Stats Literacy Sketch

Intro to AI and Colab

This session follows on from the first workshop on AI late last year, where a desire for a practical introduction to using AI methods was expressed. Brett Drury (Rothamsted) started by providing a primer on AI, including:

  • Key terms and concepts in machine learning

  • Types of learning

  • Roles of features, labels, datasets

  • Common evaluation metrics

The slides for this can be viewed here.

He also provided an introduction on using Google Colab to help with the practical session.

RNN practical

Nick Schurch (BioSS) then led the practical where we built a Recurrent Neural Network (RNN) to predict river discharge from time series data, with univariate and multivariate examples. The Colab notebook written by Nick and Joe Wilde (BioSS) also contains additional notes including:

  • Structure of RNNs and how they work

  • Activation functions and optimisers

  • Library options for AI

The notes and code can be accessed here along with the data needed for the examples.