Machine Learning

by Tom van Nuenen, Celebrating 10 years of innovation in data-intensive social science, D-Lab in collaboration with Grad Div is excited to introduce the LLM Working Group, an initiative focused on the exploration and discussion of Large Language Models (LLMs) like ChatGPT within academic research and teaching. This group aims to unite scholars, students, and data scientists to address crucial questions about AI's role in academia, including access, impact, creativity, and learning in the age of information automation. Through a series of interactive sessions, participants will gain insights into LLM capabilities, discuss ethical considerations, and explore innovative approaches to utilizing these tools in their work. Whether you're an AI veteran or a novice curious about the potentials of GenAI, the LLM Working Group offers a collaborative platform to learn, share, and shape the future of academic inquiry. Join us in navigating the world of LLMs together.

by Emma Turtelboom. What is the best Halloween candy? In this blog post, we will identify attributes of popular sweets and create a model to understand how these attributes influence the popularity of the sweet. We’ll discuss alternative model approaches and potential drawbacks, as well as caveats to interpreting the predictions of our model.

by Leïla Njee Bugha. Among its many uses, remote sensing can prove especially useful to document changes and trends from eras or settings, where traditional sources are either inexistent or infrequently collected. This is the case when one wants to study urban expansion in sub-Saharan countries over the past 20 years. To further remedy the lack of data on land cover uses from earlier time periods, classification methods can be used as well. Using easily accessible satellite imagery from Google Earth Engine, I provide here an example combining remote sensing with classification to detect changes in the land cover in Nigeria since 2000 due to urban expansion.