I really need to meet with my mentor. Due the incompatibility of our schedules, we have not been able to meet since discussing ideas for my final product. To top it all off, my mentor still hasn’t submitted the evaluation for my final product proposal, so I don’t have his opinion on the viability/feasibility of my ideas. Personally, I think that the only feasible idea of the three that I submitted is creating a neural network to pre-screen phishing emails. I can already imagine the steps I would take to complete such a product. The main problem is that this has already been done before. As for my other ideas, I have already determined that they are either far too ambitious or too vague.
Either way, I’m pretty certain that I’ll be doing something with ML. I have been re-familiarizing myself with Python and Tensorflow in preparation for this. Getting back into machine learning has been very enjoyable for me, and I’m already thinking up recreational projects (and maybe even continuing last year’s project). Determining which type of neural network to use for phishing emails, though, might be a little difficult. Current iterations of my idea have a very high true positive rate, so raising its accuracy any further will require deep analysis and fine-tuned tweaking of the neural network architecture.