if it be a figure, be able to reduce recognition rate and distinguish accuracy generally. For this purpose, the main body of a book digital recognition problem having studied the neural networks having chirps owing to BP neural networks and Hopfield. Use neural networks to design classification implement, design studying the BP network and Hopfield are network under noise environment, the scheme waits for the BP network and network realization of Hopfield.

主要任务是设计一个可行的、高效的，神经网络，并将其应用于实例0~9共10个数字，使其在标准样本加入随机噪声之后能够进行识别，并达到高识别率，低误识率。本文用了两种网络，第一种为BP神经网络字符识别，第二种为Hopfield神经网络字符识别。BP神经网络的数字识别过程分为两步：训练阶段，识别阶段。在基于BP神经网络的字符识别系统设计过程中，认真研究了网络参数设计、网络训练和识别在内的BP网络设计上的关键问题。并测试了设计的网络数字识别系统的可靠性。
The major task is feasible design one, high-effect's , neural networks, applying the person to and example 0 ~ 9 10 in total figures, makes the person be able to carry out the recognition rate distinguishing, and reaching height after the standard sample book adds random noise , the low mistake knows rates. The main body of a book has been used grow a network liang , has been a BP neural networks character identification the first kind , has been a Hopfield neural networks character identification second kinds. BP neural networks digital recognition process is two steps mark: Train a stage , distinguish a stage. In designing process owing to BP neural networks character identification system, have studied the network parameter designs , network training and the network distinguishing inclusive BP design upper $64 question carefully. And have tested network digital recognition system reliability designing that.

Hopfield神经网络应用于含噪数字识别中，Hopfield神经网络的“能量函数”的能量在网络运行过程中，具有不断地减少最后趋于稳定的平衡状态的特性，而网络一旦建立即可自动运行，无需要训练。在识别阶段，将待识别数字特征送入网络运行，待网络运行到平衡状态后，输出结果。
But Hopfield neural networks apply to contain chirp Hopfield neural networks "energy function " energy runs process middle in the network , have the characteristic property cutting down an at last unceasingly tending to stable equilibrium state in digital recognition,the network works voluntarily once building-up is OK , the nothing needs to train. Work in distinguishing a stage , sending to waiting for the characteristic distinguishing a figure in a network , export result after the network works to equilibrium state.

最后，用设计好的BP神经网络的数字识别系统对数字图片进行了识别。
本文给出了有噪数字识别的BP网络设计过程、方案、及部分源程序。
Identification having been in progress finally, to figure photograph with the digital recognition system designing good BP neural networks's. The main body of a book the BP network having given digital recognition out having chirping designs process , scheme, and part source program.

这样就可以了

when the digit mixes in the noise, usually will reduce the recognition rate and the recognition precision. Therefore, this article has studied has chirp the digital recognition question based on the BP neural network and the Hopfield neural network. With the neural network design sorter, studies under the noise environment the BP network and the Hopfield network design, the BP network and the Hopfield network realizes the plan and so on. the primary mission is designs one feasible, highly effective, the neural network, and applies it in the example 0~9 altogether 10 digit, causes it after the standard sample joins the stochastic noise can carry on the recognition, and achieves the high recognition rate, low probability of misrecognition. This article has used two kind of networks, the first kind is the BP neural network character recognition, the second kind is the Hopfield neural network character recognition. The BP neural network's digital recognition process divides into two steps: Training stage, recognition stage. In based on the BP neural network's character recognition system design process, has studied the network parameter design, the network training and in the recognition BP network design key question earnestly. And has tested the design network digit recognition system's reliability. the Hopfield neural network applies in contains chirp in the digital recognition, Hopfield neural network “energy function” the energy in the network movement process, has reduces tends the stable state of equilibrium characteristic finally unceasingly, but network, once establishes then automatically moves, does not have needs to train. In the recognition stage, will wait the character recognition characteristic to send in the network movement, treats the network moves after the state of equilibrium, output result. is final, with designed the good BP neural network the digital recognition system to carry on the recognition to the digital picture. this article gave had chirp the digital recognition BP network design process, the plan, and the partial source programs

Neural networks I would like to thank the abstract, not to machine translation
When mixed with the noise figure, it is usually to reduce the recognition rate and recognition accuracy. In this paper, based on BP neural network and Hopfield neural network identification of noise figure. Using neural network classifier designed to study the noise environment in the BP network and the design of Hopfield network, BP network and Hopfield network programs such as the realization.
The main task is to design a feasible and efficient, neural networks, and applied to examples of 0 ~ 9, a total of 10 digits to make it in the standard sample after adding random noise to identify and achieve high recognition rate, low error rate. In this paper, two types of networks, the first of a BP neural network for character recognition, the second for the Hopfield neural network character recognition. BP neural network to identify the number of two-step process: the training stage, the identification stage. BP neural network based character recognition system of the design process, carefully studied the design of network parameters, network training and recognition of the BP network design, including the key issue.
And test the design of the network the reliability of digital identification system.
Hopfield neural network with noise applied to the number of recognition, Hopfield neural network "energy function" of energy to run the process in the network, with continuous reduction of the final balance has become stable characteristics, and the network can be run automatically, once established, No need to train. In the identification stage, to identify the number of characteristics to be sent to the network until the network into a state of equilibrium, the output results.
Finally, with the design of a good number of BP neural network recognition system for a digital picture identification.
In this paper, a noise figure of the BP network to identify the design process, programs, and some source code.

自己翻的，希望有用。
打了半天呢！！～～
*呵呵*

Neural networks I would like to thank the abstract, not to machine translation
When mixed with the noise figure, it is usually to reduce the recognition rate and recognition accuracy. In this paper, based on BP neural network and Hopfield neural network identification of noise figure. Using neural network classifier designed to study the noise environment in the BP network and the design of Hopfield network, BP network and Hopfield network programs such as the realization.
The main task is to design a feasible and efficient, neural networks, and applied to examples of 0 ~ 9, a total of 10 digits to make it in the standard sample after adding random noise to identify and achieve high recognition rate, low error rate. In this paper, two types of networks, the first of a BP neural network for character recognition, the second for the Hopfield neural network character recognition. BP neural network to identify the number of two-step process: the training stage, the identification stage. BP neural network based character recognition system of the design process, carefully studied the design of network parameters, network training and recognition of the BP network design, including the key issue. And test the design of the network the reliability of digital identification system.
Hopfield neural network with noise applied to the number of recognition, Hopfield neural network "energy function" of energy to run the process in the network, with continuous reduction of the final balance has become stable characteristics, and the network can be run automatically, once established, No need to train. In the identification stage, to identify the number of characteristics to be sent to the network until the network into a state of equilibrium, the output results.
Finally, with the design of a good number of BP neural network recognition system for a digital picture identification.
In this paper, a noise figure of the BP network to identify the design process, programs, and some source code.