I've been able to get quite a bit done over the past week. With school out and most teachers lightening the workload for students, this whole quarantine thing is actually allowing me to focus my energy on more important things. I've been able to develop a pretty good blueprint for the neural network architecture I want to use for my final product. It actually ended up being a pretty standard fully connected network (FCN) with some recurrent neural network (RNN) characteristics. I also made a list of about twenty or so features that a given email has that (to me) are most indicative of whether it is intended for phishing or not. I'm very happy with the progress I've made so far, largely because I feel like I have so much more control over this project than last year's final product.
However, there are certainly many more obstacles to overcome between now and May. Firstly, I haven't actually started coding yet. I'm debating whether I should start out by implementing my neural network in a higher-level, more ML-specialized language like Octave, or just code for functionality and optimization in Python. Also, I still need to get some of the finer details of my network down, but I can't do that until I actually start coding. These next weeks will be interesting.Â