Module: Machine Learning
This module will introduce you to the world of machine learning using Galaxy
Introduction
An introduction to Machine Learning in Galaxy
Lecture Video
| Description: | The lecture explains introductory concepts in machine learning such as supervised and unsupervised learning, classification and regression, hyperparameter optimisation, cross-validation, train, test and validation sets. |
| Length: | 15 minutes |
| Captions: | Anup Kumar |
| Created: | 15 February 2021 |
| Materials: | |
| Support: |
Tutorial Video
| Description: | The talk includes a lecture followed by a hands-on session to apply multiple classification algorithms on the Quantitative structure-activity relationship (QSAR) dataset to predict the biodegradable activity of chemical compounds. QSAR models attempt to predict the activity or property of chemicals based on their chemical structure. To achieve this, a database of compounds is collected for which the property of interest is known. For each compound, molecular descriptors are collected which describe the structure (for example - molecular weight, number of nitrogen atoms, number of carbon-carbon double bonds). Using these descriptors, a model is constructed which is capable of predicting the property of interest for a new, unknown molecule. In this tutorial, we will use a database assembled from experimental data of the Japanese Ministry of International Trade and Industry to create a classification model by applying simple and complex classifiers to learn the nature of biodegradation. We will use this model to classify new molecules into one of two classes - biodegradable or non-biodegradable. Different visualisations are used to analyze the results after applying each classification algorithm. Hyperparameters of one of the classifiers are also optimised. |
| Length: | 1 hour 50 minutes |
| Captions: | Anup Kumar |
| Created: | 15 February 2021 |
| Materials: | |
| Support: | |
| Supported Servers: | Known Working Possibly Working |
| Learning Objectives: |
|
Tutorial:
Lecture & Tutorial Video
| Description: | The talk includes a lecture followed by a hands-on session to apply multiple regression algorithms on the DNA-methylation dataset to predict biological age. In this tutorial, we will build a regression model for chronological age prediction, based on DNA methylation. This is based on the work of Jana Naue et al. 2017, in which biomarkers are examined to predict the chronological age of humans by analyzing the DNA methylation patterns. Different machine learning algorithms are used in this study to make an age prediction. It has been recognized that within each individual, the level of DNA methylation changes with age. This knowledge is used to select useful biomarkers from DNA methylation datasets. The CpG sites with the highest correlation to age are selected as the biomarkers (and therefore features for building a regression model). In this tutorial, specific biomarkers are analyzed by machine learning algorithms to create an age prediction model. Multiple visualisations are also used to analyse the predictions made by simple and complex regressors and hyperparameters of one of the regressors also optimised. |
| Length: | 1 hour 29 minutes |
| Captions: | Anup Kumar |
| Created: | 15 February 2021 |
| Materials: | |
| Support: | |
| Supported Servers: | Known Working Possibly Working |
| Learning Objectives: |
|
Machine Learning with R in Galaxy
Want to go beyond the standard Machine Learning tools in Galaxy? Get a bit more control over you analysis by using R in Rstudio within Galaxy!
Speaker
Tutorial Video
| Description: | This tutorial will provide an introduction to using R with Rstudio in Galaxy. |
| Length: | 50 minutes |
| Captions: | Sébastien Fouilloux, Anne Fouilloux, Kaivan Kamali |
| Created: | 15 February 2021 |
| Materials: |
|
| Support: | |
| Supported Servers: | |
| Learning Objectives: |
|
Speaker
Tutorial Video
| Description: | This is an Introduction to Machine Learning in R, in which you’ll learn the basics of unsupervised learning for pattern recognition and supervised learning for prediction. At the end of this workshop, we hope that you will - appreciate the importance of performing exploratory data analysis (or EDA) before starting to model your data. - understand the basics of unsupervised learning and know the examples of principal component analysis (PCA) and k-means clustering. - understand the basics of supervised learning for prediction and the differences between classification and regression. - understand modern machine learning techniques and principles, such as test train split, k-fold cross validation and regularization. - be able to write code to implement the above techniques and methodologies using R, caret and glmnet. We will not be focusing on the mathematical foundation for each of the methods and approaches we’ll be discussing. There are many resources that can provide this context, but for the purposes of this workshop we believe that they are beyond the scope. |
| Length: | 1 hour 30 minutes |
| Captions: | Maria Tsagiopoulou |
| Created: | 15 February 2021 |
| Materials: |
|
| Support: | |
| Supported Servers: | Known Working |
| Learning Objectives: |
|
Speaker
Deep Learning Demystified
This 3-part session will introduce you to the world of deep learning. We cover Feedforward Neural Networks (FNN), Recurrent Neural Networks (FNN), and Convolutional Neural Networks (CNN).
Tutorial Video
| Description: | This hands-on tutorial covers Feedforward Neural Networks (FNN) |
| Length: | 1 hour 10 minutes |
| Captions: | Kaivan Kamali |
| Created: | 15 February 2021 |
| Materials: |
|
| Support: | |
| Supported Servers: | Known Working Possibly Working |
Speaker
Tutorial Video
| Description: | This hands-on tutorial covers Recurrenct Neural Networks (RNN) |
| Length: | 50 minutes |
| Captions: | Kaivan Kamali |
| Created: | 15 February 2021 |
| Materials: |
|
| Support: | |
| Supported Servers: | Known Working Possibly Working |
Speaker
Tutorial Video
| Description: | This hands-on tutorial covers Convolutional Neural Networks (CNN) |
| Length: | 1 hour |
| Captions: | Filippos Zacharopoulos |
| Created: | 15 February 2021 |
| Materials: |
|
| Support: | |
| Supported Servers: | Known Working |
Speaker
Tutorial Video
| Description: | The classification of fruits and vegetables offers many useful applications such as automated harvesting by robots, building up stocks for supermarkets, effective detection of specific defects, and determining fruit ripeness (Duong et al. 2020, Naranjo-Torres et al. 2020, Iswari et al. 2017). Machine Learning (ML) techniques such as Deep Learning (DL) are commonly used for image classification problems in various domains, including in agriculture (Kamilaris and Prenafeta-Boldú 2018). DL is a technique inspired by how a human brain operates. Due to the increased availability of compute capacity and training data, DL techniques have become very popular in recent years. In this tutorial, we will use Galaxy’s ML toolkit to build a DL model to classify fruit and vegetable images. Our DL model is trained and evaluated on Fruit 360 dataset (Mureşan and Oltean 2018) |
| Length: | 1 hour |
| Captions: | Kaivan Kamali |
| Created: | 19 January 2022 |
| Materials: |
|
| Support: | |
| Supported Servers: | Known Working |