BCCR Training Programme

Dear BCCR colleagues,

We are pleased to welcome you to the first edition of the BCCR Training programme in Machine Learning!

What is this about? What is a neural network? What does it mean when it “learns” something? Machine learning and deep learning are playing an increasingly growing role in everyday life, in the workplace, and academia. This new BCCR training programme is designed to give you the tools to understand these technologies and to get hands-on experience. We will also provide the tools to start your own machine learning framework to use it in your research. The programme will be a mix of lectures, practical activities, and project work, where you will build both understanding and confidence. It builds on last year’s successful student ML evening course. All BCCR researchers are welcome! —

The programme will consist of different learning blocks that can be done independently depending on time availability and previous experience. If you have no or very little experience in machine learning, we recommend the beginner’s course, which consists of the first four blocks. The last three blocks require knowledge in machine learning and neural networks that can be acquired with the beginner’s course.

Beginner’s course

Introduction to Machine Learning

When: Thursday April 9th - 12:30 - 15:30
Where: BCCR Seminar Room 4020 (4th floor East wing)

What:
In this block you will learn about the basic concepts of machine learning including:

• Supervised and unsupervised machine learning
• Classification and regression
• Several common machine learning methods
• scikit-learn
• Parameters and hyperparameters

Introduction to Deep Learning and Neural Networks

When: Thursday April 16th - 12:30 - 15:30
Where: BCCR Seminar Room 4020 (4th floor East wing)

What:
In this block you will learn about the basic concepts of deep learning and neural networks including:

• What is a neural network, a neuron, a layer
• How to train a neural network
• Basic types of neural networks: multilayer perceptron and convolutional network
• PyTorch

Ethical considerations and projects

When: Thursday April 30th - 12:30 - 15:30
Where: BCCR Seminar Room 4020 (4th floor East wing)

What:
This session is split into two parts. In the first part, we will discuss some ethical considerations regarding machine learning. The second part will focus on a group project to apply knowledge and skills learned in blocks 1 and 2.

Advanced ML products and projects

When: Tuesday May 12th - 12:30 - 15:30
Where: BCCR Seminar Room 4020 (4th floor East wing)

What:
This session is also split into two parts. In the first part, you will learn about some advanced machine learning applications, like large language models or weather forecasting. The second part will focus on the group project started in block 3.

BCCR Monday Seminar: Local ML Research

When: Monday June 1st - 11:00 - 12:30
Where: BCCR Seminar Room 4020 (4th floor East wing)

What:
There are already many local researchers who use machine learning and deep learning in their research at BCCR. This session will take the form of a Monday Seminar, and will showcase different examples of ML applications conducted by local colleagues. Presenters and themes will be announced later.

Advanced training

Set up your own ML framework

When: Tuesday June 16th - 12:30 - 15:30
Where: BCCR Seminar Room 4020 (4th floor East wing)

What:
In this block you will learn how to develop your own machine learning framework for your own research. This will include:

• How to set up your directories and files for a clean framework
• How to use pytorch-lightning for cleaner coding
• How to use Weights and Biases to track your experiments.

Introduction to Hyperparameter Optimisation

When: Thursday June 18th - 12:30 - 14:00
Where: BCCR Seminar Room 4020 (4th floor East wing)

What:
It is often hard to choose the best hyperparameters (learning rate, batch size, …) for your neural network or other machine learning model. In this block you will learn how to automatically select the best hyperparameters for your neural network. This will include:

• Description of some methods for finding the best hyperparameters
• Practical aspect.

Introduction to Explainable AI

When: Thursday June 18th - 14:00 - 15:30
Where: BCCR Seminar Room 4020 (4th floor East wing)

What:
Neural networks can have great predictive abilities but they are like black boxes, and it is hard to know why they make their decisions. In this block you will learn about different techniques that allow you to explain neural network decisions. These are part of the field of Explainable AI.