Attention Mechanisms: Knowing Where to Look Improves Visual Reasoning
The human visual cortex uses attention mechanisms to discard irrelevant information as well as to efficiently allocate computational resources. It has inspired modern machine learning, where attention mechanisms are a vital part of memory modules and are used for modelling object interaction as well as solving complex reasoning tasks. In this talk, we explore attention mechanisms for visual tasks and show how they can help to track objects in real-world videos when used in a recurrent framework with a hierarchy of attention mechanisms. Attention is also able to model common assumptions we make about objects - that they do not appear out of nowhere and do not disappear into thin air. This insight lends itself to unsupervised detection and tracking of multiple objects - without any human supervision.
Adam R. Kosiorek is a PhD candidate at the University of Oxford. His research interests lie on the intersection of deep learning and machine reasoning, with the goal of achieving artificial general intelligence. Over the last five years Adam has worked on various applied and research machine learning projects at IBM, Samsung, Bloomberg and he is now a research intern at Google DeepMind. In his free time, Adam reads lots of books, trains gymnastics and is a hiking enthusiast.