Robust Visual Reasoning with Adriana Kovashka - #463
Published:Mar 11, 2021 15:08
•1 min read
•Practical AI
Analysis
This article summarizes a podcast episode featuring Adriana Kovashka, an Assistant Professor at the University of Pittsburgh. The discussion centers on her research in visual commonsense, its connection to media studies, and the challenges of visual question answering datasets. The episode explores techniques like masking and their role in context prediction. Kovashka's work aims to understand the rhetoric of visual advertisements and focuses on robust visual reasoning. The conversation also touches upon the parallels between her research and explainability, and her future vision for the work. The article provides a concise overview of the key topics discussed.
Key Takeaways
- •The episode discusses visual commonsense research and its intersection with media studies.
- •It explores techniques like masking to improve context prediction in visual question answering.
- •The research aims to understand the rhetoric of visual advertisements and focuses on robust visual reasoning.
Reference
“Adriana then describes how these techniques fit into her broader goal of trying to understand the rhetoric of visual advertisements.”