Tuesday 10 March 2026 Host: Oisin Mac Aodha Speaker: Leonie Bossemeyer (School of Informatics)Title: The CleverBirds Dataset: Modelling Visual Expertise from 17M Quiz AnswersAbstract: Mastering fine-grained visual recognition, essential in many expert domains, can require that specialists undergo years of dedicated training. Modelling the progression of such expertise in humans remains challenging, and accurately inferring a human learner’s knowledge state is a key step toward understanding visual learning. We introduce and publicly release CleverBirds, a large-scale knowledge tracing benchmark for fine-grained bird species recognition. Collected by the citizen-science platform eBird, it offers insight into how individuals acquire expertise in complex fine-grained classification. More than 40,000 participants have engaged in the quiz, answering over 17 million multiple-choice questions, spanning over 10,000 bird species and long trajectories per learner. We show that tracking learners’ knowledge is challenging, especially across participant subgroups and question types. CleverBirds is among the largest benchmark of its kind, enabling the development and evaluation of new methods for visual knowledge tracing.Speaker: Philine Bommer, (Encode: AI for Science Fellow in School of GeoSciences)Title: Using deep learning to de-risk climate interventions Abstract: As global warming accelerates, Solar Radiation Management (SRM) is increasingly considered as a tool to mitigate near-term climate risks. To this end, assessing the risks of SRM is crucial, but the assessment is hindered by the high computational cost of traditional Earth System Models (ESMs). We work on a novel framework that leverages deep learning to accelerate the assessment of SRM impacts, specifically focusing on stratospheric aerosol injection and marine cloud brightening. By adapting NeuralGCM—a hybrid atmospheric model combining differentiable physics with machine learning—we aim to construct a forecast tool which provides diverse spatial and temporal scales at a fraction of the traditional computational cost. Mar 10 2026 13.00 - 14.00 Tuesday 10 March 2026 Speakers: Leonie Bossemeyer & Philine Bommer IF, G.03
Tuesday 10 March 2026 Host: Oisin Mac Aodha Speaker: Leonie Bossemeyer (School of Informatics)Title: The CleverBirds Dataset: Modelling Visual Expertise from 17M Quiz AnswersAbstract: Mastering fine-grained visual recognition, essential in many expert domains, can require that specialists undergo years of dedicated training. Modelling the progression of such expertise in humans remains challenging, and accurately inferring a human learner’s knowledge state is a key step toward understanding visual learning. We introduce and publicly release CleverBirds, a large-scale knowledge tracing benchmark for fine-grained bird species recognition. Collected by the citizen-science platform eBird, it offers insight into how individuals acquire expertise in complex fine-grained classification. More than 40,000 participants have engaged in the quiz, answering over 17 million multiple-choice questions, spanning over 10,000 bird species and long trajectories per learner. We show that tracking learners’ knowledge is challenging, especially across participant subgroups and question types. CleverBirds is among the largest benchmark of its kind, enabling the development and evaluation of new methods for visual knowledge tracing.Speaker: Philine Bommer, (Encode: AI for Science Fellow in School of GeoSciences)Title: Using deep learning to de-risk climate interventions Abstract: As global warming accelerates, Solar Radiation Management (SRM) is increasingly considered as a tool to mitigate near-term climate risks. To this end, assessing the risks of SRM is crucial, but the assessment is hindered by the high computational cost of traditional Earth System Models (ESMs). We work on a novel framework that leverages deep learning to accelerate the assessment of SRM impacts, specifically focusing on stratospheric aerosol injection and marine cloud brightening. By adapting NeuralGCM—a hybrid atmospheric model combining differentiable physics with machine learning—we aim to construct a forecast tool which provides diverse spatial and temporal scales at a fraction of the traditional computational cost. Mar 10 2026 13.00 - 14.00 Tuesday 10 March 2026 Speakers: Leonie Bossemeyer & Philine Bommer IF, G.03
