Machine Learning was the fifth module for my Master of Science. I am very interested in how machine learning works, so this was the number one module I was looking forward to! I wasn't disappointed - I found the concepts behind machine learning and artificial intelligence fascinating, and we were very lucky to be taught by our knowledgeable and caring tutor, Dr. Oladapo.
Here are the main concepts I've learned.
I already studied exploratory data analysis, correlation and regression in previous modules, so I felt quite confident with those concepts and activities. We had a short introduction to data classification in the Visualising Data module and I wanted to know more, so I was very pleased to learn about data classification in more depth. I now feel very confident using several classification methods, particularly K-Nearest Neighbors and Random Forest.
I found the maths behind machine learning absolutely fascinating. Once I understood the concept of how error functions are used to reduce prediction error, I was able to understand the way this concept is used in particular areas of machine learning, such as regression and neural networks.
I was very curious to know how natural language processing works, so I found the material on how transformer-based architectures are used for this very interesting.
We are all becoming more aware of the impact of artificial intelligence in our world, so I was very pleased to take an in-depth look at the risks associated with artificial intelligence, particularly how it can reproduce prejudice from training data, and how it can invade privacy.
I found the final project on making Spotify recommenders using different approaches a challange, particularly using a deep learning approach. I definitely learned a lot, especially about setting up a deep neural network with TensorFlow Keras.
Our team project was to use data from Airbnb to identify room prices which are too low or too high considering the room type and location. Our team met using a WhatsApp group chat, and stayed on track using a "team progress" document created by one of our members. Once we divided up the work effectively, the project came together quite well. We were able to achieve our objectives, although we received very fair feedback from our tutor which highlighted areas which could have been improved, particularly justifying our choice of specific data processing and machine learning techniques.
I learned a lot from the team project experience. Several team members didn't begin to contribute until a few weeks after the project started, which I was initially impatient with, but other team members encouraged us all to be patient and understanding. I definitely learned to give more grace to my team members!
At the end of the project I received some very nice feedback from my team members.
"Paul - thank you for your complete dedication and attention to detail in ensuring that the work is thorough and completed to the highest possible standard. I truly appreciate your efforts, and I am sure the rest of the team does as well. We are fortunate to have you as part of our team."
(Team Project Member, Qatar)
"I think Paul would be a good choice to submit the final report, as he has taken a leading role throughout the project. I really enjoyed working with you all, and special thanks to Paul for coordinating the submission."
(Team Project Member, UAE)
"Paul, your dedication and consistency throughout the work has genuinely been impressive, especially the interactive web interface and the recent visual chart updates; the project is looking very polished and professional now. You've great leadership skills and I'm glad we worked as a team. Looking forward to collaborating in future."
(Team Project Member, Pakistan)
"Thank you, Paul for the leadership and everyone else for contributions!"
(Team Project Member, Czech Republic)
| SKILLS MATRIX (0 low, 10 high) | ||
|---|---|---|
| learning outcome | proficiency | interest |
| Learn about the key paradigms and algorithms in machine learning. | 7 | 8 |
| Get an understanding of data analytics based on machine learning and using modern programming tools, such as Python or R. | 8 | 9 |
| Experience how machine learning and data analytics can be used in real-world applications. | 8 | 8 |
| Acquire the ability to gather and synthesise information from multiple sources to aid in the systematic analysis of complex problems using machine learning tools and algorithms. | 8 | 9 |
| Articulate the legal, social, ethical, and professional issues faced by machine learning professionals. | 7 | 7 |
| Understand the applicability and challenges associated with different datasets for the use of machine learning algorithms. | 7 | 7 |
| Apply and critically appraise machine learning techniques to real-world problems, particularly where technical risk and uncertainty is involved. | 6 | 8 |
| Systematically develop and implement the skills required to be effective member of a development team in a virtual professional environment, adopting real-life perspectives on team roles and organisation. | 9 | 9 |
I was definitely interested in studying machine learning algorithms, and I certainly grasped the fundamental concepts, but I could become more proficient in specific areas which are more complicated, such as convolutional networks and graph neural networks.
I loved using the Pandas Python library and the R language, and I think I was particularly good at using R for exploratory data analysis and data processing. In the applications we've studied, it seems like Python is used more commonly, but I hope I can use R in my future career as a data scientist, as I prefer that language.
I've really enjoyed applying machine learning skills to real-world problems. I feel very confident with classical machine learning methods, but I would benefit from more practice applying deep learning methods to practical tasks, while maintaining a good understanding of the uncertainties involved.
I really liked the activities which involved bringing together data from multiple sources - I think I'm very good at processing and combining large datasets, particularly using R.
We've completed several assignments on the ethics of artificial intelligence in this module and other modules - I feel like I have a good overview of the relevant issues, such as reliability, accountability, and privacy.
I enjoyed working on our team project, and I'm proud of what we achieved together. I think I was an effective team member, and I was pleased to get positive feedback from my team members.
I'll soon start applying for jobs in data science, and depending on what the responsibilities of my job are, I can focus on particular required skills. If I need to fully understand the maths behind various machine learning models, I can study the module readings again, particularly the Deep Learning book by Bishop.
I feel very confident with the R programming language, but if my job requires Python I'd like to be able to become more comfortable using it without spending so much time checking help files and manuals!
Data science jobs today are requiring more proficiency and experience with artificial intelligence and deep learning. It was great to put my new artificial intelligence skills into action by making Spotify recommenders, but before I start applying for jobs, I'd like to spend more time practising applying deep learning approaches to practical projects, and handling the uncertainties involved with those approaches.
I'm very familiar with R functions for processing and combining data from different sources, and I also have experience with combining data sources using dashboards. I'd like to practise using the Python Pandas library for the same tasks.
When we studied the ethics of artificial intelligence, I was particularly interested in training large language models to respond with awareness of prejudice and injustice - I'd like to explore this field more.
When we were working on our team project, I learned to respond more sensitively to team members who were reluctant to contribute. I think I could continue to improve in this area, by showing more understanding and empathy to team members who need encouragement.
I was very pleased to work with and learn from Dr. Oladapo. In my future career as a data scientist I aim to be able to confidently use machine learning approaches in the way Dr. Oladapo has modelled for us. Thank you Dr. Oladapo for guiding as through this Machine Learning module!