Lunch and Learn talks
Our lunch and learn talks are shorter seminars and lessons on best practices, research software engineering and data science.
You can watch recordings of all our lunch and learn talks on our YouTube playlist, or by searching for an individual talk below.
Play all lunch and learns on YouTube
If you’d like to volunteer to give a lunch and learn talk, please contact us.
Talk archive
-
Intro to the Theory of Neural Networks
- AI
- Neural Networks
- Python
- pytorch
In this lunch and learn talk, Huw Days introduces the theory of neural networks for deep learning. You'll understand how you can construct neural networks in the form of multi-layer perceptrons with non-linear activation functions. After that you'll look at how to use gradient descent and back propagation to updates the weights of your multi-layer perceptron to have your neural network learn to classify data points.
Speaker: Huw Day
Date:
-
Practical PyTest
- Python
- PyTest
- Software Engineering
In this lunch and learn talk, Richard Lane offers a practical introduction to testing, aimed at developers who are unsure about its value. It explains why testing matters, how it can improve confidence in your code, and hopefully makes a convincing case for adopting testing as part of your workflow.
Speaker: Richard Lane
Date:
-
Reproducible Data Science Projects
- Reproducibility
- Projects
- Software Engineering
In this lunch and learn talk, Léo Gorman is going to give a whistle-stop tour of some of the things you might want to consider when setting up a new project. There are many tools and philosophies around the best way to structure a data-science project, but often it boils down to a few key concepts: 1. Keep track of your data, and what you do to it 2. Make it easy for other people to download/run your code (environments) 3. Make it easy for people to understand what your code is doing (documentation) 4. Make sure if you change something and it breaks your code, that it's easy to figure out what's wrong (testing) It will not go into these topics in detail, but they key is to give you a flavour of the types of thing which are out there that you might want to consider.
Speaker: Léo Gorman
Date:
-
Classification with scikit-learn: Decision Trees and Random Forests
- Python
- Machine learning
- scikit-learn
Watch a demonstrate of how to build and interpret classification models with Python and scikit-learn. This talk covers the fundamentals of decision trees and random forest models — an ensemble technique that combines multiple decision trees trained on random subsets of data.
Speaker: Yujie Dai
Date:
-
Can others reproduce your conda results? Using 'pixi' in a Python data analysis project
- Python
- Reproducibility
- conda
Reproducible programming environments are crucial for ensuring consistent and reliable software development. They enable developers to recreate exact configurations, facilitating collaboration and long-term maintenance of projects. Watch a demonstration of pixi (a new conda environment management tool for Python) and how as a data scientist, research software engineer or scientific coder, you might take practical steps to incorporate this into your project. This talk will be of interest to people that use conda to install packages for their Python project.
Speaker: James Thomas
Date:
-
Can others reproduce your results? Using 'uv' in a Python data analysis project
- Python
- Reproducibility
Reproducible programming environments are crucial for ensuring consistent and reliable software development. They enable developers to recreate exact configurations, facilitating collaboration and long-term maintenance of projects. Watch a demonstration of uv (a new environment management tool for Python) and how as a data scientist, research software engineer or scientific coder, you might take practical steps to incorporate this into your project. This talk will be of interest to people that use pip or conda to install packages for their Python project.
Speaker: James Thomas
Date:
-
Introduction to git and GitHub
- Version control
- Reproducibility
- git
Git is a version control system that allows you to save multiple versions of a file or directory. This is useful to allow you to keep a record of all changes made to a file, and to move backwards and forwards in time through different versions of your file. GitHub uses Git providing a web-based interface for managing Git repositories and allowing multiple users to work on the same project simultaneously.
Speaker: Pau Erola
Date:
-
Pre-registration & registered reports
- Reproducibility
Gain a clear understanding of preregistration and registered reports. Learn how to implement them in your research and take a step towards more transparent and reproducible research!
Speaker: Prasad Sutar
Date: