Location:
Search - local minimum
Search list
Description: 本文深入研究了 BP 神经网络与遗传算法理论，BP 神经网络在应用过程中面临
着网络训练时间长、容易陷入局部极小值、隐层节点数不易确定等缺点，为了有效
地克服 BP 网的困难，将遗传算法与 BP 网络有机地融合，使它们之间的相互补充增
强彼此的能力，从而获得更有力的解决实际问题的能力。
-this in-depth study of artificial neural networks and genetic algorithms theory, BP neural network applications in the process of facing network training a long time and easily into the local minimum value, hidden nodes is difficult to determine such shortcomings, in order to effectively overcome the difficulties BP network, Genetic Algorithm and BP organic integration of the network, so that they complement each other between enhance mutual capability, thus more effective to solve practical problems.
Platform: |
Size: 558216 |
Author: 罗旺 |
Hits:
Description: KMEANS Trains a k means cluster model.CENTRES = KMEANS(CENTRES, DATA, OPTIONS) uses the batch K-means
algorithm to set the centres of a cluster model. The matrix DATA
represents the data which is being clustered, with each row
corresponding to a vector. The sum of squares error function is used.
The point at which a local minimum is achieved is returned as
CENTRES.
Platform: |
Size: 1926 |
Author: 西晃云 |
Hits:
Description: 本文深入研究了 BP 神经网络与遗传算法理论，BP 神经网络在应用过程中面临
着网络训练时间长、容易陷入局部极小值、隐层节点数不易确定等缺点，为了有效
地克服 BP 网的困难，将遗传算法与 BP 网络有机地融合，使它们之间的相互补充增
强彼此的能力，从而获得更有力的解决实际问题的能力。
-this in-depth study of artificial neural networks and genetic algorithms theory, BP neural network applications in the process of facing network training a long time and easily into the local minimum value, hidden nodes is difficult to determine such shortcomings, in order to effectively overcome the difficulties BP network, Genetic Algorithm and BP organic integration of the network, so that they complement each other between enhance mutual capability, thus more effective to solve practical problems.
Platform: |
Size: 558080 |
Author: 罗旺 |
Hits:
Description: bp神经网络算法是解决最优化问题的先进算法之一，本论文讨论了神经网络中使用最为广泛的前馈神经网络。其网络权值学习算法中影响最大的就是误差反向传播算法（back-propagation简称BP算法）。BP算法存在局部极小点，收敛速度慢等缺点。基于优化理论的Levenberg-Marquardt算法忽略了二阶项。该文讨论当误差不为零或者不为线性函数即二阶项S(W)不能忽略时的Hesse矩阵的近似计算，进而训练网络。-bp neural network algorithm to solve optimization problems, one of the advanced algorithm, the paper discusses the neural network in the most widely used feed-forward neural network. Its network weights learning algorithm in the greatest impact is the error back-propagation algorithm (back-propagation algorithm referred to as BP). BP algorithm for the existence of local minimum points, such as the shortcomings of slow convergence. Optimization theory based on the Levenberg-Marquardt algorithm ignores the second-order item. In this paper, the discussion when the error is not zero or not that is second-order linear function of S (W) can not be ignored when the Hesse matrix of approximate calculation, and then training the network.
Platform: |
Size: 19456 |
Author: 刘慧 |
Hits:
Description: 最小张树分类法
作最小张树(可考虑按距离给出权值)
在最张树上,确定该树的直径,并标出直径上各点的深度.
绘制直径上个点深度图,由深度图包括找出局部最小值.
去掉局部最小值的点,获得分离的二类.
-The smallest for the smallest classification Zhang Zhang (may consider the right value is given by distance) in the most Zhang, make sure that the tree diameter and diameter at various points marked depth. Drawing on the diameter of the depth map points from depth map including the identification of local minimum. remove the local minimum point, was separated from the second category.
Platform: |
Size: 3072 |
Author: 王顺祥 |
Hits:
Description: KMEANS Trains a k means cluster model.CENTRES = KMEANS(CENTRES, DATA, OPTIONS) uses the batch K-means
algorithm to set the centres of a cluster model. The matrix DATA
represents the data which is being clustered, with each row
corresponding to a vector. The sum of squares error function is used.
The point at which a local minimum is achieved is returned as
CENTRES.
Platform: |
Size: 2048 |
Author: 西晃云 |
Hits:
Description: 针对分水岭算法存在的过分割问题以及VCH-F1切片图像的特点，提出一种能够有效消除局部极小值和噪声干扰的自动分割方法。首先比较彩色分量梯度图，选择分量图像的梯度信息，达到有效提取图像边缘信息的目的；然后提出基于多阈值分割的方法消除无效梯度信息；最后介绍了算法的步骤及结果。实验结果证明，通过该方法处理的梯度图像再进行分水岭算法处理，即使不进行区域合并也能达到很好的效果。-Watershed algorithm for over-segmentation problem of the existence of VCH-F1 as well as the characteristics of the image slice, a can effectively eliminate the local minimum value and noise automatic segmentation method. Comparing the first color component gradient map, select the quantity of image gradient information, achieve an effective extraction of image edge information purposes then made based on multi-threshold segmentation method to eliminate invalid gradient information Finally introduce the steps of the algorithm and results. Experimental results show that the adoption of the method of treatment of gradient watershed algorithm for image re-treatment, even if there is no regional merger will also achieve good results.
Platform: |
Size: 3243008 |
Author: 李胖子 |
Hits:
Description: 人工神经网络系统的训练 TRAIN
BP算法存在局部极小点，收敛速度慢等缺点，改进的BP算法。-Artificial neural network training algorithm TRAINBP local minimum points, such as the shortcomings of slow convergence, improved BP algorithm.
Platform: |
Size: 2048 |
Author: q |
Hits:
Description: 在实际应用中，原始的BP算法很难胜任，因此出现了很多的改进算法。BP算法的改进主要有两种途径，一种是采用启发式学习方法，另一种则是采用更有效的优化算法。本例采用动量BP算法，来实现对网络的训练过程，动量法降低了网络对于误差曲面局部细节的敏感性，有效地抑制网络陷于局部极小。
-In practical applications, the original BP algorithm very difficult to do, so there were a lot of the improved algorithm. BP Algorithm There are two main channels, a heuristic learning method is adopted, another is the introduction of more efficient optimization algorithm. In this case the use of momentum BP algorithm to realize the network training process, momentum method to reduce the network error surface for the local details of the sensitivity of the network effectively inhibited in a local minimum.
Platform: |
Size: 3072 |
Author: zhangaixia |
Hits:
Description: 计算梯度下降法计算极值，只能找到局部最小点。可以通过调整步长实现全局最小-Calculation of gradient descent method to calculate extreme value, can only find local minimum point. By adjusting the step size can achieve the global minimum
Platform: |
Size: 1024 |
Author: 宗丹 |
Hits:
Description: SCE（shuffled complex evolution ）是一种相对较新的连续性问题的元启发搜索算法。非常适合于求解具有多个局部最小的全局优化问题。SCE算法的主要特征是通过竞争进化和定期洗牌来确保每个复形获得的信息能在整个问题空间获得共享。-SCE (shuffled complex evolution) is a relatively new meta-continuity heuristic search algorithm. Very suitable for solving with multiple local minimum of the global optimization problem. SCE algorithm is characterized primarily by evolution through competition and regular cards to ensure that each complex information obtained in the whole question of access to shared space.
Platform: |
Size: 9216 |
Author: 胡军 |
Hits:
Description: matlab格式源代码。功能：利用改进遗传优化算法解决BP神经网络中局部最小问题。-matlab source code format. Function: the use of improved genetic optimization algorithm BP neural network to solve local minimum problems.
Platform: |
Size: 9216 |
Author: magic |
Hits:
Description: 修改以后的BP人工神经网络，可以有效地避开局部最小点，收敛速度有点慢，可作为交流学习-Modified after the BP artificial neural network, can effectively avoid the local minimum points, the convergence rate a bit slow, can be used as the exchange of learning
Platform: |
Size: 1605632 |
Author: 郭龙 |
Hits:
Description: 分类判别中，bayes判别的确具有明显的优势，与模糊，灰色，物元可拓相比，判别准确率一般都会高些，而BP神经网络由于调试麻烦，在调试过程中需要人工参与，而且存在明显的问题，局部极小点和精度与速度的矛盾，以及训练精度和仿真精度间的矛盾，等，尽管是非线性问题的一种重要方法，但是在我们项目中使用存在一定的局限，基于此，最近两天认真的研究了bayes判别，并写出bayes判别的matlab程序，与spss非逐步判别计算结果一致。-Classified Identifying, bayes discriminant does have a distinct advantage, with the fuzzy, gray, matter-element and extension compared to determine the exact rate will be higher in general, and the BP neural network trouble as a result of debugging, in the need to manually debug the process of participation, but also obvious problems, the local minimum point and the accuracy and speed of contradictions, as well as simulation training accuracy and precision of the conflict between, and so on, in spite of nonlinear problems is an important method, but the use of our project there are certain limitations, based on the Here, seriously the last couple of days to study the discriminant bayes and bayes discriminant of matlab to write procedures, and non-spss stepwise discriminant calculation results.
Platform: |
Size: 4096 |
Author: lili |
Hits:
Description: Find local minimum in matlab
Platform: |
Size: 1024 |
Author: idillus |
Hits:
Description: 单纯型搜索MATLAB程序，用于搜索多变量函数的局部极小值。-Search MATLAB simple procedure used to search for multi-variable function of the local minimum value.
Platform: |
Size: 2048 |
Author: 周凯 |
Hits:
Description: Snake的初衷是为了进行图像分割，但它对初始位置过于敏感，且不能处理拓扑结构改变的问题。初始位
置的敏感性可以用遗传算法来克服，因为它是一种全局优化算法，且有良好的数值稳定性。为了更精确地进行图
像分割，本文提出了一种基于遗传算法的双T—Snake模型图像分割方法，它将双T—Snake模型解作为遗传算法的搜
索空间，这既继承了T—Snake模型的拓扑改变能力，又加快了遗传算法的收敛速度。由于它利用遗传算法的全局优
化性能，克服了Snake轮廓局部极小化的缺陷，从而可得到对目标的更精确的分割。将其应用于左心室MRI图像的分割，取得了较好的效果。-Snake s original intention was to carry out image segmentation, but it was too sensitive to initial position and can not deal with the issue of topology change. Initial position
The sensitivity of home can be used to overcome the genetic algorithm because it is a global optimization algorithm, and have good numerical stability. In order to more accurately map
Like segmentation, this paper presents a genetic algorithm based on dual-T-Snake model for image segmentation method, it will double-T-Snake model solution found as a genetic algorithm
Cable space, which not only inherited the T-Snake model the ability to change the topology, but also speed up the convergence rate of genetic algorithm. It uses genetic algorithms as a result of the overall excellent
Of performance, to overcome the local minimum of Snake contour deficiencies, which can be more precise on the target partition. Will be applied to MRI images of left ventricle
Segmentation, and achieved good results.
Platform: |
Size: 458752 |
Author: ultraqiangda |
Hits:
Description: this function will help to find local maximas of an one dimensional data,inputs are data array and minimum gap between two local maximas, it is better to use this function after filtering data using median filter
Platform: |
Size: 1024 |
Author: saneem |
Hits:
Description: optimization algorithm search local minimum function, very quick
Platform: |
Size: 16384 |
Author: zerket |
Hits:
Description: 局部最小距离分类器，性能高于knn分类器，matlab环境下，可直接调用-Local minimum distance classifier, classifier performance than knn, matlab environment, can be called directly
Platform: |
Size: 1024 |
Author: vicky |
Hits: