Location:
Search - ML-k
Search list
Description: Hill加密算法的基本思想是将l个明文字母通过线性变换将它们转换为k个密文字母。脱密只要做一次逆变换就可以了。密钥就是变换矩阵本身。即
M=m1m2……ml
Ek(M)=c1c2……cl
其中
c1=k11m1+k12m2+……+k1lml
c2=k21m1+k22m2+……+k2lml
……
cl=kl1m1+kl2m2+……+kllml
通常对于字母加解密，使用mod 26的方法。
以上线性方程可以采用矩阵表示。
Platform: |
Size: 143581 |
Author: wildkaede |
Hits:
Description: 数据挖掘领域中的一种算法－ML-KNN是一种改进的最近邻算法-Data Mining in the area of an algorithm-ML-KNN is an improved nearest neighbor algorithm
Platform: |
Size: 5120 |
Author: 李平 |
Hits:
Description: Hill加密算法的基本思想是将l个明文字母通过线性变换将它们转换为k个密文字母。脱密只要做一次逆变换就可以了。密钥就是变换矩阵本身。即
M=m1m2……ml
Ek(M)=c1c2……cl
其中
c1=k11m1+k12m2+……+k1lml
c2=k21m1+k22m2+……+k2lml
……
cl=kl1m1+kl2m2+……+kllml
通常对于字母加解密，使用mod 26的方法。
以上线性方程可以采用矩阵表示。
-Hill encryption algorithm the basic idea is to express the letter l a linear transformation through them is converted to k-secret alphabet. Off to do a secret as long as the inverse transform it. Key is the transformation matrix itself. That is, M = m1m2 ... ... mlEk (M) = c1c2 ... ... cl which c1 = k11m1+ K12m2+ ... ...+ K1lmlc2 = k21m1+ K22m2+ ... ...+ K2lml ... ... cl = kl1m1+ Kl2m2+ ... ...+ Kllml For encryption and decryption usually letters, use the method of mod 26. Above linear equations can be used express matrix.
Platform: |
Size: 143360 |
Author: wildkaede |
Hits:
Description: 实现图像的分割。改程序是在matlab环境下运行的，程序中用到了k-means算法，实现的结果很不错-This is the program of image segmentation based on the following work:
S. Chen, L. Cao, J. Liu, and X. Tang, "Image Segmentation by MAP-ML Estimations", submitted to IEEE Transactions on Image Processing.
Note that this program was only tested on 32-bit PCs installed with Windows XP or Windows Server 2003. Since the K-means algorithm used in this program is from Matlab Statistics Toolbox, the program runs slower than when a faster K-means algorithm is used which can be obtained at http://vision.ucsd.edu/~pdollar/toolbox/doc/. The program was tested using Matlab R14 and the image processing Matlab Toolbox is required to run it.
Platform: |
Size: 73728 |
Author: 培培 |
Hits:
Description: code for K Nearest Neighbor algorithm in java
Platform: |
Size: 37888 |
Author: buvi |
Hits:
Description: 通过降维处理，高维数据的分类一般可以转换为2维数据分类。此源码包含一个2维-2类数据分类工具箱。包括：ML，K-NN，SVM，LS，DB-Through the dimension reduction processing, high dimensional data classification commonly can convert to 2 d data classification. This source includes a 2 d-two kinds of data classification toolbox. Including: ML, K-NN by SVM, LS, DB.
Platform: |
Size: 988160 |
Author: 杨明 |
Hits:
Description: 提出了一种基于整体变分模型的图像修复方法，该方法通过计算像素点的梯度信息来构造一个扩散函数，再进行加权处
理，达到了对原模型的改进．实验表明，该方法修复有较大破损区域的图像效果比较好，边缘过渡更自然．-This p鼍|per p他s铷吣蚰iIIlage∞st0枷on algorinlm b鹪ed∞total v蜀一撕∞model，mis metllod证lpraves tbe original m吠ld
tb：∞ugh c∞s蜘Jcting a spmad鼬cti∞by c“culating me gradicnts of the pixel，柚d tll锄canyiIlg out the weightcd appr(舭h．Exp耐·
ments show mat tllis a190rithm、voiks k：恤玎in ttle mst0硎∞of d罩咀瞌ged in蛆ges which have rich tcxtu∞，and h鹪mo地m咖ml bound．
Platform: |
Size: 217088 |
Author: 孙红娟 |
Hits:
Description: Max Welling s Notes 原网页是PS格式的，我都转成了pdf格式方便阅读，机器学习算法相关，都是一些相对基础的算法，这个笔记把很多算法的精髓都整理的很清楚，适合初学者入门看，以及一些新的算法里面可能没有涉及，在英文描述中我把里面涉及的算法列了一下，按需求下载吧。-Max Welling s Notes。Statistical Estimation [ps]
- bayesian estimation
- maximum a posteriori (MAP) estimation
- maximum likelihood (ML) estimation
- Bias/Variance tradeoff & minimum description length (MDL)
Expectation Maximization (EM) Algorithm [ps]
- detailed derivation plus some examples
Supervised Learning (Function Approximation) [ps]
- mixture of experts (MoE)
- cluster weighted modeling (CWM)
Clustering [ps]
- mixture of gaussians (MoG)
- vector quantization (VQ) with k-means.
Linear Models [ps]
- factor analysis (FA)
- probabilistic principal component analysis (PPCA)
- principal component analysis (PCA)
Independent Component Analysis (ICA) [ps]
- noiseless ICA
- noisy ICA
- variational ICA
Mixture of Factor Analysers (MoFA) [ps]
- derivation of learning algorithm
Hidden Markov Models (HMM) [ps]
- viterbi decoding algorithm
- Baum-Welch learning algorithm
Kalman Filters (KF) [ps]
- kalman filter algorithm (very detailed derivation)
-
Platform: |
Size: 1211392 |
Author: 陈希 |
Hits:
Description: 针对单标记学习算法KNN进行改进，适用于多标记数据集改造而成的多标记K临近算法。-Improved learning algorithm for single marker KNN, suitable for multi-label data sets adapted multi-label K near the algorithm.
Platform: |
Size: 1343488 |
Author: SensorNetwork |
Hits:
Description: LCMV优化设计阵列处理信号，ML法能够很好的估计信号的信噪比,基于K均值的PSO聚类算法。- LCMV optimization design array signal processing, ML estimation method can be a good signal to noise ratio, K-means clustering algorithm based on the PSO.
Platform: |
Size: 5120 |
Author: yengjunlao |
Hits:
Description: 在这个项目中，我们的目标是建立一个识别和大小231x231图像呈现对象分类系统。我们得到了一组训练图像，每四个标签之一：1飞机；汽车2；3马，否则。我们提供了两个特点：一是方向梯度直方图（HOG），其尺寸为5408；另一个是overfeat ImageNet美国有线电视新闻网的特点，其尺寸37000。关于测试图像，我们只给出了每个图像的功能，没有标签，结果判断由平地机。我们的目标是提供二进制和多个预测。平衡错误率（BER）是我们的性能评估。为了解决这个问题，我们首先减少PCA的问题的维数，处理不平衡数据集，通过向上采样或下采样，去除异常值，通过无监督学习，如k-均值和EM算法。其次，我们使用ML方法，如二进制和多项式logistic回归，二进制和多项式SVM和神经网络。多项式SVM的证明有最好的结果。最后，我们在100分中得了92分。-In this project, our goal was to build a system that recognizes and classifies the object present in an image of size 231x231. We were given a set of training images each with one of four labels: 1 for airplanes 2 for cars 3 for horses 4 otherwise. We were provided with two sets of features: one is Histogram of Oriented Gradients (HOG), which has dimension of 5408 the other one is OverFEAT ImageNet CNN Features, which has dimension of 37,000. Concerning the test images, we were only given the features of each image without label, and the results to be judged by the grader. Our goal was to provide binary and multiple predictions. The Balanced Error Rate (BER) was our performance uator. To solve the problem, we firstly reduced the problem’s dimensionality by PCA, dealt with imbalanced datasets through up-sampling or down-sampling, and removed outliers through unsupervised learning such as K-Means and EM Algorithm. Secondly, we applied ML methods such as Binary and Multinomial Logist
Platform: |
Size: 322560 |
Author: 杨雪 |
Hits:
Description: 包括调制，解调，信噪比计算，基于K均值的PSO聚类算法,最大似然（ML）准则和最大后验概率（MAP）准则。- Includes the modulation, demodulation, signal to noise ratio calculation, K-means clustering algorithm based on the PSO, Maximum Likelihood (ML) criteria and maximum a posteriori (MAP) criterion.
Platform: |
Size: 4096 |
Author: 付平宝 |
Hits:
Description: 基于K均值的PSO聚类算法,光纤无线通信系统中传输性能的研究，ML法能够很好的估计信号的信噪比。- K-means clustering algorithm based on the PSO, Fiber Transmission wireless communication system performance, ML estimation method can be a good signal to noise ratio.
Platform: |
Size: 5120 |
Author: 文志海 |
Hits:
Description: ML-KNN，这是来自传统的K-近邻（KNN）算法。详细地，为每一个看不见的实例中，首先确定了训练集中的k近邻。之后，基于从标签集获得的统计信息。这些相邻的实例，即属于每个可能类的相邻实例的数量，最大后验（MAP）原理。用于确定不可见实例的标签集。三种不同现实世界中多标签学习问题的实验研究，即酵母基因功能分析、自然场景分类和网页自动分类，表明ML-KNN实现了卓越的性能(ML-KNN which is derived from the traditional K-nearest neighbor (KNN) algorithm. In detail, for each unseen
instance, its K nearest neighbors in the training set are firstly identified. After that, based on statistical information gained from the label sets of
these neighboring instances, i.e. the number of neighboring instances belonging to each possible class, maximum a posteriori (MAP) principle
is utilized to determine the label set for the unseen instance. Experiments on three different real-world multi-label learning problems, i.e. Yeast
gene functional analysis, natural scene classification and automatic web page categorization, show that ML-KNN achieves superior performance
to some well-established multi-label learning algorithms.
2007 Pattern Recognition Society. Published by Elsevier Ltd. All rights reserved.)
Platform: |
Size: 5120 |
Author: 玖
|
Hits:
Description: Matlab codes for knn,k means clustering,linear regression
Platform: |
Size: 228045 |
Author: swarna |
Hits: