onlincourse

Machine Learning for Image Data

Master the principles and applications of machine learning for image data to harness its potential for plant phenotyping, with this online course from the University of Nottingham.

Duration

5 weeks

Weekly study

3 hours

100% online

How it works

Unlimited subscription

Learn more

Established

1996

Location

Lincoln, Lincolnshire, UK

Become an expert in machine learning for bioscience

Machine learning has made it possible to process vast quantities of image data. That means it can enhance and facilitate the work of bioscience researchers, particularly the field of plant phenotyping.

On this five-week course from the University of Nottingham, you’ll gain an overview of the applications of machine learning for image data, focusing specifically on its use in plant phenotyping.

Gain an overview of machine learning as it applies to biological image data

You’ll start the course with an overview of machine learning, and an introduction to image data and features.

You’ll gain the background you need to understand and apply machine learning in your own bioscience research.

Master common techniques and softwares for image analysis

Once you’ve mastered the principles of machine learning for image data, you’ll start building the practical skills you need to navigate machine learning software.

Weeks 3 and 4 of the course will cover the main techniques for processing image data, some common challenges surrounding these, and useful tips and tricks to help you overcome them.

Whether you want to model data through a decision tree or create visualisations using Python, you’ll gain the hands-on experience you need for your research.

Understand neural networks and deep learning

In your last week of the course, you’ll look more closely at a specific subfield of machine learning: deep learning. You’ll learn how neural networks can be used to process biological images in the same way the human brain would.

By the end of the course, you’ll have an understanding of how machine learning can be used with biological image data, and the skills you need to harness it in your own bioscience research.

Week 1
What is machine learning?
- Introduction to machine learning
  An introduction to the course, and an introduction to image data and features.
- Example machine learning problems
  A discussion of the main types of problems you might solve with machine learning, and the kinds of problems that are specific to image data. We also look at classification and clustering applications with a set of example data.
- Supervised versus unsupervised learning
  To tackle any problem using machine learning you need to establish whether supervised or unsupervised learning is appropriate to your dataset. This activity explains the difference.
- Common tasks - classification and regression
  An overview of the common machine learning tasks classification and regression, plus a look at some other frequently used terminology.
- Software tools
  An overview to the software packages used in the course, including Scikit-Learn, Matplotlib, and Pandas.
- Summary and review
  Summary and review of week 1, with a quiz and practical activity.
Week 2
Data and features
- Types of data and features
  An overview of the types of data used in machine learning, and an introduction to features and feature extraction.
- Feature extraction
  A look at feature extraction for use in machine learning. A particular focus on feature extraction from image data.
- Labelling image data
  Image data often needs to be labelled or annotated for use in machine learning models. This activity goes over why and how you might annotate image data, and introduces some software tools.
- Pre-processing data
  How to deal with noisy and incomplete data, and a look at pre-processing of image data.
- Summary and review
  A review of the week's content, with a practical and a quiz.
Week 3
Common techniques
- Introduction
  In week 3 we will look in more detail at some common machine learning methods for clustering, classification, and regression. Plus a look at methods for model evaluation, visualisation, and selection.
- Clustering
  A closer look at clustering methods, in particular K-means clustering.
- Classification
  A closer look at the common classification methods of Decision Trees and Naive Bayes.
- Regression
  A look at regression techniques, in particular linear regression.
- Evaluation, visualisation, and selection
  How to evaluate your machine learning models. Includes accuracy, precision, recall, and F scores. Plus a look at ways to visualise your results, including confusion matrices, and some advice on model selection.
- Summary and review
  A review of the week's content, with a practical and a quiz.
Week 4
Tips and tricks
- Introduction
  An introduction to Week 4, Tips and Tricks.
- Good Training Practice
  A look at choice of features, using learning performance curves to improve model training, splitting datasets and use of cross-validation.
- Data augmentation
  A look at methods to artifically increase the size of datasets by using data augmentation.
- Common challenges
  Including overfitting, regularisation, the "Curse of Dimensionality", and class imbalance.
- Summary and review
  A review of the week's content, with a quiz and practical activity.
Week 5
Deep learning
- What is deep learning?
  A look at what deep learning is and how it compares with the machine learning learning methods we have considered previously in the course.
- Neural networks and deep learning
  An overview of how deep learning systems are constructed. Starting with perceptrons, then neural networks, and finally convolutional neural networks.
- Some simple tools
  We won't be doing any practical deep learning within this course. But to give you a taste of how we will cover it in future units, we introduce Python notebooks and Colab, and provide some links for further reading.
- Summary and review
  A review of the week and course's content, with a quiz