Major in Computer Science, undergraduate application for direct Ph.D., admitted in the fall of 2014, graduated in the summer of 2019.
Background of the application
Undergraduate:Computer Science and Technology, Tsinghua University, GPA 87 100;
In the summer of 2012, he was a visiting researcher at the Artificial Intelligence Laboratory of the Department of Computer Science at Stanford University for two months.
TOEFL:The overall score is 108, including 28 in reading, 29 in listening, 26 in speaking, and 25 in writing.
gre：v150，q168，aw4.0。
Admissions
University of Maryland, College Park, Computer Science, PhD[Accept].
Ph.D. from University of Illinois, Chicago, Computer Science
Carnegie Mellon, Language Technology Institute, Unawarded Master
Where to go after graduation
Received full-time offers from Google, Facebook, Amazon, Apple, and Alibaba upon graduation.
He joined Facebook in June 2019.
During my undergraduate years, because the Department of Computer Science of Tsinghua University has ample places for examination-free research and direct doctoral studies, I did not have much interest in going abroad.
However, in order to give myself plenty of choice, I have been preparing for the GRE and TOEFL exams since my sophomore year.
In the summer of 2012, I was fortunate enough to be selected to be a visiting scholar in the Department of Computer Science at Stanford University, under the supervision of Andrew Ng, a well-known professor in the field of machine learning.
During my two months at Stanford, I fully experienced the life, research and culture of the United States, and felt that studying abroad for a PhD would allow me to experience a different environment and broaden my horizons.
So after the Stanford exchange, I made up my mind to go abroad for a PhD.
Standardized Tests
I have been preparing for the test abroad since my sophomore year, but because I was not determined to go abroad at that time, the initial preparation process was not serious, and the first GRE and TOEFL test scores were not ideal: GRE V148, Q166, AW35, TOEFL 100.
Later, after I had the opportunity to go to Stanford, my will to go abroad began to become firmer, and I became more serious about memorizing words and doing questions. There was some improvement in all three scores of the second GRE.
My second TOEFL test was after I came back from Stanford.
I went to a foreign Chinese language school in middle and high school, so my daily English foundation is okay, plus I have been communicating in the United States for two monthsThe second TOEFL score went from 100 to 108.
[School selection].
However, while preparing for the application, I found that I knew very little about the entire application process.
The first step of choosing a school took a lot of effort.
Before applying, my knowledge of American schools was limited to the name of the school.
It was only after the application started that I learned about it one by one based on US News' ranking of computer science majors.
Since I was doing a research project in Natural Language Processing (NLP) at the time, I was also applying for NLP.
During the school selection process, I looked at the homepages of hundreds of professors from dozens of schools to learn about their research directions, skimmed their latest publications, and even saw if there were any seniors among their current students.
In the end, I chose 14 schools as my target students.
Among them, the top 10 schools in the United States are used as impact schools, such as CMU, Stanford, UIUC, etc.
I also know that the competition to apply to CMU and Stanford will be fierceSo on the CMU side, I applied directly for a master's degree, and on the Stanford side, I also checked the option to accept a master's degree when I applied for a Ph.D. program.
The schools ranked 11 to 30 in the United States are the schools I am more sure of, including the University of Maryland (UMD), Harvard University, and the University of Pennsylvania (Penn).
Schools outside the top 30 are my guaranteed schools, such as the University of Chicago and the University of Illinois at Chicago (UIC).
When choosing a guaranteed school, I made it an important criterion to have the right professors because of the school's low ranking.
[PS and letters of recommendation].
Once you've chosen a school, the most important step is to prepare your PS and letters of recommendation.
At the time, I had no idea how to write PS. The "PS Sample" on the Internet made me feel that writing PS was like writing about my own life journey, so I wrote a lot of my own stories, such as when I first came into contact with computers, why I liked computers, and so on.
Later, I looked back and found that the PS written in this way had no reference value for the admissions professors.
As a PhD applicant, the most important experience is research experience.
Brilliant scientific research experience is the stepping stone to the offer.
So I rewrote the whole PS, focusing on the scientific research projects I participated in, the problems encountered in scientific research and the process of solving them, reflecting my scientific research ability and potential.
In terms of recommendation letters, I found two teachers who had taken me to do scientific research projects and a course teacher to write me recommendation letters to introduce me from both scientific research and curriculum.
[Set magnet].
When the application materials were almost ready, I sent a set of magnetic letters to one of the professors I was most interested in at each school I applied to.
Although most of the magnetic letters were lost in the sea or only received a template response, there were some professors who expressed great interest in my application, including my later doctoral supervisors.
[Offer Selection].
After submitting all the application materials, I started to wait for feedback from the schools.
In January 2014, UIC sent me a Ph.D. offer and named the most famous professor in their computer science department as my supervisor.
But a few days later, the professor asked me to decide right away whether to accept the offer because he was going to apply for a fellowship for me, and if I accepted the offer and then regretted it, it would invalidate the fellowship.
In the absence of notification from other schools, on the one hand, I was unwilling to go to this school with a ranking of 50+, and on the other hand, I was worried that I would not receive offers from other schools in the future.
I bought the professor a day to think about it, and then emailed the other professors to ask about the status of the application.
Luckily, the professor at UMD had an impression of me and immediately arranged an interview for me as soon as he received my email.
After the interview, he told me that the probability of admission was very high.
This put me at ease at once, so I turned down UIC's offer.
Luckily, I also received a master's offer from CMU LTI before the official offer from UMD arrived.
In the final choice, although CMU LTI is a NLP palace-level scientific research unit, considering that there is still uncertainty about transferring from master to doctoral degree after admissionCoupled with the fact that UMD's professors and my research direction were more compatible, I finally chose to accept UMD's PhD offer.
It also brought my application process to a successful conclusion.
I applied for a PhD in Natural Language Processing (NLP).
NLP is a subfield under the currently popular field of artificial intelligence, which mainly studies the language that people use every day.
The reason for the emphasis on "natural language" here is to distinguish it from "programming language" in the computer field.
There are many research directions in NLP, such as part-of-speech tagging, syntactic analysis, semantic analysis, question answering system, sentiment analysis, topic modeling, etc.
Traditional NLP methods are mostly based on statistics and probability, which requires researchers to have a solid mathematical foundation, especially knowledge of probability and statistics.
In recent years, with the rise of deep learning, NLP has gradually become the world of deep learning.
Various deep learning technologies designed for NLP are emerging, such as recurrent neural networks (RNNs), word embeddings, and so on, which have greatly refreshed the benchmark for various tasks.
With the increase in computing power, pre-trained models are beginning to emerge.
Google came up with the famous pre-trained model BERT, which relied on a model to sweep all major benchmarks.
As the greatest invention of mankind, natural language can express infinite meanings through various combinations, although the vocabulary of a language is limited.
This makes it difficult for machines to understand natural language, but with the emergence of new algorithms and the improvement of computing power, there is still a lot of work that can be done in NLP, and many researchers need to continue to cooperate and work hard to push this field forward.
The following experience comes from my own experience of applying and how I felt about being involved in admissions on the department's admissions committee during my PhD.
Looking for a study abroad agency or DIY?
Generally speaking, most students may only have one experience of applying to study abroad, but this application is enough to change the direction of their lives in the future.
At the same time, the application process is extremely complicated, and if you are DIY, you have to step on countless pitfalls in person, and when you look back, it is full of blood and tears, so finding a study abroad service agency is a good option.
Excellent study abroad service agencies, such as admitwrite, have a lot of experience in helping students apply, know how to help students avoid the pitfalls in the application, how to maximize the potential of students, so as to get the chance of admission.
For students with strong ability, you can also choose a semi-DIY way, and the teacher of Admitwrite will guide you to complete the entire application process, which can not only participate in the application process in person, but also reduce detours under the guidance of teachers, seniors, and seniors, and improve the application efficiency and success rate.
Master's or PhD?
The master's and doctoral programs in the United States are very different.
A master's degree can be graduated with enough credits, while a doctorate requires achievements in scientific research in addition to the course.
If you don't have much interest in scientific research, then you can apply for a master's degree.
However, if you want to participate in scientific research, then it is recommended to apply for a PhD.
A PhD is essential if you want to work in research-related fields after graduation, such as working in a large research institute or university.
In addition, the scholarship conditions for master's and doctoral degrees are also very different.
PhD scholarships in science and engineering are generally guaranteed, and most of them come from the income of teaching assistants (funded by faculties) or research assistantships (funded by supervisors' research projects).
A small number of outstanding doctoral students can apply for fellowship.
Most of the master's students have to pay their own tuition, and only a few cases will be funded, such as participating in scientific research projects or the department has more teaching assistant positions.
GPAGPA is an important metric that directly reflects an applicant's learning at the undergraduate level.
Although generally speaking, the higher the GPA, the betterBut when it comes to applying for a PhD, GPA is not the most important criterion.
Since research experience is the most important thing when applying for a PhDSo GPA is just a secondary standard.
As long as the GPA score and ranking are passable, it generally won't have much impact on the application (except when applying to top schools).
When applying for a master's degree, since the master's program generally does not require scientific research experience, the importance of the GPA will be higher than that of the doctorate.
Scientific research and**.
Research experience is essential when applying for a PhD program.
Research experience can reflect the applicant's attitude towards problems and problem-solving skills.
Based on the applicant's research experience, the admissions professor will determine whether the applicant has the potential to study for a PhD.
Impressive research experience will bring great benefits to doctoral applications.
In general, research experience can be obtained through participation in laboratory research projects and off-campus research internships.
As for the ** requirement, most doctoral programs generally do not require applicants for undergraduate direct Ph.D., but if they have been published in a first-class journal or conference during the undergraduate stage, the success rate of applying for a PhD will be greatly increased.
I would like to mention here that the field of artificial intelligence in which I work has risen rapidly in recent years, and conferences in all directions are creating new highs in the number of submissions and acceptances every year.
There has also been a significant increase in the number of applicants who have presented at top-level conferences at the undergraduate level.
Therefore, if doctoral applicants in the field of artificial intelligence want to get offers from top universities, they should try their best to publish before applying**.
When applying for a master's program, there are not many requirements for the applicant's scientific research experience since the graduation requirements do not include ** requirements.
However, a good research experience will still greatly increase the probability of admission.
[PS and letters of recommendation].
P.S. and letters of recommendation are actually introductions to oneself and others to oneself.
For doctoral applicants, what the professors want to see most is that the applicant is interested, capable, and has good scientific research quality.
In PS, you can write about your own scientific research experience or other related project experience (such as coursework), describe the problems you encounter and the process of solving them, so as to reflect your ability and quality.
Letters of recommendation can also be written along these lines, but from the recommender's point of view.
You can also mention the professors you are interested in in ps.
In my experience with admissions committees, if an applicant with a good background mentions one or more professors in the PS, the committee member responsible for the initial review will forward the application materials to the mentioned professor for further review.
Due to the sheer volume of applications, it is impossible for the admissions committee to read all the PS and letters of recommendation verbatim during the initial review.
In order to impress the members of the admissions committee in a short period of timePS and letter of recommendation are very important.
Focused, grammatically correct, and fluent PS and letters of recommendation can quickly allow the admissions committee to catch the highlights of applicants, so the language and content of PS and letters of recommendation must be revised and polished repeatedly.
In this regard, a study abroad program like admitwrite can provide applicants with high-quality essay editing services, making the language more authentic and in line with English language habits, and highlighting the applicant's highlights.
Interview. When you get an interview invitation, congratulations that you're one step away from an offer.
Every year, each school's department receives a large number of applications, so it is impossible for the admissions professor to interview all the applicants one by one.
From the professor's point of view, he will select some applicants who are close to his research direction and have a more prominent background for interview.
During the interview, the professor will usually ask the applicant to talk about the research projects he has done and ask some detailed questions, mainly to see if the applicant has actually participated and done the work done in the PS and recommendation letters.
If the applicant's application materials are true, then the applicant generally does not need to prepare much for the interview in this regard.
In addition, the professor may temporarily come up with some questions in related fields to examine the applicant, which requires the applicant to have a relatively solid basic skills, and it is best to understand the professor's field before the interview, and have a simple understanding of this field.
[Set magnet].The main function of the magnetic letter is to greet the interested professor in advance and let the professor pay attention to his application.
However, every professor receives a lot of emails about this every day, so don't expect every professor to open every set of magnetic letters.
However, as an applicant, you should still send a set of magnetic letters to the professors you are interested in.
When writing a magnetic letter, try not to use a template other than polite statements.
Template sets of magnetic messages generally do not receive a reply or will only receive a template reply.
In the set of magnetic letters, you can briefly introduce your background and scientific research experience, and list some of the professor's recent work, explain that you and the professor's scientific research direction is close, and express your willingness to cooperate with the professor.
For example, when I sent a set of magnetic letters to more than a dozen professors when applying, only three or four of them replied, most of them were template responses, but one professor was very happy with my application and replied directly "I am very excited about your application".
It also paved the way for my later application.
The application documents required for each program are similar and limited.
When the applicant has more content to show (such as **, demo, etc.), the applicant can create a personal homepage, hang these contents on the homepage, and then write the homepage address in the application materials.
A common and free way is to host the homepage on GitHub, which should be familiar to students with programming experience.
Resume. Most of the computer science majors know about LaTeX.
This is a very powerful typesetting software. The resume compiled and typeset with LateX will be very beautiful, so it is recommended to try it.
It is especially recommended for students majoring in computer science, which can reflect their professionalism.
In addition, latex is also a common tool for typesetting, and mastering it in advance also provides convenience for future writing.