£325Registration Fee
Register Now- Overview
- Instructors
- Schedule
Course Description
This course is aimed towards researchers that want to analyse animal GPS-tracking data using deep learning, particularly for when the goal is to generate movement predictions and downstream outputs such space use and landscape connectivity.
Animal movement behaviour is a sophisticated process that varies across multiple timescales, often with complex interactions to the surrounding environment. Modellers are then faced with many choices about how to specify these dynamic relationships and interactions. Deep learning provides tools that naturally handle these complexities and interactions, with minimal a priori specification required of the models. However, although they are flexible and powerful, deep learning architectures and the code required to implement and fit them can be intimidating.
Here, we will cover the deep learning components (convolutional layers, multilayer perceptions (neural networks), mixture density networks) that comprise the deepSSF approach, showing that they are often simple and straightforward to fit to data. Our aim is then to combine these components into bespoke and customised models for a range of movement data and applications. We will focus on active learning through hands-on programming exercises.
We will use publicly available data sets to demonstrate model training and predictions and use freely available and open-source Python packages. The expected outcome of the course is a solid foundation in certain areas of deep learning, which can be directly applied to animal telemetry data and used to generate predictions.
What You’ll Learn
- A background of predictive animal movement modelling, in particular of step-selection functions
- Understanding deep learning components such as convolutional layers, multilayer perceptrons (neural networks), and mixture density networks
- Backpropagation, loss functions and training procedures
- Using the deepSSF Python package to train deep learning models on different datasets
- Assessing model performance using withheld data
- Generating movement trajectories from fitted models, which can be used for downstream predictions such as animal space use and landscape connectivity
Course Format
Interactive Learning Format
Each day features a well-balanced combination of lectures and hands-on practical exercises, with dedicated time for discussing participants’ own data, time permitting.
Global Accessibility
All live sessions are recorded and made available on the same day, ensuring accessibility for participants across different time zones.
Collaborative Discussions
Open discussion sessions provide an opportunity for participants to explore specific research questions and engage with instructors and peers.
Comprehensive Course Materials
All code, datasets, and presentation slides used during the course will be shared with participants by the instructor.
Personalized Data Engagement
Participants are encouraged to bring their own data for discussion and practical application during the course.
Post-Course Support
Participants will receive continued support via email for 30 days following the course, along with on-demand access to session recordings for the same period.
Who Should Attend / Intended Audiences
This course is intended for academics and postgraduate students working on projects involving the analysis and/or prediction of animal telemetry data, as well as applied researchers and analysts in public, private, or third-sector organizations. Users should have a basic understanding of statistical concepts such as regression, and some experience with data analysis and manipulation including the ability to import and export data, manipulate data frames, fit basic statistical models, and generate exploratory and diagnostic plots. Although the language used will be Python, and familiarity with Python will help, an understanding of similar programming languages (e.g. R) will also be beneficial.
Equipment and Software requirements
A laptop or desktop computer with a functioning installation of Python, and usually an Integrated Development Environment (IDE) such as VSCode is required. Both Python and VSCode are free, open-source programs compatible with Windows, macOS, and Linux systems.
While not essential, using a large monitor—or ideally a dual-monitor setup—can significantly enhance your learning experience by allowing you to view course materials and work in R simultaneously.
A working webcam is recommended to support interactive elements of the course. We encourage participants to keep their cameras on during live Zoom sessions to foster a more engaging and collaborative environment.
All necessary R packages will be introduced and installed during the workshop.
Dr. Scott W. Forrest
Scott is a quantitative ecologist working at the interface of habitat selection, animal movement, and landscape ecology. His research focuses on methods for scaling up from fine-scale decision processes to longer term properties such as their home ranges, distribution and landscape connectivity. Central to this is the use and development of predictive approaches such as individual-based simulations and deep learning, and the representation of dynamic behaviours. He also uses a range of statistical methods to model animal movement and habitat selection for conservation applications, and has worked with data from a range of terrestrial species. Recently, he also has been analysing remote sensing data using computer vision, detecting and classifying waterholes across vast areas of tropical savanna.
Scott earned his PhD in Mathematics from the Queensland University of Technology (QUT) in collaboration with the Commonwealth Scientific and Industrial Research Organisation (CSIRO) (supervised by Prof. Michael Bode, Prof Christopher Drovandi, Dr Andrew Hoskins and Dr Dan Pagendam), following a Master’s degree in Wildlife Management from the University of Otago (supervised by Prof. Philip Seddon and A.Prof. Michael Paulin) and an Honours degree in Engineering (Naval Architecture). He is currently a postdoctoral researcher at the University of Tasmania with Dr James Pay, where he is analysing behavioural risk and adaptations of wedge-tailed eagles to wind energy facilities, a research assistant in computer vision at QUT, and a self-employed statistical consultant.
Education & Career
- PhD in Mathematics, Queensland University of Technology
- M.Sc. in Wildlife Management, University of Otago
- BEng (Hons) in Naval Architecture, Flinders University, Australian Maritime College
- Postdoctoral Researcher, University of Tasmania
- Research Assistant, Queensland University of Technology
- Statistical consultant
Research Focus
Scott’s work centers on capturing fine-scale and behavioural processes in animal movement models, and how they change across daily and seasonal timescales, often with a view to generating predictions. These analyses and predictions are often to inform conservation management, such feral species control, quantifying risks from wind energy facilities or human infrastructure, and how animals respond to conservation translocations.
Current Projects
- Developing methods for predicting animal movement with deep learning
- Predicting distributions and landscape connectivity of feral species such as pigs and buffalo for improved management practices
- Quantifying risk and avoidance behaviour of eagles to wind energy facilities
- Detecting and classifying waterholes from satellite imagery across north Australia’s tropical savannas
Teaching & Skills
- Delivered workshops in statistical methods for movement ecology and species distribution modelling,
- Expert in applying statistical modeling, telemetry data, GIS, and simulation approaches to movement ecology
Links
Session 1 – 02:00:00 – Introduction to SSF and deepSSF
Overview of step selection functions as the foundation for deepSSF. Combining animal movement and habitat selection to learn about behaviour and generate predictions. Covers some introductory deep learning concepts in the context of animal movement.
- SSF concepts
- Create selection surfaces using SSFs
- Increasing flexibility in SSFs
- deepSSF
- Movement process (neural networks)
- Habitat selection process (convolution layers)
- How these combine into the next-step probability surface
Break – 00:30:00
Session 2 – 01:00:00 – Introduction to the deepSSF package
- Overview of the deepSSF package and its functions.
Session 3 – 01:30:00 – Data preparation – from CSV to Dataloader
- Movement data (CSV) and spatial environmental data (raster/tif) preparation. Creating Dataset and Dataloader objects using PyTorch.
Break – 00:30:00
Session 4 – 01:30:00 – deepSSF model architecture
- Creating the deepSSF model in PyTorch. Combining the model components, and selecting the number of neural network layers and weights and the number of convolution layers and kernels. How this affects training and performance.
Session 5 – 01:30:00 – Training models – loss functions, optimisers, schedulers and early stopping
- Overview of the loss function, optimiser, schedulers and early stopping functions required for training deepSSF models, and how these combine into a training loop.
Break – 00:30:00
- Session 6 – 01:30:00 – Assessing and validating model performance
- Testing predictive performance of both the movement and habitat selection processes against with-held data. Extracting probability weights and diagnostic plots to determine what the model has learned.
Session 7 – 02:00:00 – Simulating animal movement from deepSSF models
- Using the trained deepSSF model to simulate animal movement trajectories. How these might be used to generate space use estimates, landscape connectivity and scenario testing.
Break – 00:30:00
Session 8 – 01:00:00 – Conclusions and future possibilities
A recap of deepSSF models and how they’re trained, and how they might be further developing into the future.
Frequently asked questions
Everything you need to know about the product and billing.
When will I receive instructions on how to join?
You’ll receive an email on the Friday before the course begins, with full instructions on how to join via Zoom. Please ensure you have Zoom installed in advance.
Do I need administrator rights on my computer?
I’m attending the course live — will I also get access to the session recordings?
I can’t attend every live session — can I join some sessions live and catch up on others later?
I’m in a different time zone and plan to follow the course via recordings. When will these be available?
I can’t attend live — how can I ask questions?
Will I receive a certificate?
When will I receive instructions on how to join?
You’ll receive an email on the Friday before the course begins, with full instructions on how to join via Zoom. Please ensure you have Zoom installed in advance.
Do I need administrator rights on my computer?
I’m attending the course live — will I also get access to the session recordings?
I can’t attend every live session — can I join some sessions live and catch up on others later?
I’m in a different time zone and plan to follow the course via recordings. When will these be available?
I can’t attend live — how can I ask questions?
Will I receive a certificate?
Still have questions?
Can’t find the answer you’re looking for? Please chat to our friendly team.








