Do you want to get your research published in an academic journal but don’t know where to begin? Horizon alumnus Serena Pei, a high school junior from California who participated in our Spring 2021 Applications of Machine Learning section, recently shared her story of how she got published in the Journal of High School Science having a limited background in machine learning (ML) prior to the program. Serena’s research used deep learning models to classify emotions using a dataset of facial expressions from Kaggle’s 2013 emotion recognition competition. We congratulate her on her journey. Her experience shows that, while publication isn’t the best or only measure of a quality research project, publication is a possibility for Horizon students who work hard, thoughtfully select a topic, write an excellent paper, and select the right target journal. 

With the guidance of her mentor, PhD candidate Angelina Wang from Princeton, her work analyzed various confusion matrices, studied data augmentation methods, and applied a multiplier technique to the images of diverse facial expressions in the dataset for increased accuracy. As emotion recognition mechanisms are regularly harnessed to better grasp human interactions and social behavior, Serena’s project contributed to the field by increasing the precision with which machines detect one set of feelings from another via an image. Her project accounted for minute nuances of facial expressions which filled in some of the gaps in the current data. Reflecting back, Serena highlighted that her Horizon experience “was really hands-on, I could code my own code and then use that for my paper and that was really cool!”

Serena pointed to the onset of the program as the most exacting — “I didn’t know a single thing about ML and all the different terms [...] Angelina had to give me a course online. It was called Fast.AI. It was a full-on course with a lot of chapters, and I had to go through all of them. They were really long too.” Serena did not shy away from asking questions to solidify her understanding of ML. In fact, she intentionally used her 1-on-1 time with Angelina to create learning opportunities for herself by drawing on Angelina’s experiences. Serena emphasized, “I had to ask [Angelina] for explanations and she was really good at answering my questions and you could tell she was very professional. She was a really big help for this project.”

All too often do students underestimate their own merit and that of their research manuscripts — finding it hard to believe someone of their credentials could get published. So akin to the majority of high school students, Serena had her hesitations about publication. Publication came second to conducting the research itself: writing a 20 page research manuscript is a challenge in and of itself. Only after she finished her research paper did Serena consider “‘so maybe I should try to get it published.’ So I went and did that.” She admits that her mentor was a driving force behind her efforts to get published; stressing that “getting an A from [Angelina] kind of gave me confidence. She had pretty good feedback for my paper so that’s what motivated me to submit it to journals to get published.”

Once she was convinced that publication was within arm’s reach, she “really wanted to be published so I really tried hard submitting those papers.” Serena’s “go-get-’em” attitude coupled with her persistence and patience navigating the publishing world ultimately proved worthwhile. When we inquired further in an effort to get a student’s insights on the publication process, Serena first recalled the anxiety during the waiting period. She found it nerve-wracking wondering if journals would actually read her paper “but after they go back to me about the edits I’m supposed to make, it was easier for me to communicate with them.” Serena continued to note that the “hardest part was that each journal requires different formatting. There were a lot of formatting issues like for the bibliography and works cited. Also, just for the entire research paper, they have different fonts they want.” Serena’s advice to her fellow peers, and her greatest takeaway from the publication experience at large, was that “you have to cater for [each] journal and what they want as a requirement.” 

While publication and similar accolades undoubtedly carry benefits in resume-building for university admissions, there are also broader gains that are priceless. In Serena’s case, she discovered her own academic potential, surprised by just “how much I could do and write in my paper. I think I’ve grown a lot since then.” She felt a novel determination and capacity within herself to get things done when she’s passionate — making her more confident in her choice of college major and her career aspirations in machine learning. And she improved as a writer and computational thinker, benefits which will serve her in university and her future career.