LiuSpark-CSDN博客

转载 Inception Score&Mode Score

Inception ScoreMode ScoreReference[1]Tong Che, Yanran Li, Athul Paul Jacob, Yoshua Bengio, Wenjie Li. Mode Regularized Generative Adversarial Networks. In ICLR2017. [2]Improved Techniques for Trainin

2017-05-19 21:00:09 6903

转载 Wasserstein distance&测度

Wasserstein distance维基百科https://www.zhihu.com/question/41752299/answer/147394973测度https://www.zhihu.com/question/28367115

2017-05-15 20:20:25 1422

转载度量空间，赋范空间，內积空间，希尔伯特空间

上海交通大学公开课：数学之旅——3函数空间距离范数内积在前半段各种空间在公开课后半段基本上把概念讲的很清楚了

2017-05-15 19:49:53 1803

转载拉格朗日乘子法和KKT条件

http://www.cnblogs.com/zhangchaoyang/articles/2726873.html

2017-05-14 13:18:44 454

转载再生核Hilbert空间(RKHS)&SVM

Reproducing Kernel Hilbert Space（RKHS）http://www.cnblogs.com/murongxixi/p/3480851.htmlhttp://blog.pluskid.org/?page_id=683

2017-05-14 13:15:43 1232

https://github.com/vdumoulin/conv_arithmetichttps://www.zhihu.com/question/43609045/answer/98699795https://www.zhihu.com/question/43609045/answer/132235276https://www.zhihu.com/question/43609045/answer

2017-04-23 19:34:18 528

转载熵

信息熵相对熵交叉熵各种形式的熵函数，KL距离

2017-04-09 15:32:37 639

原创 20170402#cs231n#13.Others

Segmentation 分割上图中没有对四只牛进行具体的分割，因为四只牛混在一块了，这是 Semantic Segmentation Instance Segmentation 实例分割又称为实时检测与分割需要把每个pixel分到具体的类上面去实际中语义分割和实例分割是分开进行的Semantic Segmentation 语义分割multi-scaleRefinementUpsampling

2017-04-02 09:34:13 950

原创 20170328#cs231n#12.CNNs in Practice

得到更多数据在小数据集用数据增强和迁移学习是很合适的Data augmentation数据增强这是一种很容易实现的方法。在图片输入CNN之前先做一些变化，应该是改变了每一个pixel，但是图片的label保持不变。可能减小过拟合的发生。Horizontal flips 水平翻转Random crops/scales 随机裁剪/缩放图片（在test的时候要对test set进行裁剪缩放水平翻转

2017-03-28 21:29:55 638

原创 20170326#cs231n#11.Recurrent Neural Networks 循环神经网络RNN

主要参考资料简书——[译] 理解 LSTM 网络 csdn——循环神经网络(RNN, Recurrent Neural Networks)介绍 cs231n RNN PPTRNNkarpathy/min-char-rnn.py https://gist.github.com/karpathy/d4dee566867f8291f086 图片转文字描述CNN→RNN把CNN最后的分类器去掉，然后

2017-03-26 14:48:44 2072

原创 20170325#cs231n#10.Understanding and Visualizing Convolutional Neural Networks

Visualize patches that maximally activate neurons这个意思是是把数据输入某一层中，然后看数据的哪一部分最能激活这层的神经元Visualize the filters/kernels (raw weights) 但对高层的weight可视化的意义就不是特别大了Visualizing the representationt-SNE visualizati

2017-03-25 15:32:09 1777

原创 20170324#cs231n#9.ConvNets for spatial localization & Object detection

Computer Vision Tasks

2017-03-24 11:08:25 893

原创 #cs231n#Assignment2：ConvolutionalNetworks.ipynb

根据自己的理解和参考资料实现了一下 ConvolutionalNetworks.ipynbConvolutional NetworksSo far we have worked with deep fully-connected networks, using them to explore different optimization strategies and network archit

2017-03-20 09:32:34 2031

原创 #cs231n#Assignment2：Dropout.ipynb

根据自己的理解和参考资料实现了一下 Dropout.ipynbDropoutDropout [1] is a technique for regularizing neural networks by randomly setting some features to zero during the forward pass. In this exercise you will implement

2017-03-19 10:25:17 1489

原创 #cs231n#Assignment2：BatchNormalization.ipynb

根据自己的理解和参考资料实现了一下 BatchNormalization.ipynbBatch NormalizationOne way to make deep networks easier to train is to use more sophisticated optimization procedures such as SGD+momentum, RMSProp, or Adam.

2017-03-19 10:24:06 2058

原创 #cs231n#Assignment2：FullyConnectedNets.ipynb

根据自己的理解和参考资料实现了一下 FullyConnectedNets.ipynbFully-Connected Neural NetsIn the previous homework you implemented a fully-connected two-layer neural network on CIFAR-10. The implementation was simple but

2017-03-19 10:20:22 4236

原创 20170316#cs231n#8.Convolutional Neural Networks

Convolutional Neural Networks (CNNs / ConvNets)

2017-03-16 19:41:41 2008

原创 20170307#cs231n#7.Neural Networks Part 3: Learning and Evaluation

Gradient Checks/momentum/AdaGrad/RMSProp/Adam/...

2017-03-07 14:22:46 953

原创 20170304#cs231n#6.Neural Networks Part 2: Setting up the Data and the Loss

Data Preprocessing 数据预处理Mean subtraction 零均值化：对每一个特征减去均值 X -= np.mean(X, axis = 0)Normalization 归一化：把数据的所有维度归一化，使其数值范围近似相等，归一化操作只有在确定不同输入特征有不同数值范围（或单位）才有意义，但是图像处理中由于范围为[0,255]所以这个数据预处理的步骤意义不是特别大。归一化大

2017-03-04 18:12:30 774

原创 20170228#cs231n#5.Neural Networks Part 1神经网络1 /Assignment1-NeuralNetwork

Neural Networks Part 1: Setting up the Architecture这个是NeuralNetwork的计算公式s=W2max(0,W1x)s = W_2 \max(0, W_1 x) xx [3072×1] W1W_1[100×3072] W2W_2 [10×100] max函数是一个非线性函数，与我们之前的不同，正是因为这个改变使得其与线性函数不同。参数W

2017-02-28 10:28:13 639

原创 20170226#cs231n#4.Backpropagation反向传播

Backpropagation反向传播反向传播是一种利用链式法则chain rule去递归计算表达式梯度的方法训练集的xix_i的梯度仍然是有用的，例如在将神经网络所做的事可视化以便于理解的时候参考资料：CS231n课程笔记翻译：反向传播笔记-知乎智能单元-杜客ChainRule链式法则∂f∂x=∂f∂q∂q∂x\frac{\partial f}{\partial x}=\frac{\par

2017-02-26 23:12:10 712

原创 20170225#cs231n#3.最优化问题

Optimization:Stochastic Gradient DescentOptimization是寻找一个WW使得LossFunction最小化的过程SVMcostFunction就是一个convex function（凸函数），然后这就涉及到了 ConvexOptimization（凸优化）但是神经网络的代价函数就是non-convex的了然后会有很多lossfunction是Non

2017-02-25 16:56:04 606

原创 #cs231n#相关资源

CS231n Notes CS231n教学计划网易公开课CS231n—深度学习与计算机视觉 xieyi4650的博客CS231n专题

2017-02-25 10:31:02 362

转载 main函数的参数问题

带参数的main函数 main函数的第三个参数【C语言】C语言中main函数问题

2017-02-12 01:04:29 377

原创 20170202 Coursera Stanford-MachineLearning/Week9

Week9:Anomaly detection/Recommender Systems 异常检测/推荐系统Anomaly detection 异常检测训练样本在中心的概率最大所以test如果在中心表明正常Gaussian Distribution 高斯分布(正态分布)x∼N(μ,σ2)P(x;μ,σ2)=12π‾‾‾√σe−(x−μ)22σ2x \sim N(\mu,\sigma^{2}) \\

2017-02-02 20:52:28 1097

原创 20170129 Coursera Stanford-MachineLearning/Week8

Week8:Unsupervised Learning非监督学习Supervised learning & Unsupervised learning 监督学习与非监督学习Supervised learning 监督学习：给出y值，已经为数据提供label了Unsupervised learning 非监督学习：从未标记的数据中学习，要求算法替我们分析出数据的结构Clustering 聚类算法聚

2017-01-29 18:06:14 888

转载线性代数相关

线性代数

2017-01-29 12:05:13 380

原创 20170125 Coursera Stanford-MachineLearning/Week4-5

Week4/5:Neural Networks: Representation/Learning

2017-01-25 11:58:01 484

原创 20170123 Coursera Stanford-MachineLearning/Week7

Week7:Support Vector Machine(SVM) 支持向量机SVM又称为大间距分类器(Large Margin Classifier)

2017-01-23 17:22:11 731

原创 Linux 命令和基础

Linux入门

2017-01-19 15:52:52 338

原创 20161206#cs231n#2.线性分类器 Assignment1--SVM&Softmax

Linear classification: Support Vector Machine, SoftmaxLinear Classifier线性分类器

2016-12-08 12:50:28 600

原创 20161202 Coursera Stanford-MachineLearning/Week10-11

Week10：Large Scale Machine LearningStochastic Gradient Descent

2016-12-05 00:10:57 770

原创 20161129 Coursera Stanford-MachineLearning/Week6

Week6:Advice for Applying Machine Learning&Machine Learning System Design

2016-11-29 21:45:09 468

原创 20161106#cs231n#1.最近邻分类器 Assignment1-KNN

原创 20161113#CVPR#学习笔记

20161113CVPR学习笔记

2016-11-24 18:29:43 384

原创 20161124 Coursera Stanford-MachineLearning/Week1-3

学习笔记

2016-11-24 18:25:40 422

LiuSpark

原创 GAN——ModeCollapse