Advanced Machine Learning

Programming & Embedded Systems

Skill owner: Dr Dorian Florescu

Purpose: When neural networks (NNs) were first introduced many decades ago, the original aim was to mimic how the brain learns. Meanwhile, thanks to discoveries in neuroscience, we learned that biological neural networks operate in a quite different way. Nevertheless, the old model of the neural network, now known as an artificial neural network, has become a very powerful prediction tool in ML, being used in modern high impact applications such as driverless cars and natural language processing (Chat GPT). By mastering this skill, you will understand what a neural network is, how learning happens and how it is similar to biological learning. You will be able to find which ML problems are suitable for NNs, and how to evaluate their prediction performance. Lastly, you will be able to train NNs yourself and use them to solve a range of ML problems.

Requirements: To claim this skill, you will use Python or MATLAB to train and validate a neural network (NN) on a chosen dataset. You should show:

  1. A brief description of the aims of the study, and the choice of the particular ML model (50-100 words).
  2. A brief explanation (50-100 words) of what makes the chosen dataset suitable for supervised learning with a NN.
  3. Screenshots of the training process clearly marked, accompanied by a brief description of the results (50-100 words). The training should test out different NN hyperparameters on various ranges, and the description should provide relevant error metrics in each case.
  4. A concise discussion (100-250 words) will also be included commenting on the performance achieved, the model's generalization capability, and what further analyses /improvements could be made to the chosen approach.

Skills Framework Levels

The purpose of this Skills Framework is to evidence your acquisition of important engineering skills. Thus, it combines a combination of technical and transferable skills in eleven broad categories. It is designed so that the final portfolio can be used as a showcase of your skills attainment.

To reflect the differing levels of skills attainment, each of the skills in this framework can be demonstrated at three different levels:
Knowledge: Achieved when you follow instructions to demonstrate the skill.
Application: Achieved when you demonstrate the skill at request without instructions, and you have reflected on the skill's success. A critical reflection demonstrates your understanding of the skill by highlighting what went well and what could be improved (50-100 words).
Synthesis: Achieved when you have demonstrated the skill without guidance or instruction for a specific project, justifying your choice of using the skill and you have reflected on the skill's success (see above). Your justification should include an explanation of why you have used the skill to contribute towards a defined objective for a whole system / project. The choice of the skill must be supported by evidence showing that it is the best solution compared to other options. This allows you to demonstrate your understanding of when this skill is appropriate and how it fits within a wider context (50-100 words). There is also an expectation that you are demonstrating this skill effectively to achieve synthesis.

Each skill will have specific requirements for the skill to be satisfactorily endorsed. In addition, there are overarching requirements for all pages of your ePortfolio. If any page does not satisfy these requirements, the page will not be considered in any further detail.

  1. The page comprises an introduction that summarises all the digital artefacts on the page.
  2. Every digital artefact (e.g. photos, figures, videos and other non-text items) must clearly show the username of everyone contributing to the work in such a way as to authenticate the intellectual ownership of the artefact.
  3. All the skills being claimed are arranged below a 'Skills Mapping' block at the bottom of the page, as per the template.
    4. n
  4. Within the annotation block you have clearly specified what skill level you are claiming and why.</li>
  5. The page is of sufficient quality to present to people external to the University.
  6. The page has fewer than 5 errors such as spelling mistakes or other typographical errors.
  7. Any evidence in audio format must be recorded in a quiet, stable environment, with clear speech at a moderate pace, to ensure clarity and comprehensibility.
    8. </ol> # Labs Demonstrating This Skill

Citations