Gabriela Csurka presents Language based Explainability in Computer Vision
On 2025-01-28 11:00:00 at G205, Karlovo náměstí 13, Praha 2
Explainable AI focuses on creating AI models that provide clear explanations
about their outcomes, decisions, or predictions with the goal of making these
systems more interpretable and understandable to humans. In the field of
computer vision, significant progress has been made through deep neural
networks, however, these models can be considered “black box” systems due to
their complexity, making them difficult for humans to understand. In this talk,
I will delve into the concept of explainability and interpretability in computer
vision, exploring both visual based and language-based explanations.
Subsequently, I will present our simple yet effective solution to predict and
describe via natural language potential failure modes of computer vision models
about their outcomes, decisions, or predictions with the goal of making these
systems more interpretable and understandable to humans. In the field of
computer vision, significant progress has been made through deep neural
networks, however, these models can be considered “black box” systems due to
their complexity, making them difficult for humans to understand. In this talk,
I will delve into the concept of explainability and interpretability in computer
vision, exploring both visual based and language-based explanations.
Subsequently, I will present our simple yet effective solution to predict and
describe via natural language potential failure modes of computer vision models