Location:
Search - neural units
Search list
Description: 论文标题:自适应模糊系统在手写体数字识别中的应用研究
作者:张镭
作者专业:计算机软件人工智能
导师姓名:黄战
授予学位:硕士
授予单位:暨南大学
授予学位时间:19990501
论文页数:59页
文摘语种:中文文摘
分类号:TP18 TP391.4
关键词:手写体数字 自适应 模糊逻辑 神经网络 模式识别
摘要:该文针对模式识别的特点,构造了适合于模式识别问题的自适应模糊系统,对三种不同学习算法加以改进,在手写全数字识别上对分类器进行了实现,并与采用BP算法训练的三层前馈神经网络分类器相比较,分析其优劣.仿真实验表明,在该文的样本集条件下,自适应模糊分类吕的识别性能优于神经网络分类器,这充分体现了自适应模糊技术用于数字识别的优越性和潜力.-thesis entitled : adaptive fuzzy system in a handwritten numeral recognition of applied research Author : Zhang Lei professional author : artificial intelligence computer software instructor Name : Huang war conferred degrees : Master award units : Jinan University conferred degrees : 19990501 page thesis : Abstracts 59 languages : Chinese Digest Key words : TP18-image Keywords : handwritten digital adaptive fuzzy logic neural network pattern recognition Abstract : In this paper the characteristics of pattern recognition, suitable for the construction of adaptive pattern recognition fuzzy systems, three different learning algorithm to improve the handwriting recognition on digital classification of the device to achieve, BP and with a three-tiered training algorithm for neural network clas
Platform: |
Size: 4194405 |
Author: 成东 |
Hits:
Description: 论文标题:自适应模糊系统在手写体数字识别中的应用研究
作者:张镭
作者专业:计算机软件人工智能
导师姓名:黄战
授予学位:硕士
授予单位:暨南大学
授予学位时间:19990501
论文页数:59页
文摘语种:中文文摘
分类号:TP18 TP391.4
关键词:手写体数字 自适应 模糊逻辑 神经网络 模式识别
摘要:该文针对模式识别的特点,构造了适合于模式识别问题的自适应模糊系统,对三种不同学习算法加以改进,在手写全数字识别上对分类器进行了实现,并与采用BP算法训练的三层前馈神经网络分类器相比较,分析其优劣.仿真实验表明,在该文的样本集条件下,自适应模糊分类吕的识别性能优于神经网络分类器,这充分体现了自适应模糊技术用于数字识别的优越性和潜力.-thesis entitled : adaptive fuzzy system in a handwritten numeral recognition of applied research Author : Zhang Lei professional author : artificial intelligence computer software instructor Name : Huang war conferred degrees : Master award units : Jinan University conferred degrees : 19990501 page thesis : Abstracts 59 languages : Chinese Digest Key words : TP18-image Keywords : handwritten digital adaptive fuzzy logic neural network pattern recognition Abstract : In this paper the characteristics of pattern recognition, suitable for the construction of adaptive pattern recognition fuzzy systems, three different learning algorithm to improve the handwriting recognition on digital classification of the device to achieve, BP and with a three-tiered training algorithm for neural network clas
Platform: |
Size: 4194304 |
Author: 成东 |
Hits:
Description: 详细说明了如何实现基于bp神经网络的手写数字识别。神经网络对于参数的设置是敏感的，尤其是隐藏层的单元个数，本文列出了一系列bp神经网络的应用的参数设置。结果表明，可以实现较好的模式识别功能-Detailed description of how to realize bp neural network-based handwritten numeral recognition. Neural network for parameter setting is sensitive, especially the number of hidden layer units, the paper sets out a series of bp neural network applications, parameters setting. The results show that pattern recognition can be achieved better functional
Platform: |
Size: 1120256 |
Author: 胡存英 |
Hits:
Description: ADIAL Basis Function (RBF) networks were introduced
into the neural network literature by Broomhead and
Lowe [1], which are motivated by observation on the local
response in biologic neurons. Due to their better
approximation capabilities, simpler network structures and
faster learning algorithms, RBF networks have been widely applied in many science and engineering fields. RBF network is three layers feedback network, where each hidden unit implements a radial activation function and each output unit implements a weighted sum of hidden units’ outputs.
Platform: |
Size: 114688 |
Author: u123xz |
Hits:
Description: 步态识别图像的初步裁切，对视频监控提取的图像进行裁切-The structures of the neural networks were designed using a constructive algorithm where the basic idea was to start with a small network,then add hidden units and weights incrementally until a satisfactory solution be found
Platform: |
Size: 5120 |
Author: maliguo |
Hits:
Description: 基于bp神经网络的车牌字符识别源程序，采用字符比对的方式进行切割-The structures of the neural networks were designed using a constructive algorithm where the basic idea was to start with a small network,then add hidden units and weights incrementally until a satisfactory solution be found .The higher efficiencies were for a 30×17×10 number network and letter network
Platform: |
Size: 488448 |
Author: maliguo |
Hits:
Description: neural-network模型中，在產生一個輸出值前units轉換它們的net-input數值為一個activation value並視為一個中介的步驟。很多架構省略這個中介的步驟並且直接到輸出值的產生。在這裡，先忽略這個activation value的複雜度，我們首要的工作是output value輸出值的產生。我們以一個微分方程式的形式來表示一個unit的output value。就好像是生物學中所提的同等事物一樣，units的輸出值是時間的動態函數。-neural-network model, generating an output value before the units converted to their net-input values for an activation value, and as an intermediate step. Many architecture omitted intermediate steps and go directly to the output value of production. Here, first, ignore the activation value of the complexity, our primary task is to output value output value generation. We are in the form of a differential equation to represent a unit of output value. Biology is like the same things mentioned in the same, units of output value is the dynamic function of time.
Platform: |
Size: 79872 |
Author: 曾琪騰 |
Hits:
Description: Radial Basis Function Neural Controller for Tanker Ship Heading Regulation (using only 9 receptive field units)
Platform: |
Size: 14336 |
Author: ffault |
Hits:
Description: 这是一个matlab程序用于构建人工神经网络模型，可以随意设置层数和单元个数！-This is a matlab program for building artificial neural network model can arbitrarily set the number of layers and units!
Platform: |
Size: 1531904 |
Author: 龙陈 |
Hits:
Description: 神经网络建立的空调系统空气处理单元模型；simulink模型mdl，附带相关数据及归一化处理程序-Air-conditioning system neural network model of air handling units simulink model mdl, with relevant data and normalization process
Platform: |
Size: 13312 |
Author: 王政 |
Hits:
Description: 文章介绍如何使用CUDA实现神经网络，并把他应用在GPU图像处理单元上。
-An Artificial Neural Network is an information processing method that was inspired by the way biological nervous systems function, such as the brain, to process information. It is composed of a large number of highly interconnected processing elements (neurons) working in unison to solve specific problems. Neural Networks have been widely used in "analogous" signal classifications, including handwriting, voice and image recognitions. Neural network can also be used in computer games. It enables games with the ability to adaptively learn from player behaviors. This technique has been used in racing games, such that opponent cars controlled by computers can learn how to drive by human players.
Since a Neural Network requires a considerable number of vector and matrix operations to get results, it is very suitable to be implemented in a parallel programming model and run on Graphics Processing Units (GPUs). Our goal is to utilize and unleash the power of GPUs to boost the performance o
Platform: |
Size: 1392640 |
Author: 尘封 |
Hits:
Description: 提出了一种基于自适应 Chebyshev 多项式神经网络(ACNN)的 Logistic 混沌系统控制算法。该算法采用 Chebyshev
正交多项式作为神经网络的激励函数, 构建 Logistic 混沌系统的预测与控制模型。为了保证算法的稳定性, 提出和证明了收敛定
理, 并利用自适应学习率算法提高神经网络的学习效率和收敛速度。通过采用自适应 Chebyshev 神经网络直接学习 Logistic 混
沌系统的动态特性, 并对系统实施目标函数控制。实验仿真结果表明, 该算法在 Logistic 混沌系统有外部干扰的情况下仍能对其
进行有效控制, 网络学习时间为 0.178 s, 训练步长为 10, 均方误差达到 1.15×10
− 4 , 与其他常见算法相比具有计算量小、速度快、
精度高和网络结构简单等优点。 - A novel algorithm for controlling Logistic chaotic system based on adaptive Chebyshev polynomials
neural networks (ACNN) is presented. In the algorithm, the activation function of hidden units is defined by Chebyshev
orthogonal polynomials in the neural networks, and the forecast and control model of Logistic chaotic system is estab-
lished. In order to ensure stability of the algorithm, the convergence theorem of the algorithm is proposed and proved.
Then the adaptive learning rate algorithm is used for improving the learning efficiency and convergence speed. The
adaptive Chebyshev neural networks directly learn dynamic characters of Logistic chaotic system and control it to target
function. The simulation results show that the algorithm is still effective when there are external disturbance in the Lo-
gistic chaotic system, now the learning time is 0.178s, training steps is 10 and mean square error is 1.15×10 − 4 . Com-
pared with other ordinary
Platform: |
Size: 1381376 |
Author: |
Hits:
Description: Training Artificial Neural Network.
XOR Problem.
Summation Units, Log-Sigmoid Neurons with Biases.
Input Layer: 2, Hidden Layer: 2, Output Layer: 1 neurons.
Returns mean square error between desired and actual outputs.
Reference Papers:
D. Karaboga, B. Basturk Akay, C. Ozturk, Artificial Bee Colony (ABC) Optimization Algorithm for Training Feed-Forward Neural Networks, LNCS: Modeling Decisions for Artificial Intelligence, 4617/2007, 318-329, 2007.
D. Karaboga, C. Ozturk, Neural Networks Training by Artificial Bee Colony Algorithm on Pattern Classification, Neural Network World, 19(3), 279-292, 2009.
*/
- Training Artificial Neural Network.
XOR Problem.
Summation Units, Log-Sigmoid Neurons with Biases.
Input Layer: 2, Hidden Layer: 2, Output Layer: 1 neurons.
Returns mean square error between desired and actual outputs.
Reference Papers:
D. Karaboga, B. Basturk Akay, C. Ozturk, Artificial Bee Colony (ABC) Optimization Algorithm for Training Feed-Forward Neural Networks, LNCS: Modeling Decisions for Artificial Intelligence, 4617/2007, 318-329, 2007.
D. Karaboga, C. Ozturk, Neural Networks Training by Artificial Bee Colony Algorithm on Pattern Classification, Neural Network World, 19(3), 279-292, 2009.
*/
Platform: |
Size: 5120 |
Author: ehsan |
Hits:
Description: In this work, Artificial Neural Networks are employed in a Fault Detection and Isolation scheme for robotic manipulators. Two networks are utilized: a Multilayer Perceptron is employed to reproduce the manipulator dynamical behavior, generating a residual vector that is classified by a Radial Basis Function Network, giving the fault isolation. Two methods are utilized to choose
the radial unit centers in this network. The first method,
Forward Selection, employs Subset Selection to choose
the radial units from the training patterns. The second employs the Kohonen’s Self-Organizing Map to fix the radial unit centers in more interesting positions.
Simulations employing a two link manipulator and the Puma 560 manipulator indicate that the second method gives a smaller generalization error.
Platform: |
Size: 149504 |
Author: fad |
Hits:
Description: An Artificial Neural Network is a network of many very simple processors ("units"), each possibly having a (small amount of) local memory. The units are connected by unidirectional communication channels ("connections"), which carry numeric (as opposed to symbolic) data. The units operate only on their local data and on the inputs they receive via the connections.
Platform: |
Size: 37888 |
Author: kumar |
Hits:
Description: BP神经网络预测公路运量
1.问题的描述
公路运量主要包括公路的客运量和公路货运量两个方面。据研究，某地区的公路运量主要与该地区的人数、机动车数量和公路面积有关，表1给出了20年得公路运量相关数据，表中人数和公路客运量的单位为万人，机动车数量单位为万两，公路面积的单位为万平方千米，公路货运量单位为万吨。
根据有关部门数据，该地区2010年和2011年的人数分别为73.39和75.55万人，机动车数量分别为3.9635和4.0975万辆，公路面积将分别为0.9880和1.0268万平方米。请利用BP神经网络预测该地区2010年2011年得公路客运量和公路货运量。
-BP neural network to predict the road traffic (1) description of the problem of road traffic including road passenger and road freight volume two. According to research, an area of road traffic and the number of the region, the number of motor vehicles and road area, Table 1 shows the 20 years have to road traffic-related data, the table number and road passenger traffic units for people , million two units of the number of motor vehicles, road area in square kilometers, the highway freight volume was 10 000 tons. According to relevant data, the number of people of the region in 2010 and 2011 were 73.39 and 75.55 million, the number of motor vehicles were 3.9635 and 4.0975 million units, the road area were 0.9880 and 1.0268 million square meters. Please use the BP neural network in the region in 2011, 2010 have highway passenger volume and road cargo.
Platform: |
Size: 17408 |
Author: 蔡恩启 |
Hits:
Description: 最小RMSE神经网络方法计算Lyapunov指数的matlab函数。-This M-file calculates Lyapunov exponents with minimum RMSE neural network. After estimation of network weights and finding network with minimum BIC, derivatives are calculated. Sum of logarithm of QR decomposition on Jacobian matrix for observations gives spectrum of Lyapunov Exponents. Using the code is very simple, it needs only an scalar time series, number of lags and number of hidden unites. Higher number of hidden units leads to more precise estimation of Lyapunov exponent, but it is time consuming for less powerful personal computers. Number of lags determines number of embedding dimensions. Therefore, please give number of lags equal to number of embedding dimension. The codes creates networks with various neurons up to user supplied value for neurons and lags up to user specified number lags. Total number of networks are equal to number of neurons times number of lags. this modeling strategy is complex but helps to user select embedding dimension based on minimum BIC.
Platform: |
Size: 2048 |
Author: miaomiao |
Hits:
Description: BP神经网络隐单元个数不同造成误差精度以及训练时间不同-BP neural network the number of hidden units caused the error accuracy and training time
Platform: |
Size: 1024 |
Author: 郜苑 |
Hits:
Description: CLLIB is a varied collection of Common lisp tools and routines in CLOCC.
-CLLIB is a varied collection of Common lisp tools and routines in CLOCC.
Includes:
■ "guess the animal" game simple neural net (AI)
■ autoload function and snarfing autoloads from other files
■ basic definitions: package and path
■ base64 encoding and decoding (data format)
■ Rolodex: BBDB/vCard handling
■ check values and types of the elements of a list
■ Common Lisp HyperSpec access
■ read/write CLOS objects (serialization)
■ read/write comma-separated values
■ CVS diff and log parsing (version control)
■ data analysis and visualization
■ date/time
■ dated lists
■ answer questions automatically requires metering.lisp
■ Load and run Emacs-Lisp ELisp code in Common Lisp
■ read/write lists etc
■ Financial functions: mortgage calculations, Luhn algorithm, Black-Scholes, Solow (mathematics)
■ geography, weather, etc (units)
■ command line option processing- for Lisp scripting
■ Gnuplot interface (plotting)
■ GetQuote- stock quotes over the In
Platform: |
Size: 456704 |
Author: 张茜 |
Hits:
Description: Neural networks typically consist of multiple layers or a cube design, and the signal path traverses front to back. Back propagation is where the forward stimulation is used to reset weights on the front neural units and this is sometimes done in combination with training where the correct result is known. -Neural networks typically consist of multiple layers or a cube design, and the signal path traverses front to back. Back propagation is where the forward stimulation is used to reset weights on the front neural units and this is sometimes done in combination with training where the correct result is known.
Platform: |
Size: 852992 |
Author: crowalt |
Hits: