←
Princeton Computer Science1,846 posts
Princeton Computer Science avatar

Princeton Computer Science

@PrincetonCS

The Department of Computer Science at Princeton University

◷ Joined August 2012▧ 559 Photos and videos
Princeton Computer Science avatar
Princeton Computer Science@PrincetonCS·Nov 17

“I find it fascinating that ML models can be incredibly good at making predictions but still struggle when something unexpected happens,” she said.

Princeton Computer Science avatar
Princeton Computer Science@PrincetonCS·Nov 17

Yu’s work looked at ways to adjust the model as things change, without having to train the whole model again. “Imagine a model trying to predict how much electricity a city will use tomorrow,” Yu said, “but a sudden heat wave throws off its predictions.”

Princeton Computer Science avatar
Princeton Computer Science@PrincetonCS·Nov 17

The research explored ways to help time-series machine learning models make better predictions. An energy company might use a time-series ML model to forecast future demand, for example. Other uses include climate forecasting and fraud detection.

Princeton Computer Science avatar
Princeton Computer Science@PrincetonCS·Nov 17

Shirley Yu, a senior majoring in computer science, won a department award for her independent work project, "Spectral-Gated Test-Time Adaptation for Foundation Time-Series Models under Regime Shifts." She was advised by @HazanPrinceton.

Post attachment
Princeton Computer Science avatar
Princeton Computer Science@PrincetonCS·Nov 16

The research shows that advisers used information unavailable to the algorithm in two out of three actions taken during student meetings. While the AI tool may have prompted an adviser to check in, Liu said, advising a student usually requires human expertise.

Princeton Computer Science avatar
Princeton Computer Science@PrincetonCS·Nov 16

Colleges use AI tools for academic advising, analyzing data to predict which students need additional help. But new research from Lydia T. Liu suggests that helping students requires judgment and expertise that go beyond the algorithm. bit.ly/4iLlKyr

Post attachment
Princeton Computer Science avatar
Princeton Computer Science@PrincetonCS·Nov 12

The work is part of the 3TMAAP project, a partnership led by Prof. Emmanuel Kreike, an expert in African history. The project that uses AI to label archival photos from 7 countries. The goal is to better understand environmental change that happened before satellite imagery.

Princeton Computer Science avatar
Princeton Computer Science@PrincetonCS·Nov 12

“The goal is to see if a machine learning model can reliably find households and trees across large areas and multiple decades,” Yeh said. “This allows us to record how communities and landscapes have changed over time.”

Princeton Computer Science avatar
Princeton Computer Science@PrincetonCS·Nov 12

Annika Yeh, a senior majoring in COS, won a dept award for her independent work project. Her research looks at whether deep learning models can label trees and vegetation on archival aerial photographs of Namibia. 📸 Photo from Yeh's project, with trees marked as pink dots.

Post attachment
Princeton Computer Science avatar
Princeton Computer Science@PrincetonCS·Nov 11

Congrats to Nicolaas Kaashoek, PhD student in CS, on being awarded a Wallace Memorial Fellowship from @PrincetonGrad! 🎉 The fellowship is the highest award conferred to graduate students at @EPrinceton. Kaashoek's research focuses on distributed systems and cloud storage.

Post attachment
Manoel avatar
⟳ Princeton Computer Science retweeted
Manoel@manoelribeiro·Nov 10

What does ChatGPT tell folks about the midterm elections? It depends on who's asking! We audited ChatGPT inside 160 real users' own accounts in the days before the midterms. With memory on, we find that answers shifted toward each user's own party! elections.citp.princeton.edu…

Post attachment
Xindi Wu avatar
⟳ Princeton Computer Science retweeted
Xindi Wu@cindy_x_wu·Nov 5

Excited and honored to be selected as a 2026 Rising Star in Data Science! datascience.uchicago.edu/res… My research focuses on building a better data foundation for more capable multimodal systems, particularly world models that can interact with the physical world. Looking forward to meeting and learning from an amazing group of researchers and peers at @UChicago next week!

Post attachment
Princeton Computer Science avatar
Princeton Computer Science@PrincetonCS·Nov 4

The competition challenges participants to train control policies on simulated buildings and then evaluate them on an unseen set of buildings that vary in size and materials. The goal is to create HVAC control policies that can adapt to new buildings with minimal changes.