Some time ago, I went to an interview with an intern of byte skipping algorithm , Flow scheduling direction , Record the death interview .
First of all, a fixed self introduction , Then I asked about the written face recognition project , It's used SVM after , Just a push SVM The process of , What is support vector , Specific in the project SVM How to achieve it （ It's equivalent to the feeling of knowing the formula and doing the practical problems ）, Then I asked why we should grayscale the picture , Loss of information （ I'm talking about acceleration ）, How to speed up the calculation without losing information .
And then I asked CNN,RGB What is the channel , What is convolution , Why can I use color pictures directly .
And then it starts to enter the dead computer mode ：
1, You said you learned python, Then you know python Global lock ?
2, You know? TCP agreement , Three handshakes （ And some computer knowledge that I can't understand ）?
After hearing that I didn't know anything , The interviewer said that the basic knowledge of computer network should be understood and mastered . And then I came up with two math questions .
1, hold 52 A playing card is divided into three piles , Ask big and small Wang the probability of appearing in a pile at the same time
2, There is an uneven coin , Positive upward probability 0.7, Negative upward probability 0.3, How to design rules so that two people can play a coin toss game fairly ?
The preparation for this interview is very poor , This is my first interview for an internship , But when I put in my resume for this position job description
It's about data analysis , But when I went, I was interviewed for algorithm , The interviewer and I are helpless , Finally, he suggested that I take a look at the content of data structure , Brush more leetcode. In fact, we should write more python, See more about data structure , Machine learning algorithm must be able to deduce , There are many basic math interview questions , You can also have a look at it more often .