Thursday, February 13
2:30 — 5:00 p.m.
Amy Gutmann Hall | 3333 Chestnut Street, Philadelphia, PA 19104
Join us at Amy Gutmann Hall for two engaging workshops as part of the Women in Data Science (WiDS) @ Penn Conference.
Whether you’re passionate about AI, fascinated by cutting-edge vision-language models, or simply eager to expand your data science skills, these workshops are the perfect start to your WiDS experience!
Please note that tours and workshops are available on a first-come, first-served basis.
Guided Tour: 2:30 – 3:00 p.m.
Completed in 2024, Amy Gutmann Hall serves as a hub for cross-disciplinary collaborations that harness research and data across Penn’s 12 schools and numerous academic centers. Join us for a guided tour to explore how Amy Gutmann Hall is shaping the future of education and discovery.
Choose Your Workshop at Registration
Students can select their preferred workshop upon registering. Please note that workshops and tours are available on a first-come, first-served basis, so plan to arrive early to secure your spot!
We look forward to seeing you at Amy Gutmann Hall for an inspiring and educational start to WiDS @ Penn!
Workshop Bridging Language and Vision: 3:00 – 4:30 p.m.
Workshop From Data to Discovery: 3:00 – 5:00 p.m.
Bridging Language and Vision: A Hands-On Introduction to Vision-Language Models
Bridging Language and Vision: A Hands-On Introduction to Vision-Language Models
3:00 – 4:30 p.m.
Amy Gutmann Hall, Room 414
Led by Artemis Panagopoulou, SEAS’20, SAS’20, doctoral student in Computer and Information Science (CIS) within the School of Engineering and Applied Science (Penn Engineering)
Explore the fascinating world of vision-language models in this interactive workshop! We’ll dive into the evolution of these models, from earlier Convolutional Neural Network architectures (e.g., AlexNet) to more recent transformer-based models. We will explore both contrastive approaches like CLIP and generative systems such as LLaVa. You’ll gain a clear understanding of how these models work, their role in the current AI landscape, and the challenges and ethical implications they present. The workshop culminates in an interactive Google Colab demo, where you’ll learn how to apply these models in practice. Whether you are new to vision-language integration or looking to deepen your expertise, this session will provide valuable insights and practical skills.
Prerequisites:
– Understanding: Python coding experience. Familiarity with deep learning concepts.
– Preparation: A laptop is highly recommended for hands-on participation during the demo.
From Data to Discovery: Exploring AI with a Patent Case Study, ChatGPT, and Generative Models
From Data to Discovery: Exploring AI with a Patent Case Study, ChatGPT, and Generative Models
3:00 – 5:00 p.m.
Amy Gutmann Hall, Room 306 (Hub Lab)
Led by Linda Zhao, Professor of Statistics and Data Science, and Xinyu Xie, Ph.D. Student, both from the University of Pennsylvania. The session will be moderated by Lynn Wu, Associate Professor of Operations, Information and Decisions at the Wharton School.
In today’s data-driven landscape, vast unstructured data sources—such as documents, texts, and electronic health records (EHRs)—demand advanced AI tools to unlock their full potential. Generative AI, powered by Large Language Models (LLMs), is becoming indispensable for processing and extracting insights from complex language-based data. Applications like patent evaluation or transforming healthcare through EHR analysis are just a few examples of how AI is reshaping industries.
This interactive workshop, centered on a patent approval case study, will introduce students to the full pipeline of solving real-world problems using AI and data science. Participants will explore:
- How to identify key problems that lend themselves to AI-driven solutions.
- Collecting and preparing data, building models, and running state-of-the-art algorithms.
- Validating models and interpreting their results.
While the workshop will include live demonstrations using RMarkdown, students will have the opportunity to interact with tools like ChatGPT and HuggingFace, a platform hosting thousands of pre-trained Transformer models. Due to package limitations, certain code segments will be demonstrated, but key concepts and steps will be clearly explained.
The workshop will provide a deep dive into how LLMs evolved from basic concepts such as regression (understanding how weights and coefficients are estimated) to Neural Networks (NN), and finally to powerful Transformer-based models like those found on HuggingFace. By the end of the session, students will have a solid understanding of how modern AI technologies—like ChatGPT and Transformers—help tackle large, unstructured datasets for applications such as patent approvals, EHR analysis, and beyond.
*All attendees should bring a laptop or device.