Last month the CDS Leadership Circle, our student representative group that plans events and activities for students, hosted the first ever CDS Academic Showcase on May 7. The event featured several research projects presented by CDS students. The showcase took place digitally via a tool called Topia.io, which provides a customized virtual world and included booths and live presentations that allowed attendees to enjoy live interactions with each presenter. Projects were voted on at the time of the event.
Let’s meet the two showcase winners.
In CDS undergraduate student Daniyal Khalid Chawla’s presentation, “Real-Time Data Visualization for SpaceSuit Design”, he summarized his internship experience at Final Frontier Design, a dynamic small business of artists and engineers that creates garments for space (particularly for NASA). During his internship, Daniyal designed a program that:
- visualizes the data coming out of the (many) sensors inside space suits in real time
- sets off alarms if anything becomes unsafe for the astronaut such as if the oxygen level falls or if the suit becomes overheated, etc.
- tries to predict if a certain parameter may be leaving the safe zone (to prevent any potential dangers for the astronaut before they even occur).
“This experience taught me how important it is for a data scientist to adapt. Traditional data science (where data is analyzed retrospectively) wouldn’t be very helpful in my case. Everything needed to happen in real time. Therefore, this project inspired me to re-learn my data science concepts, and come up with new ways to visualize, analyze, and predict any potential dangers to the astronauts wearing my suit,” says Daniyal.
In CDS MS student Johnny Ma’s project, “Comedian Identity in Stand-up Comedy: Text and Audio Analysis of Netflix Routines”, he collected 100 stand-up comedy routine audio and subtitles from Netflix. By creating textual and audio features, he built a set-up line vs. punchline prediction model with 60% accuracy, slightly improving on the literature. Finally, he wanted to investigate whether a female comedian telling a joke that used female gender humor would resonate more with audiences compared to a male comedian telling the same female gendered joke. He used a pre-trained model to tag gendered jokes, and found that audiences laughed 35% longer when comedians told jokes matching their own gender identity.
“I really loved going to the comedy clubs in NYC (pre-COVID), seeing great comedians such as Ali Wong, Dave Chapelle, and Hannah Gadsby. Since I was reading a lot on digital humanities and cultural analytics, I wanted to apply the NLP tools I used at my job to analyze an art form that combines creative writing and stage performance…While punchline prediction is a typical NLP task, I wanted to look into how audiences react to comedian and joke identity parity… I learned from the book “The Comedians: Drunks, Thieves, Scoundrels, and the History of American Comedy” that comedy has a long history of excluding certain groups… and wondered about how identity works in comedy in contemporary times… I was able to establish that identity determines both who can tell a joke, and how hard audiences laugh at a joke,” says Johnny.
For more information about the CDS Leadership Circle and other student groups at CDS, please visit our Student Groups webpage.
By Ashley C. McDonald