Library AI Conversations

Library AI Conversations

Teenie Harris Archive, August 21, 2018

Dominique Luster (Teenie Harris Archivist at the Carnegie Museum of Art),  Caroline Record (Creative Technologist, Carnegie Museums of Pittsburg), and Golan Levin (Associate Professor of Art at CMU and Director, Frank-Ratchye STUDIO for Creative Inquiry)  share their experience applying machine learning and crowdsourcing techniques to an image collection. The original project was funded by an NEH grant . Following that work they organized what they called  "Teenie Week of Play," to explore auto shortening titles, cleaning up existing subject headings, extracting names and and locations from descriptions using Named Entity Recognition, and verifying face recognition data using Amazon Mechanical Turk.

Londa Schiebinger, James Zou, Dan McFarland, Bas Hofstra, June 5, 2018

Our guests, Bas Hofstra (Postdoctoral scholar in Education) and Dan McFarland (Education, Sociology, Business), Londa Schiebinger (History) and James Zou (Biomedical Data Science, Computer Science, Electrical Engineering) will meet with us to discuss a project to analyze the effects of diversity on scholarship. The study will require building a corpus of scholarship across disciplines, so we will also discuss how the library might contribute to the project.

We will also learn about a project funded by the Human-centered AI seed grant program at Stanford that James Zou and Londa Schiebinger recently completed with Nikhil Garg and Dan Jurafasky. It takes a historical look gender and ethic stereotypes in word embeddings (See 

Ashok Popat, April 16, 2018

Ashok Popat is a research scientist at Google (previously at Xerox PARC) leading the Google Cloud Vision team. Popat and a colleague will talk with us about how we might use the Google Cloud Vision API for image analysis of Library collections. The Google Cloud Vision API helps generate metadata and can perform sentiment analysis. It enables image classification, detection of objects and faces, and reading printed words images.   

Harini Kannan, April 12, 2018

A resident at Google Brain working on Ian Goodfellow's team, Harini shared her work on adversarial examples for neural networks. Though her research does not overlap with any projects we are pursuing in the library, we learned from Harini how AI research teams are set up at Google Brain, how projects are organized and what makes them so successful.

Michael Bernstein and Ranjay Krishna, February 12, 2018

We spoke about the Visual Genome project and how to use computer vision and crowd-sourcing to make the most of digitized library image content. Michael Bernstein is an Assistant Professor of Computer Science at Stanford University, where he is a member of the Human-Computer Interaction group. His research focuses on the design of crowdsourcing and social computing systems. Ranjay Krishna is Stanford PhD Candidate/Researcher in the Artificial Intelligence Laboratory, co-advised by ​Prof. Fei-Fei Li and ​Prof. Michael Bernstein. His research interests lie at the intersection of artificial intelligence, computer vision, machine learning and human computer interaction. Krishna teaches a class on computer vision at Stanford in the Fall quarter, exposing students to applications such as self driving cars, object detection and tracking, etc. You can find details about that course and course materials here

Read the Visual Genome paper here

Bryan Catanzaro, January 24, 2018

Bryan Catanzaro, runs a research lab at NVIDIA where he is Vice President of Applied Deep Learning Research. We spoke with Bryan about how to configure a lab and how to determine which projects are worth pursuing. We talked through some of the ideas for applications of AI that staff across the Library have shared in the "Your AI Ideas for the Library" document.

Peter Norvig, November 27, 2017

Peter Norvig, Director of Research at Google, was our first "Conversations" guest. Peter talked with us about how software development is changing and the changing role of the library when so much of our search activity happens in Google search, Google scholar and Google books. We talked about bias, reliability, and whether we can know why an AI arrived at a particular result (with reference to this article).