chenshen03 / chenshen03.github.io Goto Github PK

https://chenshen.xyz/categories/deep-learning/%E7%A5%9E%E7%BB%8F%E7%BD%91%E7%BB%9C%EF%BC%9A%E9%9D%99%E6%80%81%E9%83%A8%E5%88%86/

白化白化操作的输入是特征基准上的数据，然后对每个维度除以其特征值来对数值范围进行归一化。该变换的几何解释是：如果数据服从多变量的高斯分布，那么经过白化后，数据的分布将会是一个均值为零，且协方差相等的矩阵。该操作的代码如下： 123# 对数据进行白化操作:# 除以特征值 Xwhite = Xrot / np.sqrt(S + 1e-5)

https://chenshen.xyz/categories/software-engineering/%E9%A2%86%E9%AA%91%E8%A1%AB%E7%9A%84%E6%9C%9D%E5%9C%A3%E4%B9%8B%E6%97%85/

前言因为变态的考试月加上放假后嗨过头，这篇领骑衫的博客姗姗来迟，看到小明和林锦辛苦整理的博客书里面，领骑衫的博客独缺我一个，真是蜜汁尴尬，为自己死活改不掉的拖延症说一声抱歉。刚看到大史助教里面提到的拖延症患者，我可能就是其实一位吧。完美主义者加拖延癌晚期，活该自己天天熬夜哈哈哈~

https://chenshen.xyz/categories/machine-learning/%E6%94%AF%E6%8C%81%E5%90%91%E9%87%8F%E6%9C%BA/

最大间隔分类器函数间隔：$γ^{i} = y^{i}(w^{T} x + b)$， 改变w和b的量级，对分类结果不会产生任何影响，但是会改变函数间隔的大小。因此，直接对函数间隔求最大值是没有任何意义的。因为可以通过任意改变w、b的量级，使得函数间隔任意大。

https://chenshen.xyz/categories/algorithms/POJ-2393-%E7%AE%80%E5%8D%95%E8%B4%AA%E5%BF%83%E6%B0%B4%E9%A2%98/

题目DescriptionThe cows have purchased a yogurt factory that makes world-famous Yucky Yogurt. Over the next N (1 <= N <= 10,000) weeks, the price of milk and labor will fluctuate weekly such that

https://chenshen.xyz/categories/algorithms/POJ-3176/

题目###Description The cows don’t use actual bowling balls when they go bowling. They each take a number (in the range 0..99), though, and line up in a standard bowling-pin-like triangle like this:

https://chenshen.xyz/categories/algorithms/POJ-3255-%E6%AC%A1%E7%9F%AD%E8%B7%AF%E5%BE%84/

题目DescriptionBessie has moved to a small farm and sometimes enjoys returning to visit one of her best friends. She does not want to get to her old home too quickly, because she likes the scenery along

https://chenshen.xyz/categories/machine-learning/%E7%BA%BF%E6%80%A7%E5%9B%9E%E5%BD%92%E5%92%8C%E9%80%BB%E8%BE%91%E5%9B%9E%E5%BD%92/

线性回归批量梯度下降法 每次对参数进行一次迭代时，都要扫描一遍输入全集 算法可以收敛到局部最优值 当迭代多次之后，每次迭代参数的改变越小

https://chenshen.xyz/about/

https://chenshen.xyz/categories/algorithms/POJ-3187-%E6%9E%9A%E4%B8%BE%E5%85%A8%E6%8E%92%E5%88%97/

、题目DescriptionFJ and his cows enjoy playing a mental game. They write down the numbers from 1 to N (1 <= N <= 10) in a certain order and then sum adjacent numbers to produce a new list with one

https://chenshen.xyz/categories/machine-learning/%E7%BB%8F%E9%AA%8C%E9%A3%8E%E9%99%A9%E6%9C%80%E5%B0%8F%E5%8C%96/

经验风险最小化有限假设类情形对于Chernoff bound 不等式，最直观的解释就是利用高斯分布的图象。而且这个结论和中心极限定律没有关系，当m为任意值时Chernoff bound均成立，但是中心极限定律不一定成立。

https://chenshen.xyz/categories/algorithms/POJ-3669-BFS/

题目DescriptionBessie hears that an extraordinary meteor shower is coming; reports say that these meteors will crash into earth and destroy anything they hit. Anxious for her safety, she vows to find he

https://chenshen.xyz/categories/study/%E3%80%8A%E8%AE%A1%E7%AE%97%E6%99%BA%E8%83%BD%E3%80%8B%E7%AC%94%E8%AE%B0/

最优化算法广度优先搜索总是在某一深度上先搜索所有节点，之后搜索下一个深度的节点，能保证一定可以得到最优解，但需要生成大量节点，并且有可能导致组合爆炸，搜索效率低。

https://chenshen.xyz/categories/algorithms/POJ-2718-%E5%85%A8%E6%8E%92%E5%88%97-%E6%9A%B4%E5%8A%9B/

题目DescriptionGiven a number of distinct decimal digits, you can form one integer by choosing a non-empty subset of these digits and writing them in some order. The remaining digits can be written down

https://chenshen.xyz/categories/essay/2018-2019%E5%AD%A6%E5%B9%B4%E7%A0%94%E4%B8%80%E4%B8%8A%E5%AD%A6%E6%9C%9F%E6%80%BB%E7%BB%93/

该文被密码保护。

https://chenshen.xyz/categories/machine-learning/%E5%BA%94%E7%94%A8%E6%9C%BA%E5%99%A8%E5%AD%A6%E4%B9%A0%E7%9A%84%E5%BB%BA%E8%AE%AE/

偏差方差权衡 使用较小的神经网络，类似于参数较少的情况，容易导致高偏差和欠拟合，但计算代价较小使用较大的神经网络，类似于参数较多的情况，容易导致高方差和过拟合，虽然计算代价比较大，但是可以通过归一化手段来调整而更加适应数据。

https://chenshen.xyz/categories/algorithms/POJ-3616/

题目###Description Bessie is such a hard-working cow. In fact, she is so focused on maximizing her productivity that she decides to schedule her next N (1 ≤ N ≤ 1,000,000) hours (conveniently labeled 0.

https://chenshen.xyz/categories/algorithms/POJ-3040-%E5%B1%80%E9%83%A8%E6%9C%80%E4%BC%98%E5%88%B0%E5%85%A8%E5%B1%80%E6%9C%80%E4%BC%98/

题目DescriptionAs a reward for record milk production, Farmer John has decided to start paying Bessie the cow a small weekly allowance. FJ has a set of coins in N (1 <= N <= 20) different denomina

https://chenshen.xyz/categories/algorithms/POJ-3050-%E5%9F%BA%E7%A1%80DFS/

题目DescriptionThe cows play the child’s game of hopscotch in a non-traditional way. Instead of a linear set of numbered boxes into which to hop, the cows create a 5x5 rectilinear grid of digits paralle

https://chenshen.xyz/categories/algorithms/%E4%B9%9D%E5%A4%A7%E6%8E%92%E5%BA%8F%E7%AE%97%E6%B3%95%E6%B1%87%E6%80%BB/

算法说明数据结构中经常需要用到各种排序算法，故参考网上代码，将九个排序算法整合在一起，以便日后使用。算法运行时，可以选择所要采用的排序算法，并会输出每一趟的排序过程，更利于对排序算法的理解。

https://chenshen.xyz/categories/deep-learning/%E5%A4%9A%E7%B1%BBSVM/

梯度求导SVM的损失函数在某个数据点上的计算：

https://chenshen.xyz/categories/deep-learning/softmax%E5%88%86%E7%B1%BB%E5%99%A8/

wiki百科：softmax函数的本质就是将一个K维的任意实数向量压缩（映射）成另一个K维的实数向量，其中向量中的每个元素取值都介于（0，1）之间。 softmax公式意义在softmax函数，输入向量z的值有正有负，正数表示对应的特征对分类结果是积极的，负数则表示是消极的。因此，在softmax函数中，要 先计算$e^z$, 目的是为了把所有的输入先处理到大于0的空间内，比如负数经过计算后会

https://chenshen.xyz/categories/algorithms/POJ-3262/

题目DescriptionFarmer John went to cut some wood and left N (2 ≤ N ≤ 100,000) cows eating the grass, as usual. When he returned, he found to his horror that the cluster of cows was in his garden eating

https://chenshen.xyz/categories/machine-learning/%E9%AB%98%E6%96%AF%E6%B7%B7%E5%90%88%E6%A8%A1%E5%9E%8B/

高斯混合模型 软分类算法，即对每一个样本，计算其属于各个分布的概率，概率值最大的就是这个样本所属的分类。 对于训练样本的分布，看成为多个高斯分布加权得到的。其中每个高斯分布即为某一特定的类。

https://chenshen.xyz/categories/deep-learning/%E7%A5%9E%E7%BB%8F%E7%BD%91%E7%BB%9C%EF%BC%9A%E5%8A%A8%E6%80%81%E9%83%A8%E5%88%86/

训练技巧损失函数可视化损失函数一般都是定义在高维度的空间中，这样要将其可视化就很困难。然而办法还是有的，在1个维度或者2个维度的方向上对高维空间进行切片，例如，随机生成一个权重矩阵W，该矩阵就与高维空间中的一个点对应。然后沿着某个维度方向前进的同时记录损失函数值的变化。换句话说，就是生成一个随机的方向$W_1$并且沿着此方向计算损失值，计算方法是根据不同的$\alpha$值来计算$L(W+aW_1

https://chenshen.xyz/categories/algorithms/POJ-3190-%E5%8C%BA%E9%97%B4%E9%97%AE%E9%A2%98/

题目DescriptionOh those picky N (1 <= N <= 50,000) cows! They are so picky that each one will only be milked over some precise time interval A..B (1 <= A <= B <= 1,000,000), which include

https://chenshen.xyz/categories/essay/%E6%88%98%E4%BA%89%E4%B8%AD%E7%9A%84%E9%81%93%E5%BE%B7/

**发明了火药，且用火药制造烟花用于庆礼；火药传入西方国家后，西方国家用火药制造弹药、炸弹，进行军事改革，并且用**发明的火药炸开了**的国门。面对这一史实，许多**人无不痛恨当时政府的无能，嘲笑当时政府的迂腐可笑，抱怨他们为什么只懂得享受，没有看到世界之大，不会利用火药去强大国家，导致**跟不上世界工业革命的脚步，以至于传承五千多年的文明古国在如今却要在仅仅建国一百多年的美国下唯唯诺诺！

https://chenshen.xyz/categories/algorithms/POJ-2229/

题目###Description Farmer John commanded his cows to search for different sets of numbers that sum to a given number. The cows use only numbers that are an integer power of 2. Here are the possible sets

https://chenshen.xyz/categories/computer-vision/CNNH/

[AAAI 2014] Supervised Hashing via Image Representation Learning [paper] [code]Rongkai Xia , Yan Pan, Hanjiang Lai, Cong Liu, Shuicheng Yan. Overcome之前的哈希方法，大都使用手工的图像特征（如GIST等）作为图像的特征表达， 但是这些手工特征是采用

https://chenshen.xyz/categories/algorithms/POJ-1328-%E5%8C%BA%E9%97%B4%E9%97%AE%E9%A2%98/

题目DescriptionAssume the coasting is an infinite straight line. Land is in one side of coasting, sea in the other. Each small island is a point locating in the sea side. And any radar installation, loc

https://chenshen.xyz/categories/algorithms/POJ-2385/

题目###Description It is a little known fact that cows love apples. Farmer John has two apple trees (which are conveniently numbered 1 and 2) in his field, each full of apples. Bessie cannot reach the a

https://chenshen.xyz/categories/algorithms/POJ-1862-%E7%AE%80%E5%8D%95%E5%93%88%E5%A4%AB%E6%9B%BC/

题目DescriptionOur chemical biologists have invented a new very useful form of life called stripies (in fact, they were first called in Russian - polosatiki, but the scientists had to invent an English

https://chenshen.xyz/categories/software-engineering/%E7%BB%93%E5%AF%B9%E7%BC%96%E7%A8%8B1%EF%BC%9A%E9%9C%80%E6%B1%82%E5%88%86%E6%9E%90%E4%B8%8E%E5%8E%9F%E5%9E%8B%E8%AE%BE%E8%AE%A1/

需求分析（NABCD模型）N（Need，需求） 师生互选，改变导师被动分配学生的局面 每个学生分配到预想的导师，避免分配了志愿以外的导师的情况出现 学生可以了解到所有导师的课题和研究方向

https://chenshen.xyz/categories/software-engineering/%E6%AF%95%E8%AE%BE%E9%80%89%E5%AF%BC%E7%B3%BB%E7%BB%9F%EF%BC%9A%E5%AE%A3%E4%BC%A0%E6%96%87%E6%A1%88/

产品概述　　毕设导师智能分配系统是一个用来简化传统手工匹配繁琐操作的系统。本系统将学生报志愿、系负责人收集整理数据、相关人员进行手工分配、反馈选择结果等繁琐的操作转移到线上。把毕设导师互选的所有流程，传化对本系统的操作。减少了相关人员的工作量，降低了流程中由于手工操作而出现错误的可能。学生的志愿选择、导师分配、数据统计、结果查看及导出等操作均可在上系统完成，提高了毕设导师选择的效率。

https://chenshen.xyz/categories/algorithms/POJ-2718-%E8%B4%AA%E5%BF%83-%E6%9E%9A%E4%B8%BE/

题目DescriptionGiven a number of distinct decimal digits, you can form one integer by choosing a non-empty subset of these digits and writing them in some order. The remaining digits can be written down

https://chenshen.xyz/categories/algorithms/LeetCode-155/

题目Design a stack that supports push, pop, top, and retrieving the minimum element in constant time. push(x) – Push element x onto stack. pop() – Removes the element on top of the stack. top() – Get t

https://chenshen.xyz/categories/software-engineering/%E8%BD%AF%E5%B7%A5%E5%AE%9E%E8%B7%B5%E4%B8%AA%E4%BA%BA%E6%80%BB%E7%BB%93/

驻足当下课程期望对比开学初的课程期望——软件工程的实践项目课程的自我目标： 对项目的整个开发流程有一个全面专业的了解 掌握《构建之法》的精髓，懂得如何管理一个开发团队 代码能力以及代码规范得到进一步的提升 做出一个具有网站端又有移动端的项目 项目能在各类比赛中大放异彩，并进一步商业化开发，投入市场 实践课保持依旧的魔鬼强度，又有所新的创新，让我们在敲代码的时候享受到快乐

https://chenshen.xyz/categories/machine-learning/%E5%9B%A0%E5%AD%90%E5%88%86%E6%9E%90%E6%A8%A1%E5%9E%8B/

协方差矩阵协方差矩阵为对称矩阵。 在高斯分布中，方差越大，数据分布越分散，方差越小发，数据分布越集中。 在协方差矩阵中，假设矩阵为二维，若第二维的方差大于第一维的方差，则在图像上的体现就是：高斯分布呈现一个椭圆形，且主轴对应的就是方差大的第二维度。简而言之，若对角线元素相等，则高斯分布的图形是圆形，反之则分布图形为椭圆形。

https://chenshen.xyz/categories/software-engineering/%E7%BB%93%E5%AF%B9%E7%BC%96%E7%A8%8B2%EF%BC%9A%E6%AF%95%E8%AE%BE%E9%80%89%E5%AF%BC%E7%B3%BB%E7%BB%9F/

问题重述编码实现一个毕设导师的智能匹配的程序。提供输入包括：30个老师（包含带学生数的要求的上限，单个数值，在[0,8]内），100个学生（包含绩点信息），每个学生有5个导师志愿（志愿的导师可以重复但不能空缺）。实现一个智能自动分配算法，根据输入信息，输出导师和学生间的匹配信息（一个学生只能有一个确认导师，一个导师可以带少于等于其要求的学生数的学生）及未被分配到学生的导师和未被导师选中的学生。

https://chenshen.xyz/categories/deep-learning/%E7%A5%9E%E7%BB%8F%E7%BD%91%E7%BB%9C%EF%BC%9A%E6%BF%80%E6%B4%BB%E5%87%BD%E6%95%B0/

激活函数Sigmoidsigmoid将输入实数值“挤压”到0到1范围内。更具体地说，很大的负数变成0，很大的正数变成1。它对于神经元的激活频率有良好的解释：从完全不激活到在求和后的最大频率处的完全饱和（saturated）的激活。然而现在sigmoid函数实际很少使用了，这是因为它有两个主要缺点：

https://chenshen.xyz/categories/algorithms/POJ-3723-%E6%9C%80%E5%A4%A7%E7%94%9F%E6%88%90%E6%A0%91/

题目DescriptionWindy has a country, and he wants to build an army to protect his country. He has picked up N girls and M boys and wants to collect them to be his soldiers. To collect a soldier without a

https://chenshen.xyz/categories/software-engineering/%E6%AF%95%E8%AE%BE%E9%80%89%E5%AF%BC%E7%B3%BB%E7%BB%9F%EF%BC%9A%E6%B5%8B%E8%AF%95/

测试用例1 测试内容 获取下拉框的输入测试 测试代码 12345678910$("#sub-confirm").click(function() { gradeSelected = document.getElementById("gradeSelect").value; departmentSelected = document.getElementById("depar

https://chenshen.xyz/categories/algorithms/POJ-3169-%E5%B7%AE%E5%88%86%E7%BA%A6%E6%9D%9F%E7%B3%BB%E7%BB%9F/

题目DescriptionLike everyone else, cows like to stand close to their friends when queuing for feed. FJ has N (2 <= N <= 1,000) cows numbered 1..N standing along a straight line waiting for feed. T

https://chenshen.xyz/categories/algorithms/POJ-1017/

题目DescriptionA factory produces products packed in square packets of the same height h and of the sizes 11, 22, 33, 44, 55, 66. These products are always delivered to customers in the square parcels o

https://chenshen.xyz/categories/machine-learning/%E5%9F%BA%E4%BA%8E%E6%9C%B4%E7%B4%A0%E8%B4%9D%E5%8F%B6%E6%96%AF%E7%9A%84%E4%B8%AD%E6%96%87%E5%A4%9A%E5%88%86%E7%B1%BB%E5%99%A8/

算法说明 为了便于计算类条件概率$P(x|c)$,朴素贝叶斯算法作了一个关键的假设：对已知类别，假设所有属性相互独立。 当使用训练完的特征向量对新样本进行测试时，由于概率是多个很小的相乘所得，可能会出现下溢出，故对乘积取自然对数解决这个问题。 在大多数朴素贝叶斯分类器中计算特征向量时采用的都是词集模型，即将每个词的出现与否作为一个特征。而在该分类器中采用的是词袋模型，即文档中每个词汇的出现次数作为

https://chenshen.xyz/categories/technique/Linux%E7%B3%BB%E7%BB%9F%E7%9F%A5%E8%AF%86%E6%B1%87%E6%80%BB/

系统相关静态IP地址配置 Ubuntu配置和修改IP地址 硬盘分区和挂载 parted分区和挂载及非交互式操作

https://chenshen.xyz/categories/essay/%E6%B5%85%E8%B0%88%E6%95%99%E8%82%B2/

十几天的寒窗苦读难道仅仅只是为了高考吗？ 我们从一出生就被各式各样的“考”所缠身。小考、中考、高考，这其间还夹杂着太多太多的测试！老师们不停地强调着各类考试的重要性，将其与人生航向联系起来，还自创什么“此考若失利，人生即悲剧”来激励我们。背负了太多压力的我们不得不整日伏案提笔，做着那永远也做不完的考卷。

https://chenshen.xyz/categories/computer-vision/%E8%AE%A1%E7%AE%97%E6%9C%BA%E8%A7%86%E8%A7%89%E9%A1%B6%E5%B0%96%E6%9C%9F%E5%88%8A%E5%92%8C%E4%BC%9A%E8%AE%AE/

计算机视觉会议A类 CVPR: International Conference on Computer Vision and Pattern Recognition ICCV: International Conference on Computer Vision

https://chenshen.xyz/categories/essay/%E8%84%91%E9%83%A8%E5%BC%80%E5%8F%91%E5%A5%87%E6%83%B3/

该文被密码保护。

https://chenshen.xyz/categories/algorithms/POJ-2376-%E5%8C%BA%E9%97%B4%E9%97%AE%E9%A2%98/

题目DescriptionFarmer John is assigning some of his N (1 <= N <= 25,000) cows to do some cleaning chores around the barn. He always wants to have one cow working on cleaning things up and has divi

https://chenshen.xyz/categories/algorithms/POJ-3009-DFS-%E5%9B%9E%E6%BA%AF/

题目DescriptionOn Planet MM-21, after their Olympic games this year, curling is getting popular. But the rules are somewhat different from ours. The game is played on an ice game board on which a square