Currently, I am four weeks into an REU (Research Experience for Undergraduates) program that lasts ten weeks. I am participating in an REU that is for computer science majors or anyone who has a strong interest in computer science. When I was applying for this, I had absolutely no idea what I was getting myself into.
Applying for an REU
I lucked out. My university sent me an email saying that an REU at Texas State University was accepting applications. But hey, if you are looking for different REUs, just click here. The day that the application was due, I decided to apply. I did not see any reason why I should not. Although, I would not recommend waiting until last minute because these applications generally require a reference, which I was not able obtain before the deadline.
Lesson Learned: Always apply! Even if you do not believe that you have a chance. Oh, and apply early and make sure you have a reference on hand.
A few weeks later, I had a Zoom call with the professor that is running the project that I am currently on. She did not ask any technical questions, just normal behavior interview questions. I remember her telling me that as long as I was comfortable diving into a subject that I had zero knowledge on, I would be fine. I told her that I was more than prepared to learn everything and I ended up getting the position!
The project that I am working on pertains to deep learning and machine learning. I can assure you, I did not even know that Tensorflow existed before I started working here.
Lesson Learned: Do not let impostor syndrome get the best of you. You can do a lot more than you think you can!
Getting Started
When I arrived to Texas State University, I felt way out of my comfort zone. Coming from the Northeast, Texas is a huge culture shock. I have never been away from home for longer than a month and a half and this is far from home.
The REU that I am participating in deals with smart & connected communities. I’ve attached a link to their website here. The project that I am working on deals with identifying buildings and structures on aerial imagery from drones utilizing a deep learning framework. This problem has been solved with computer vision, but we are exploring the idea of whether or not it can be solved with deep learning.
The thing about research is that you do not know if what you are researching is going to lead to anything or not. The first two weeks was actually just me learning the fundamental basics. I never used Linux or Bash before coming here. I like to think four weeks in, I am pretty decent with using terminal (I am actually starting to like using it!). Everything is baby steps. My mentor knew that I was going in with little to no experience and so she saw that I was putting in effort and learning things.
With deep learning, I decided to start at the most basic tutorial that I could find. That was the handwritten digit recognizer competition from Kaggle. This fundamentally helped me understand what I was jumping into. Think of this as the “Hello World” program for machine learning. For the most part, my first two weeks was reading and reading and reading a ton of research papers.
Lesson Learned: Baby steps! You cannot just dive right into research when you do not even understand the fundamentals. And, get ready to read everything and anything. I surpassed my article limit on Medium within the first four hours of work (bless incognito mode though)!
Diving In
Two weeks in, my mentor bought us Nvidia Jetson computers. In particular, the Jetson Nano and the Jetson Tx2. These single-board computers are used for deep learning. Think, a way more powerful Raspberry Pi computer.
When she gave us these computers, she asked me to download Matterport onto them and try to get a lightweight deep learning framework working on them. Honestly, I had absolutely no idea how to operate either computer and I this was truly me jumping right into the deep end.
The challenges I faced were extremely evident the moment I tried to install software onto the Nano. My partner and I could not figure out why particular software that should have been compatible would not install. After two days of work, we figured out that we were trying to install the wrong version.
Once we realized that implementing our framework is just absolutely not compatible with the Nano, I had to switch over to the Tx2. Looking back, I wasted around four days trying to figure out the Nano when we realized that we could not use it; however, I learned so much from that experience. And now I know how to use a Nano pretty well. We even got a security doorbell working on it using a Raspberry Pi camera!
Lesson Learned: Take the little victories. Seriously, debugging is tiring. Also, do not get attached to doing something one way. Make sure that you are open-minded. Do not get tunnel vision. Be ready for something that you have been working on for a while just completely flop.
Expectations and Experience
This REU is not over yet, but in the beginning, I expected to be way further into my project than I am right now. I also expected to write a paper that would get published, but like I said earlier, you never know if your research is going to work out. I do not know if I will walk out of this REU with a publishable paper. I actually seriously doubt that. You cannot do absolutely groundbreaking research in just ten weeks and somehow write a paper that is well thought out and executed.
I am excited for the next six weeks. We got the opportunity to visit tech companies and I met a lot of incredibly smart people that come from all over the United States. I would definitely do another REU. This is such a great experience for someone who does not know if they want to go into industry or research or if they have little direction in the computer science field.
I know that I have a great work-life balance here. I appreciate that so much. Generally, at my university, I am working 24/7. But here, we make time to do other things. For instance, go to Torchy’s Tacos, float down the San Marcos River, go to the gym every morning (I am really doing that now!), or start a book club (which I did). This experience has been fantastic so far. Research is stressful and tiring at times but it is also super rewarding. I know that at the very least, by the end of the ten weeks, I will have a vast amount of knowledge on a topic I knew nothing about coming in.
