Explore question answering techniques for language and vision in this 40-minute talk by Richard Socher from MetaMind (A Salesforce Company). Delve into topics such as single joint models, multitask learning, recurrent neural networks, and visual question answering. Learn about attention mechanisms, world knowledge integration, and language patterns. Gain insights into computational neuroscience and speech tagging. Witness a live demonstration showcasing the practical applications of these advanced techniques in AI and machine learning.
Overview
Syllabus
Introduction
Question Answering
Single Joint Model
Single Architecture
Multitask Learning
Recurrent Neural Networks
compute
neuroscience
answer module
speech tagging
visual question answering
attention
world knowledge
language patterns
live demo
Taught by
MITCBMM
