Machine learning ecosystem models

Unlike other ecosystem models, machine learning is built solely from the information it is presented (as opposed to other models where important biological or environmental relationships are built in).

Thus, the models are assumed to contain all the information needed to represent an ecosystem.

Illustration of a feed-forward artificial neural network (ANN).

Model summary

While there are many different types of machine learning, within ecosystem modelling the most commonly used is artificial neural networks (ANN). ANN algorithms can be likened to the neural connections of the brain. Inputs are given to the brain, which are run through a series of nodes (connections) which interpret and make sense of the data. The outputs are new interpretations of the same data from which predictions can be made.

Contact

Principal Scientist - Fisheries
Research subject: Ecosystems