Baidu 2024 Campus Recruitment Interview Experience in Machine Learning, Data Mining, and Natural Language Processing

  This article introduces the basic interview situation, questions, coding questions, etc. for Baidu 's machine learning/data mining/natural language processing engineer position in the 2024 autumn recruitment .

  In early August, I participated in the interview for the position of machine learning/data mining/natural language processing engineer approved by Baidu in advance , and my department was in the search direction. After one side is over, you will know that it is cool. Let me share the sutra of one side.

  Among them, I feel that the questions in the question-and-answer session will be very detailed, and the interviewer will ask further questions based on your answer to the previous question to test your detailed understanding of the principles of machine learning and deep learning algorithms. The interviewer was very kind and I gained a lot of new understanding about machine learning algorithms from the entire interview process.

  This is also the first interview for the autumn recruitment, and I really feel that compared with the intern interview, the autumn recruitment interview pays more attention to the understanding of the code, the basic principles of the algorithm, and the bottom layer (of course, this also depends on the specific job direction you are applying for) ; At the same time, you must brush up on the questions - during this interview, the interviewer also kept emphasizing that when preparing for the autumn recruitment interview, you must pay attention to the accumulation of brushing up on the questions. You must not be careless about this.

Interview situation

• Starts in the afternoon 15:00and lasts 45about minutes.

• Online video interview 1with an interviewer, a department leader; the interviewer was very kind.

• First, you are asked to introduce yourself, and then you ask questions for 1 minute, including 25algorithm questions .220

Ask a question

• What did you do during your summer internship? What specific work did you participate in? Was it mainly focused on development rather than algorithms?
• Are you currently exposed to NLP a lot? Which areas of deep learning are you mainly exposed to? What are the models that are used more frequently?
• What have you been exposed to in machine learning and what models have you used?
• Have you ever been exposed to a large-scale deep learning project? What exactly was done? What model was used in the process, RNN or DNN ?
• What did you do with your master's thesis and thesis, how was the progress, and when did you graduate?
• Please introduce in detail the content of deep learning in your graduate graduation project, what exactly it does, what the input data and output data are like, what is the background of the project, is it just regression analysis without classification analysis, use What is the loss function of ?
• Why is RNN not as effective as DNN in your graduate project ? Tell me what you think?
• Why do we need to predict remote sensing image data? What is its application value and how to verify it?
• Now that other satellite data are available, do you want to predict another one?
• Do you know what our department does? How do you understand the field of NLP ?
• Have you ever used convolutional neural networks and do you understand the theoretical principles?
• Have you ever done a classification task? What algorithms were used to do it, and what language was used to implement it?
• Let’s talk about the theoretical principles of random forest. How to judge the quality of the corresponding results of each node?
• Will dropout cause neurons to fail randomly? If the same data is substituted into the model after the model is determined, will Dropout still randomly cause neurons to fail, and will the results obtained change? Once the model structure of the neural network is determined, will Dropout not change?
• Let’s talk about the principle of BatchNorm and what does it do?
• What does overfitting mean? Why does overfitting occur? Let’s talk about some possible reasons why it occurs?
• If the data is concentrated in a certain range, will the neural network overfit?
• How to alleviate over-fitting? How does Dropout alleviate over-fitting?
• How does CNN neural network alleviate overfitting?
• Can overfitting be alleviated by adjusting the loss function, and can L1 and L2 regularization be alleviated?
• How do L1 and L2 regularization alleviate overfitting?

Algorithm questions

• C++ finds the longest continuous sequence.
• C++ bracket matching.

End of interview questions

• Is deep learning currently more mainstream mainly in word processing, such as NLP algorithms?

Feedback

• About a few days later, the official website showed that the process was over.

Welcome to follow: Crazy Learning GIS