Over the last week, I have continued to develop my Python and machine learning skills while refining the direction of my research. My project is now beginning to focus on the mechanistic interpretability of machine learning models applied to electrical circuits.

After building my first machine learning model last week, my supervisor and I discussed how best to move the project forward. Initially, it was difficult to decide which direction to take, so we explored two possible approaches. The first was to continue with my existing model while using covariance analysis to investigate the behaviour of the neural network. The second was to build a transistor-based model to explore how a neural network learns from non-continuous voltage behaviour.

Following some initial research and further discussions with my supervisor, we decided to build a new, simpler circuit model based solely on resistors. Since resistor circuits have well-defined and predictable behaviour, they provide a better foundation for investigating what is happening inside the model's "black box" and understanding what the network is actually learning.

One of the biggest challenges so far has been narrowing the scope of the project. While exploring different ideas has been valuable, deciding on a clear research direction has been more difficult than I expected. However, these discussions have helped shape a more focused research question and given me a stronger foundation for the next stage of the projec