TAIAO Workshop 2023

Join the TAIAO Workshop: Celebrating 50 Years of Computing at UoW (AI Month)

Prepare for an extraordinary event as we invite you to the highly anticipated TAIAO Workshop, in celebration of 50 years of Computing at the University of Waikato. Join us in August, during the AI Month, at the vibrant Tauranga Campus for an unforgettable gathering. Immerse yourself in the forefront of data science developments, with a special focus on cutting-edge environmental-driven use cases.

Mark your calendar with the event details:

Venue: The University of Waikato, Tauranga Campus, 101 Durham Street, Tauranga 3110

Date: 24 August 2023 (Thursday)

Time: 9am - 5pm

Workshop Objectives:

  • Ignite engaging conversations and collaborations as we delve into the latest and future trends in data science.
  • Discover the TAIAO's innovative environmentally driven use cases, sparking connections with potential organisations and stakeholders.
  • Unleash your knowledge and insights, as you learn, network, and establish valuable relationships in an inspiring environment.
Click here to register.

Registration for this event is completely free but act fast, spots are limited! Secure your place by registering now and mark your calendar. The deadline for registration is Thursday, August 17, 2023. Can't attend in person? No worries! The entire workshop will be live-streamed and you don’t need to register for this. You can find the link below.

TAIAO Workshop programme


Download the programme here.

TAIAO Workshop speakers

Presenting the dynamic lineup of speakers for this workshop, all prepared to share invaluable insights and knowledge with you:

Plenary speakers

Vanessa Clark, Executive Director Kanapu, Ngā Pae o te Māramatanga Māori Centre of Research Excellence

Kanapu - Expanding the Impact of Vision Mātauranga

Under the korowai of Ngā Pae o Te Māramatanga, Kanapu is a 6-year MBIE-funded programme with a mission to connect, attract, nurture and accelerate Māori talent and leadership across the research, science and innovation ecosystem. Vanessa will share learnings, insights and emerging themes from Kanapu's foundation year and growing call from hapori Māori and local practitioners to develop their skills and lead community projects.

Vanessa is Pouhere Kanapu | Executive Director of Kanapu, under the korowai of Ngā Pae o te Māramatanga Māori Centre of Research Excellence and based at the University of Waikato. Vanessa spent the majority of her career in the Information and Communications Technology sector in Hong Kong, the UK, Australia, and the USA. Since returning to NZ in 2011, Vanessa has held a variety of roles in the private and public sectors; including Kā Hao Māori ICT Development Fund (Advisor), Te Māngai Pāho (Board member) and Science for Technological Innovation National Sciences Challenge (Kāhui Māori). In 2021, she was elected to Te Arataura, the tribal executive of Waikato-Tainui.

Her research interests include Māori data sovereignty and entrepreneurship. She has a Bachelor of Business Studies in 1992 (Massey) and Masters of Management Studies in 1998 (Waikato). Proud māma of Matteo and Nadia, Vanessa lives in Whatawhata.

Christo Rautenbach, Coastal and Estuarine Physical Processes Scientist, NIWA

Interpretable Deep Learning Applied to Rip Current Detection and Localization

Christo will talk about his research on the application of interpretable deep learning to rip current detection and localisation. While there have been some significant advances in AI for detecting and localising rip currents, there is a lack of research ensuring that an AI algorithm can generalise well to a diverse range of coastal environments and marine conditions. The study utilised an interpretable AI method, known as gradient-weighted class-activation maps (Grad-CAM), which represents a novel approach for detecting amorphous rip currents. Christo is a Coastal and Estuarine Physical Processes Scientist at NIWA. He holds dual PhDs, the first in Applied Mathematics and the second in Physical Oceanography, with a wide-ranging background in numerical modelling. His primary focus revolves around coastal and ocean hydrodynamics as well as wave dynamics. He has over 12 years of experience as a senior scientist in the fields of operational physical oceanography, coastal engineering, and coastal dynamics research.

Presentations by TAIAO project members

Jack Julian, PhD Student, University of Auckland

Anomaly Detection with Continual Learning for Maritime Trajectories

Jack studies irregular movements of maritime vessels, which can indicate illegal fishing activity, oil spills and potential biosecurity risks. With the help of continual learning techniques, these anomalies can be detected in real time to identify and prevent environmental hazards.

Guilherme Cassales, Postdoctoral Research Fellow, University of Waikato

Forecasting Plantation Forests Growth: Current State and Future Directions

In plantation forests, where returns take more than a decade to materialize, proactive planning is crucial. Current technologies can often provide real-time spatio-temporal data, which require appropriate methods to successfully extract knowledge and use these large and complex databases. In this talk Guilherme will discuss the utilization of Machine Learning (ML) tools to optimize resource allocation with forestry hydrology data and the exciting opportunities for future research, particularly related to the analysis of intricate edaphoclimatic data collected throughout the forests.

Olivier Graffueille, PhD Student, University of Auckland

Exploring Machine Learning methods for Water Quality Remote Sensing

Olivier’s research focuses on using machine learning techniques for water quality remote sensing. He explores characteristics of this problem from a data perspective, which allows him to link these to relevant machine learning ideas. This allows him to better model this important environmental problem, while also advancing fundamental machine learning research.

Ding Ning, PhD Student, University of Canterbury

Graph-Based Deep Learning for Sea Surface Temperature Anomaly Forecasts

Ding will talk about his research on using graph re-sampling on the gridded ERA5 product and a graph neural network for learning from large graphs that evolve over time, to forecast global monthly mean sea surface temperatures and their anomalies.

Moritz Lehmann, Senior Oceanographer, Starboard Maritime Intelligence

Panel/open discussion: Generative AI and ChatGPT: How does seemingly sentient technology relate to Environmental Data Science

Generative AI, like ChatGPT, is valuable for researchers due to its ability to efficiently handle large datasets, simulate environmental scenarios, and simplify complex scientific content into understandable plain language. It has even demonstrated the capability to generate complete research articles, including data analysis, within just one hour. This panel discussion led by Moritz will delve into how TAIAO researchers are using generative AI, evaluating its creativity, and discussing the legitimacy of the current hype surrounding it. The discussion will incorporate audience questions and aims to shed light on future generative AI advancements, offering guidance to students and practitioners on its effective utilization.

Yun Sing Koh, Associate Professor, School of Computer Science, University of Auckland

Open discussion: Growing capabilities in Environmental Data Science

As the world grapples with ever more intricate environmental challenges, an escalating imperative emerges: the need to harness the formidable power of data to gain profound insights into ecological systems, climate trends, and the prudent administration of natural resources. With the continuous evolution of technology and the vast expansion of datasets, expertise in environmental data science assumes a progressively vital role, equipping us to proactively address emerging environmental threats and unveil patterns and correlations that were once concealed. This open discussion delves into the ongoing evolution of capabilities and anticipates future requirements for enhancing proficiencies within the realm of Environmental Data Science.

Nick Lim, Postdoctoral Research Fellow, AI Institute, University of Waikato

Video-based Annotation and Classification Tool for Underwater Habitat Classification

Recent advances in storage and technology have made available large volumes and a variety of data accessible for training machine learning algorithms and conducting data analysis. However, a significant challenge with handling these data is the accurate annotation of the complex and dynamic scenes, especially in very specialised domains such as underwater habitats. Existing publicly available video annotation tools lack the flexibility to annotate the multiple hierarchical labels and uncertainty in the labels, while bespoke solutions can be inflexible and can be costly to implement and deploy. In this talk, Nick will showcase a web video-based annotation software designed for the Department of Conservation for undersea habitat classification. He will present some preliminary classification results for automatic annotations to reduce the workload of the annotators and describe how the system can be used for other domains that use videos for classification.

Phil Mourot, Senior Research Fellow, University of Waikato; Nick Lim, Postdoctoral Research Fellow, AI Institute, University of Waikato; Corey Sterling, Research Programmer, AI Institute, University of Waikato

Deploying Deep Learning Models for Real-time Flood Prediction

Floods are one of nature's most costly and deadly natural disasters. Accurate and reliable forecasting of floods are invaluable to mitigate and manage these deadly disasters and artificial intelligence have demonstrated that it is an invaluable tool for flood forecasting. However, leveraging on the advances in machine learning and artificial intelligence for flood prediction can be challenging especially for users who are not well versed in deploying machine learning systems. This talk will present an end-to-end flood prediction system, including the deployment process, to address these challenges.

Gregory Pearson, Meteorological Design Engineer, MetService

Impact of Machine Learning on Weather Models

Drawing from his extensive experience in atmospheric modelling, forecasting, and statistical techniques, Greg Pearson offers an introduction to weather modelling and will discuss the impacts of some recent work in machine learning.

Karin Bryan, Professor in Earth and Environmental Science, University of Waikato

Classification to Augment Numerical Modelling Results for Ohiwa Harbour

Planning for future changes to climate in our coastal environment is one of the most challenging tasks facing coastal managers. Yet, many of our modelling systems are too complex and detailed to undertake projections at appropriate timescales. In this talk, Karin explores classification and neural networks as a way of using coastal dynamic models to emulate future climate systems, using Ōhiwa Harbour as a case study.

Simna Rassak, PhD Student, University of Waikato and Institute of Environmental Science and Research (ESR)

Leveraging Multi-Source Data for Comprehensive Analysis of Climate Change, Environment, Health, and Socio-Economic Impacts

Simna will talk about her work where she conducted a comprehensive analysis of multisource data on manuka-dominated ecosystem, which integrates remote sensing observations, data from soil sensors, and ground truth information. The purpose of the analysis is to systematically explore the pivotal role played by this plant species in mitigating water pollution resulting from farming activities. The study also assesses the impacts of introducing native species on soil health and biodiversity, with the overarching goal of contributing to the restoration of Lake Waikare's mauri.

Varvara Vetrova, Senior Lecturer, University of Canterbury

Anomaly Detection: From Biosecurity to Climate

Detecting anomalies on time can help us on very different scales. On one hand, it could prevent invasive species from taking hold in our cities. On the other hand, we can better understand what happens with the climate system, such as extremes in Antarctica. In this talk, Varvara will delve into deep learning methods behind these applications.