Skip to content

Upcoming Events (Fall ’26 Schedule Coming Soon)

Flyer with the following text: Wildlife Seminar Fall ‘26
Professor Yun Sing Koh,
Director of the Centre of Machine Learning for Social Good, Te Kura Matai Rorohiko School of Computer Science, The University of Auckland
AI for Wildlife Monitoring and Ecological Intelligence: A New Zealand Perspective
Artificial intelligence has achieved remarkable advances in perception and recognition, yet many AI systems continue to struggle in dynamic, open-world environments where data distributions evolve, labelled examples are limited, and new concepts emerge over time. Wildlife conservation presents precisely these challenges. Building AI that can operate reliably in natural ecosystems therefore requires advances beyond conventional supervised learning, spanning continual learning, open-world recognition, multimodal foundation models, uncertainty estimation, and trustworthy AI.
New Zealand provides a unique testbed for this research. Its distinctive biodiversity, together with ambitious conservation programmes targeting introduced predators and endangered native species, creates opportunities to develop AI methods that must adapt to changing environments while supporting real-world conservation decisions.
In this talk, I will present recent AI research from the Centre for Machine Learning for Social Good at the University of Auckland, where we are developing intelligent systems for long-term wildlife monitoring in dynamic, open-world environments. Our work spans computer vision, multimodal foundation models, continual learning, open-world recognition, and trustworthy AI to enable the identification of individual animals, adaptation to changing ecosystems, and learning from limited and evolving data. These challenges are motivating new advances in representation learning, robust multimodal reasoning, and AI systems capable of reliable long-term deployment beyond benchmark datasets.
The talk will conclude by examining how wildlife conservation is driving advances in artificial intelligence. The need for AI systems that can recognise new individuals, adapt to changing environments, learn continuously, and make reliable decisions under uncertainty is motivating research that extends beyond conservation and contributes to the development of more general, robust, and trustworthy intelligent systems.
Artificial intelligence has achieved remarkable advances in perception and recognition, yet many AI systems continue to struggle in dynamic, open-world environments where data distributions evolve, labelled examples are limited, and new concepts emerge over time. Wildlife conservation presents precisely these challenges. Building AI that can operate reliably in natural ecosystems therefore requires advances beyond conventional supervised learning, spanning continual learning, open-world recognition, multimodal foundation models, uncertainty estimation, and trustworthy AI.
New Zealand provides a unique testbed for this research. Its distinctive biodiversity, together with ambitious conservation programmes targeting introduced predators and endangered native species, creates opportunities to develop AI methods that must adapt to changing environments while supporting real-world conservation decisions.
In this talk, I will present recent AI research from the Centre for Machine Learning for Social Good at the University of Auckland, where we are developing intelligent systems for long-term wildlife monitoring in dynamic, open-world environments. Our work spans computer vision, multimodal foundation models, continual learning, open-world recognition, and trustworthy AI to enable the identification of individual animals, adaptation to changing ecosystems, and learning from limited and evolving data. These challenges are motivating new advances in representation learning, robust multimodal reasoning, and AI systems capable of reliable long-term deployment beyond benchmark datasets.
The talk will conclude by examining how wildlife conservation is driving advances in artificial intelligence. The need for AI systems that can recognise new individuals, adapt to changing environments, learn continuously, and make reliable decisions under uncertainty is motivating research that extends beyond conservation and contributes to the development of more general, robust, and trustworthy intelligent systems.
Friday 9/4/26, 12noon-1pm, 
Mulford Room 36 or https://berkeley.zoom.us/j/94180561590 info: wildlife.berkeley.edu

Professor Yun Sing Koh


Director of the Centre of Machine Learning for Social Good, Te Kura Matai Rorohiko School of Computer Science, The University of Auckland


AI for Wildlife Monitoring and Ecological Intelligence: A New Zealand Perspective


Mulford Room 36 or https://berkeley.zoom.us/j/94180561590

To be added to our list, email us at berkeleywildlife@berkeley.edu

Previous Events