This week, we were introduced to machine learning and neural networks. While reading about them in my weekly reading, I found the concepts quite complicated and difficult to understand at first. However, Diana’s explanation in class helped me understand the basics and made them feel more approachable.

A part of the workshops I really enjoyed was looking at the maths behind machine learning and neural networks, even though it was quite basic. As I studied maths in my A-levels, and it’s a subject I really miss doing.

When we began coding our own neural networks, my model’s accuracy was much lower than that of most of my classmates, even after implementing ResNet18. I’m not sure why this was happening to me and a couple of others, but hopefully, we’ll figure this out together next week.

This week’s lecture was from Helen Bear, a Machine Learning Engineer. It was interesting to hear about the real world applications, processes, and difficulties of machine learning. None of her examples were really artistic, but listening to her process made me start to think about how machine learning could be applied to a creative task.

To end the week, Diana tasked us to use our own data to train our neural networks. I was assigned to take photos of seating, while my other classmates took pictures of windows and trees. It was really exciting and intriguing to collect our own data. I was surprised that we only needed to take 15 pictures each, meaning our whole dataset would only have about 90 photos, as I thought you would need much more data to have an accurate network. Still, I’m looking forward to seeing how it works out.