Skip to article frontmatterSkip to article content

Course outline

Download the slides here


Topics

Part 1 - Structure of the brain

  • Neurons
  • Synapses and networks
  • Anatomy of the brain

Part 1 is about the structure of the brain, how it’s composed of brain cells called neurons, connected via synapses, and divided into different regions. We’ll also talk about how these cells communicate and compute with these signals, and the models we use to try to understand this.

Part 2 – Learning in brain and machine

  • Learning rules
  • Training “spiking” neural networks

In part 2, we’ll focus specifically on learning. What we know about how learning happens in the brain, some models of this, and how those models might relate to learning in machines.

Part 3 – Theoretical approaches

  • Understanding neural networks
  • Various topics

Part 3 of the course is about some of the theories that have been proposed about how the brain puts all these mechanisms together. We’ll start from some methodological approaches to how you might go about understanding the brain. These also apply to understanding an artificial neural network. Then we’ll look at some theories of how the brain might be solving some particular sorts of tasks.

Part 4 – Future developments

  • Neuromorphic computing
  • Recent developments

In the final part, we’ll discuss prospects for the future. We’ll talk about neuromorphic computing, that is, specialised hardware designed to mimic some aspects of how the brain functions. We’ll also talk about some more recent developments in neuroscience that we still don’t quite know what to make of.

Weekly structure

This course uses a flipped classroom approach. That means rather than sitting down and listening to us talk for a couple of hours each week and then go off and try to understand the material on your own, we flip that:

  • Each week you watch a series of short videos that cover the essential material. You might also want to do some extra reading around these topics.

  • Then, in our scheduled class time you’ll work in small groups doing coding-based exercises, and we will be around to support you more interactively.

  • You can also continue after the scheduled time if you don’t finish or you want to take it further.

Unfortunately, we won’t be able to do interactive sessions online, but you will have access to all the exercises.

There will also be a discussion group on Teams if you’re at Imperial, or Discord otherwise.

Weekly Structure

Figure 1:N4ML Weekly Structure

Assessment

Finally, for those studying at Imperial this course will be assessed. There are 3 assessment points:

  • Coursework 1 - 40%

  • Coursework 2 - 40%

    • Groups of two
    • Coding-based
    • Python notebooks, e.g. Google Colab
    • Can claim costs of Google Colab Pro
    • Peer assessment
  • End of term quiz - 20%

    • Multiple choice

Firstly, there are two pieces of coursework, each worth 40% of the final grade. These will be done in groups of two and will be coding-based. We would encourage you to use Python notebooks.

The easiest way to do so is to use Google Colab which we will set up for you with zero installation required, and you can claim back the cost of a Google Colab Pro account for the term if you would like (although the free version is fine for this course).

Assessment will partly be done by peers, that is everyone will read and assess each others’ notebooks. The reasoning behind this is to encourage clear presentation of your work, and to let you learn from the different approaches taken by different groups.

We will make a detailed guide available when you are given the first coursework assignment.

The third assessment point is an online multiple choice quiz at the end of the term, worth 20%.