CHAI Advanced Ph.D. Student Talk: Manoj Acharya
RIT's Center for Human Aware AI (CHAI) 2022 Spring Seminar Series
CHAI Advanced Ph.D. Student Talk: Manoj Acharya
Title: Autonomous discovery of object categories with deep neural networks
Abstract: Over the past decade, deep neural networks have enormously advanced object classification, object detection, vision and language learning, and much more. However, most of these tasks assume a closed-world, i.e., that they know all of the labels to be predicted during evaluation and that the train distribution is the same as the test distribution. Many real-world applications cannot make this assumption. Robust systems need the ability to operate in a dynamic changing world, where these assumptions do not hold. This is known as open-world learning. In open-world learning, the system must detect examples of unseen labels and continually update the system with new knowledge, without retraining from scratch. In this talk, I review progress in open-world learning, the gaps in the literature, and discuss my work toward enabling efficient multimodal open-world learning in deep neural networks.
Bio: Manoj Acharya is a PhD candidate in Prof. Christopher Kanan's lab at RIT. He has been working on designing vision and language systems with a focus on handling open-world learning challenges. He has published papers in top-AI conferences, including AAAI, NAACL, ECCV, and BMVC. He recently won a continual learning competition held at ICCV-2021. During his PhD, he completed an internship at SRI International where he worked on Graph Neural Networks for detecting out-of-context objects in scenes.
Website: https://www.manojacharya.com/
Event Snapshot
When and Where
Who
Open to the Public
| Cost | FREE |
Interpreter Requested?
Yes