资源列表
TSP_hopfield
- 这个代码可以有效地解决旅行商问题的组合优化问题(This code can effectively solve the combinatorial optimization problem of traveling salesman problem.)
2
- Hopfield神经网络在matlab实例(Hopfield neural network)
mnist98
- 改进的dnn，准确率达到了百分之98，有注释(The accuracy of the improved DNN is ninety-eight percent, with annotations)
pyDogVsCat
- 识别率85%，kaggle上有名的猫狗大战算法，可以很方便的查看分类结果。每一个epch需要22s左右（GTX1050Ti 4G）(The recognition rate is 85%. The famous dog and dog algorithm on kaggle is very convenient for us to see the classification results. Each epch needs about 22s (GTX1050Ti 4G))
Pattern Recognition and Machine Learning
- for machine learning
co-training
- matlab实现co-training算法，数据集mutiple+Features(Matlab implements co-training algorithm, data set mutiple+Features.)
lle
- lle用来处理高维数据降维，经检验此程序运行效果很好(LLE is used to deal with dimensionality reduction of high-dimensional data. It is proved that the program runs well.)
SparseAutoencoder-master
- 自动编码器是一种数据的压缩算法，其中数据的压缩和解压缩函数是数据相关的、有损的、从样本中自动学习的。在大部分提到自动编码器的场合，压缩和解压缩的函数是通过神经网络实现的。(An autoencoder, autoassociator or Diabolo network is an artificial neural network used for unsupervised learning of efficient codings. The aim of an autoencoder is
新建文件夹
- 基于元胞自动机实现密集人群流动模型，用于大型社会活动风险踩踏评估。(A dense crowd flow model based on cellular automata is applied to risk stampede assessment for large-scale social activities.)
MNIST_data
- Mnist完整数据集，已经实验过，可以应用。(Mnist complete dataset has been tested and applied.)
基于MATLAB的肤色分割和匹配的人脸识别研究
- 对人脸识别的基本进程进行综述,介绍了空间转换及识别的方法；采用低通滤波处理策略,在总结和分析这些处理方法的基础上,运用填孔处理和边缘检测的人脸识别方法进行人脸识别工作；将几种边缘检测算法与人脸识别相结合进行检测,其识别率有可观的提升。(The basic process of face recognition is reviewed, and the method of space conversion and recognition is introduced. On the basis of
simpleCNN
- 在anaconda+opencv+tensorflow平台下，利用简单的CNN卷积神经网络进行手写字符识别(Under the anaconda+opencv+tensorflow platform, we use simple CNN convolution neural network to handwritten character recognition.)