Claire Dennis of the Princeton School of Public and International Studies is deep in math and computer code this spring. Algorithms and computers She plans to enter the world of policy rather than programming, but it is important to understand how technology is changing the way we process knowledge. She felt there was.
“We are witnessing an explosion of technology and the impact on policy is huge,” said Denise, who is preparing to complete her master’s degree in public affairs in May. rice field. “I have heard many times that there is a big divide between policy makers and engineers, and very few people speak both languages.”
Dennis, who plans to pursue a career in technology policy, is filling a gap in his knowledge through a new graduate course, Machine Learning: A Practical Introduction for Humanists and Social Scientists. The course, taught by her Sarah-Jane Leslie, the philosophy professor of the 1943 class, provides an introduction to “deep learning” for graduate students.
This class assumes that students do not have extensive knowledge of calculus, linear algebra, or coding experience. By the end of the semester, students were able to code various models themselves, including language and image recognition models, giving them a better understanding of the use of machine learning, especially in the humanities and social sciences. Her last two weeks of the course focused on understanding how complex language models such as ChatGPT work.
“This course isn’t the best opportunity for me to become a programmer, but it is an opportunity to familiarize myself with models, the challenges of these models, their common tensions and trade-offs, and to be the best mediator possible.” It’s a great opportunity, it’s time for me to graduate,” said Denise. “It’s becoming more and more relevant every day.”
Portal to Brand New Scholarships
Psychology graduate student Rachel Metzgar (left) discusses ways to apply machine learning to her research with Sarah-Jane Leslie (right), professor of the 1943 Class of Philosophy.
Leslie, who served as dean of the graduate school from 2018 to 2021, conceived the course shortly after stepping down from the role when he wondered how to use machine learning in his research. This class will be offered again next spring.
“This is an incredibly exciting frontier that opens up new research possibilities across disciplines, and it’s doing it incredibly fast,” Leslie said. Leslie was particularly inspired by the work of Marina Rastow, Kedri A. Zirka Professor of Jewish Civilizations in the Near East and Professor of Near Eastern Studies and History. WHO Using deep learning and computer vision A treasure trove of ancient documents, Princeton Jenniza Project.
“Even in fields that seem far removed from machine learning, these techniques can be used to do unprecedented research,” says Leslie.
Leslie probably Ten students attended the course, with a maximum of 25. Instead, she enrolled her 35 students from all four of her departments at the university. Taking this class are PhD students in Comparative Literature, Near Eastern Studies, History, Politics, Psychology, Neuroscience, Civil and Environmental Engineering, Mechanical and Aerospace Engineering.
Four students, including Dennis, are graduate students at the School of Public and International Affairs.
“Normally, especially at the graduate level, there are pretty hefty prerequisites for taking a machine learning class, so I challenged myself to teach this course without requiring a coding background or college-level mathematics. It was a challenge,” said Leslie. “I see it as a zero-to-hero course.”
crack the code
Students from different disciplines work together on their machine learning homework during their break. Foreground: Nancy Tan (right), Political Science graduate student, Jamie Chiu (left), Psychology. Background: João Carvalho (right) and Teddy Becker-Jacob (left), graduate students in philosophy.
Given that Leslie planned to teach students with no machine learning experience, she needed to develop some method by which they could gain a quick and substantive understanding of the technology and its uses. there was.
She started classes with a crash course in Python, a programming language used in machine learning. In many of her early sessions, she projected her computer screen into the class, demonstrating how she wrote code, and narrated every step while composing each line. She sometimes highlights potential traps, bugs, and errors in her messages and how to respond to them, allowing students to follow her thought process.
In addition to teaching students how to code, Leslie has also led in-depth discussions on the core mathematical operations that enable machine learning, and the core concepts of computer vision and state-of-the-art natural language processing. I was. Throughout, she aimed to demystify artificial intelligence such as ChatGPT by helping students understand how such models actually learn and behave.
Dennis said the course was engaging and challenging in a way that was different from her policy class, but not so difficult that she felt she was struggling. I’m a little surprised at how much I’ve actually caught up with when I didn’t have one,” she said.
Chelsea Clark, a comparative literature graduate student, said she was intrigued by the concept of the course after hearing Leslie speak at the Center for Digital Humanities (CDH) last fall.
“At the Center for Digital Humanities, there has been a great deal of discussion about large-scale language models,” said Clark. “That was probably my biggest motivation: I want to be fluent in these models.”
Clark also ponders his career prospects after completing his Ph.D. “I’m looking at a tech career and tenure her non-track career,” she said. “I asked her research software engineer at CDH if she would hire someone who was self-taught, or who had taken a course. knows where the learning gaps are.”
Clark said he will at least graduate from class with an understanding of machine learning literacy and its applications and uses.
One of Leslie’s hopes is that the course will help her students become more fluent collaborators.
“My own research career has involved a great deal of interdisciplinary collaboration. In my experience, it is easy to have the illusion that two people who know nothing about each other’s research can work together productively. is,” she said. People who are somewhat fluent in other people’s fields of expertise. ”
“I hope that students see this course as a starting point rather than an ending point,” she added. “I told them on day one that what I wanted more than anything else was to go out and learn more about machine learning and feel like they could navigate this burgeoning space. rice field.”
AI through a philosopher’s lens
Acquiring large numbers of students through the mathematics and theory-oriented courses required to develop expertise in machine learning is challenging, especially as artificial intelligence evolves rapidly, says Gordon YS Wu School of Engineering. Professor Peter Ramudge said. , Professor of Electrical and Computer Engineering and Director of the Center for Statistics and Machine Learning.
“I think what Sarah-Jane is doing is bringing machine learning to people who live outside of the machine learning pond. Look, roll up your sleeves, here’s a hands-on exercise and see how it all works,” Ramaj said. “And along the way she can learn how to program in Python, which I think is good for everyone these days. I hope you will think, “I want to.”
Philosophy professor and chair of the department Benjamin Morrison said when Leslie first pitched the idea for the course, he was excited not only for its practical benefits, but also as a way to broaden the skill sets of graduate students. to future employers.
Given the long tradition of philosophers with expertise in adjacent fields such as linguistics, logic and mathematics, it makes sense that such courses would be taught by a philosophy faculty, he said. I was. “Philosophers use philosophy as a lens to think about their expertise. Philosophy can take knowledge from other areas and think seriously about what the nature of that knowledge is.” I’ll do it,” Morrison said.
“I think it was really inspiring for Sarah-Jane to know that if our students are going to solve AI problems, they really need to understand these technologies,” says Morison. added Mr. “And, of course, that goes for other humanist scholars and beyond who attend her seminars. If they are all going to solve these problems, they have to know the basics.” it won’t.”
