# QA Official

### OrdinaryLeader Squares is What Do You Mean?

Ordinary Least Square是普通最小二乘法，简称OLS。是应用最多的参数估计方法，也是从最小二乘原理出发的其他估计方法的基础。英语的解释： In statistics, ordinary least squares (OLS)

### Reasons and Solutions for loss Equal to 87.3365 and Remaining the Same in CAFFE (Convolutional Architecture for Fast Feature Embedding) Training

https://qaofficial.com/post/2019/04/28/69309-reasons-and-solutions-for-loss-equal-to-87.3365-and-remaining-the-same-in-caffe-convolutional-architecture-for-fast-feature-embedding-training.html 2019-04-28
Such as the topic, when training in CAFFE (Convolutional Architecture for Fast Feature Embedding), caffe meeting this special number, loss will always be this number. Although there are many trick on the Internet, there is little detailed analysis. Therefore, it is necessary to study the source code of CAFFE (Convolutional Architecture for Fast Feature Embedding). Softmax's formula is p_k= \frac{exp(x_k)}{\sum_{i}exp(x_i)} Where x is the output of the previous layer of softmax

### Role of Mentality

https://qaofficial.com/post/2019/04/28/63994-role-of-mentality.html 2019-04-28

//选择排序（有头节点） template<typename T> void chainWithHeader<T>::selectionSort() {chainNode<T>*pd,*pf,*pe,*pa,*pmax;pf = headerNode;pd = NULL;bool sorted = false; //已经排好序之后退出循环while (pf->next->next!=pd&&!sorted){pmax = pf;pa= pf->next;sorted = true;T temp;while (pa->next!= pd){if (pmax->next->

### Some Precautions for Use of OpenCV&amp;#x27;s reshape Function

https://qaofficial.com/post/2019/04/28/69153-some-precautions-for-use-of-opencvamp#x27s-reshape-function.html 2019-04-28
1) In matlab, reshape comes in columns, while in opencv, reshape of mat comes in rows. 2)reshape requires the transformation object matrix A to be continuous and can be judged by A.iscontinous ().If it is not continuous, it will not be able to devote all one's efforts to reshape, and it will report an error.Generally, the cropped Mat images are no longer continuous, such as cv::Mat B = A(rect);Crop_img, cv:: Matb = a (range (begin _ row, end _ row), etc.

### Start learning speech recognition technology

https://qaofficial.com/post/2019/04/28/69009-start-learning-speech-recognition-technology.html 2019-04-28
As a front-end ordinary employee working in telecom, inside decided to devote himself to the study of speech recognition in the next day because he was more interested in technical research (in fact, it was also an escape from insufficient marketing ability) and solved the problems encountered in his working life. Continue to keep learning progress updated on a daily basis. The text here is used to urge you to hurry up.

### Use and Difference of DES and md5

https://qaofficial.com/post/2019/04/28/68880-use-and-difference-of-des-and-md5.html 2019-04-28
1.DES algorithm has three entry parameters: Key, Data and Mode.The Key is 7 bytes and 56 bits, which is the working key of DES algorithm.Data is 8 bytes and 64 bits, which is data to be encrypted or decrypted;Mode is DES's working mode, there are two kinds: encryption or decryption.Des is reversible and can be decrypted using a seven-bit key. <span style="font-family:Comic Sans MS;"><span style="font-size:12px;">package com.sica.des;  import com.

### [ Scientific Research Notes ] Considerations on Artificial Intelligence and Algorithm Projects

https://qaofficial.com/post/2019/04/28/69182-scientific-research-notes-considerations-on-artificial-intelligence-and-algorithm-projects.html 2019-04-28
Preface [ scientific research notes ] series is my essays and thoughts on the road of scientific research. the content is not limited. it is an open series of articles, which also gives me some free space. This article is some thoughts on artificial intelligence technology and algorithm project management based on my own experience.You are welcome to criticize and correct me and put forward valuable opinions. The subsection of the article is arranged as follows: