Human-AI Teamwork for Image Analysis

by Dr. Travis Mandel

Getting AI to work well requires humans to tediously label numerous examples before the AI begins learning. How can we design a more interactive process, where humans and AI systems learn and work together in real-time? We are applying these techniques to real-world problems affecting Hawaii Island such as identifying outbreaks of invasive plant species.

Object Tracking in low-data environments

by Dr. Travis Mandel

Tracking animals and objects as they move through space is something that is typically easy for a human and hard for a computer. State-of-the-art tracking methods rely on large amounts of video data, but what about situations where we don't have nearly as much data, such as tracking a particular species of fish off Hilo Bay? We are developing tracking systems that can cope with these challenging limited-data scenarios.

AI-Assisted Scientific Data Collection

by Dr. Travis Mandel

Collecting data in fields such as psychology and behavioral economics is expensive, which limits the progress we can make in these fields. How can we use AI systems to help determine where each new datapoint should be collected for maximum effectiveness? We are exploring methods that interact with scientists and others to determine how to collect data in a way that best fulfills scientists' objectives.

Using the Crowd to Prevent Harmful AI Behavior

by Dr. Travis Mandel

AI has the potential to cause a great deal of harm - but who decides what is harmful? Dr. Mandel and his team presented the first study of how best to involve non-AI experts in the process of preventing real-world AI harm. In addition to testing on real-world scenarios, Mandel et al. developed CarefulCar, a novel video game testbed for AI safety.

Hawaii's Coral Reefs in VR

by Francis Cristobal

Imagine diving deep into the Pacific Ocean to explore unique Coral Reefs without ever getting wet or running out of air! The Hawaiian Coral Reefs in VR project integrates 3D models derived from thousands of images collected from the field by Dr. John Burns' UH MEGA Lab into a Virtual Reality platform.

A New Way to Look at Coral Reefs

by Dr. John Burns

The multi-scale environmental graphical analysis (MEGA) lab is a next generation research facility that emphasizes the importance of science, technology and fun. We constantly strive to develop innovative techniques to monitor coral reefs by creating 3D models of the benthic environments. These models then are used to interpret the physiochemical, molecular and geophysical factors that affect coral reefs and the humans that depend on these valuable ecosystems.

Leveraging Data Science To Enhance Coral Disease Research

by Dr. John Burns

Can digital 3D reconstructions be used to survey coral disease? UHH students, supported by EPSCoR Ike Wai and the MEGA Lab, conducted a formal analysis to compare conventional human-based SCUBA surveys to digital assessments of coral health using photomosaic imagery. This project resulted in a publication which showed the strengths and weaknesses of both approaches and how data science can help optimize future studies to study coral reef health.

Predicting Students' Success in College Using AI

by Dr. Sukhwa Hong

Predicting students' pathways in college is important because it helps direct their success. This project aims to identify students' critical pathways to graduation and see how they are connected as a network using AI. Through these connections, we will be able to identify what factors result in successful graduation.

Wa'a, A'a, and Ono: A Digital Exploration of Hawaiian History

by Dr. Sukhwa Hong

Collecting, processing, and analyzing large datasets especially unstructured data such as text are economically expensive and time-consuming. In this project, we use natural language processing models to study the old Hawaiian newspaper archive printed in the 19th century, to find recurring topics or themes. With this information, we can understand and grapple with complex contemporary questions and dilemmas by examining how the past has shaped societies and cultures.

Old Newspapers, New Learning: Ho'Ike and Rediscovering Old Hawai'i

by Dr. Sukhwa Hong

Neural Machine translation (NMT) is the process by which machines such as computers are used to translate a text from one natural language to another using neural networks. In this project, we develop neural machine translation to automatically translate newspapers that were printed in Hawaiian. Our neural machine translation allows us to better understand the knowledge and opinion in Hawaiian newspapers.



In Partnership With