December 4, 2024 Seminar - UF/IFAS AI and Data Science Seminar Series
Автор: UF/IFAS Agricultural and Biological Engineering
Загружено: 2025-01-28
Просмотров: 104
Описание:
Speaker: Dr. Alina Zare
The Strength to Say “I Don't Know”: expanding neural networks to address out of distribution samples and ambiguous inputs
Machine Learning methods, and specifically neural networks, have shown impressive capabilities on a wide range of application areas. However, these models are often brittle when seeing unexpected or novel data that differs in significant or meaningful ways from the data that was used to train them. Often, this is overcome with ever-increasing training data volumes. However, there are many applications that are inherently data limited or scenarios in which systems encounter ambiguous or out-of-distribution samples. In these cases, the ability for a model to be able to produce the output “I Don’t Know, Never seen anything like this before” can be critical. In this talk, Dr. Alina Zare will discuss methods that her group has developed to flag out-of-distribution and address ambiguous inputs in test data.
Повторяем попытку...
Доступные форматы для скачивания:
Скачать видео
-
Информация по загрузке:
