Introduction to the course

Welcome to Deep Learning for Bioscientists.

The course is designed to take around two to three hours a week over five weeks to complete. It is aimed at bioscience professionals, particularly in the field of plant phenotyping, who are interested in learning more about deep learning, and how it can help their research.

Learning outcomes

The learning outcomes for the course are as follows:

• Compare machine learning with deep learning
• Practice the use of GPU acceleration with Google Colab
• Explain each of the key components of convolutional neural networks
• Create a simple convolutional neural network using PyTorch
• Perform the basic steps needed to train a deep learning model
• Improve the deep learning training process by adjusting hyperparameters
• Create a simple image classifier using PyTorch
• Modify a deep learning network for use in regression
• Explore the use of deep learning for image segmentation
• Compare other deep learning approaches such as heatmap regression and multi-task learning

DataCampp

Development and delivery of this course is supported by a UKRI Large-scale data training grant MR/V038850/1, “Data CAMPP (Innovative Training in Data Capture, Analysis and Management for Plant Phenotyping)”, and is part of a series of courses developed for that project.

If you haven’t already, we recommend taking two of the previous courses in this series before starting this course, especially if you are new to programming with Python:

• Introduction to Image Analysis for Plant Phenotyping
• Machine Learning for Image Data

Otherwise, if you have some experience with Python, image analysis, and understand the basic concepts of machine learning, and want to expand that knowledge to deep learning, you may be able start here. If you find some of the concepts and discussion hard to follow however, we would recommend looking at the material in the courses listed above.

Software

All the practicals in the course are done using Python, in particular the deep learning package PyTorch (https://pytorch.org/), which we will be run in an internet browser using Colab (https://colab.research.google.com/). The latter will require a free account with Google, and we will explain the use of both as we go through the material.

The people behind the course

Your lead educators are: Michael Pound, Andrew French and Nathan Mellor.

We would also like to thank Sean Riley of Boardie Video Production for the video production itself.

You might like to introduce yourself in the comments section below and let us know why you are taking the course, and what you are hoping to learn.