Let's talk about the "Interdisciplinary Approach to Machine Learning" module
Let's talk about the "Interdisciplinary Approaches to Machine Learning" module
Having the liberty to choose optional modules is one of the biggest and most exciting perks of being a Master's student at the Centre of Interdisciplinary Methodology department. In this blog post, I will be sharing my reason for choosing the Interdisciplinary Approaches to Machine Learning module in the Big Data and Digital Futures Masters program. I will also share what I have learnt and my experience putting those learnings into practice in my assessments.
Having taken a couple of courses on Data and AI in the past, I was shockingly left unsatisfied. There was a gaping hole in the knowledge gained. I really never understood the "why" or the theory behind certain practices. I could replicate it, but it was as though I could not critically engage with the nuances of transferring the knowledge in other applications. So when I saw the Interdisciplinary Approaches to Machine Learning optional module, I was intrigued. Prior to this time, I had never really thought of Machine Learning from a non-technical perspective (specifically Computer Science and Statistics). The course outline was very interesting, mentioning the historical and social context of Machine Learning, and this was it. Voila!!
Here's a little spoiler: The module focuses on Deep Learning.
One of the most outstanding things about this module is that it starts off with the history of Artificial Intelligence and Machine Learning, which I found very interesting. It was pretty amazing to find out that most of the AI practices used today were discovered quite a long time ago, like in the 20th century. Learning about the research that paved the way for the amazing tools we use today felt really good 😃. Also, the combination of studies from other disciplines creates an excellent and complete foundational knowledge of how advancements in other fields contribute to the advancements in machine learning. Furthermore, the module then moves from relatively old but useful deep learning models like the vanilla neural network to the more recent models like the Transformers architecture that disrupted the Natural Language Processing (NLP) subfield of linguistics and AI.
"In the Interdisciplinary Approach to Machine Learning Module, we learn the history of AI up till up-to-date state of the art practices in machine learning."
The literature covered in the Interdisciplinary Approach to Machine Learning module is rich and relevant to current machine learning practices in the industry. We take a deep dive into the workings of models such as Convolutional Neural Networks (CNNs), Recurrent Neural Networks (RNNs), and Long-Short Term Memory (LSTM) Neural Networks through the literature. Also, the module structure includes weekly coding labs to implement and get a feel of each model type. Moreover, current innovations in AI like the ChatGPT and the Creative AI models powering platforms like DALL-E are broken down to study the components underpinning how they work, what makes them so good and their limitations.
Student name: Ayokunle Adeniyi
Subject area: PGT in Big Data and Digital Futures
What's learning without assessments?!!! The Interdisciplinary Approach to Machine Learning module has multiple assessments of 2 types. There were the presentation assessment and written assessments. The presentation assessment is group based, and the group is expected to show a critical understanding of any topic of choice within the module and express this deep understanding or argument orally with PowerPoint slides as a guide. For the written assessments, there were 2: the first one was an experiment and a report that helped me get a feel of practical deep learning while making use of the academic literature. The second one gave me the freedom to pursue my interests and passion, where it could be more conceptual, giving me the opportunity to have an argument (a position) on any aspect of machine learning, or it could be empirical, where you build a model that does something interesting.
Fun fact: My presentation was on Attention, Self-Attention and Multi-Head Attention in Deep Learning
Overall, it has been an exciting and engaging learning experience. The Interdisciplinary Approach to Machine Learning module has provided me with a perfect balance between theory (from multiple disciplines) and practice of machine learning. I got to engage deeply with literature in the field in ways that I can apply what I have learnt in other fields. I also gained hands-on experience by building deep learning models and hyperparameter optimisation through my assessments. I can say that the module provides a solid foundation for contemporary machine learning and shows an awareness of current industry practices in machine learning.