The semester has been rolling along so quickly that I haven't found the time to log my experiences officially, if this even counts as officially.
Computer Science 1 has been a good experience, and is mostly review other than some of the lower level memory allocation stuff that C entails. It's been great to learn these concepts from a different point of view. At Berkeley I was doing mostly projects in python and Java. It's interesting to learn about these same concepts in C, and the pointers and dynamic memory make many of the concepts a bit more intuitive, but also more tedious. I do admit that although the class is engaging, I'm itching for more of a challenge and find myself quickly working through homework problems. Unfortunately, the CS department at UCF is based around a foundation exam that pulls directly from CS1 content; and although these Berkeley classes covered similar topics, they didn't transfer for CS1.
In my Linear Regression Analysis class, I've been working with a NYC Airbnb listing dataset for the entire semester. I've learned how to use the R programming language further after being introduced to it last semester in Nonparametric Statistics. Each assignment in the class is a further analysis of our dataset. I've went through the data cleaning processes, transformed variables, normalized residuals, and I'm finally working with creating a multivariate linear regression model using price as the response variable. It is great to really get to know the ins and outs of the dataset and learn the process of creating a professional statistical report.
I joined the ucfAI club this year and started attending their data science meetings. I've been leading a small team as hope to use deep learning to categorize art by it's movement (e.g. Impressionism, Surrealism, Realism). At the moment we are training a fairly simple convolutional neural net on three categories of art movements. This net, with the help of some learning rate scheduling, is reaching around 46% validation accuracy. We hope to raise this accuracy by exploring the use of a more complex model, or even transfer learning from a pre-trained net. This paper is a similar to our endeavor, and their final accuracy was 62% using transfer learning on a residual neural net. I look forward to continuing this project over the rest of the semester!
I'm also taking a security in computing class that is mostly cryptography and cypher math. I also read a great book called The Cuckoo's Egg, about Cliff Stoll tracking and eventually catching a hacker. It was coincidentally set in Berkeley, where Stoll was working as systems administrator for the Lawrence Berkeley Astronomy Labs. This class is pretty straightforward, and while cybersecurity is interesting to me, I feel as if this is just the basics. I learned some really neat Euler Gradus function math that can also be applied to set theory with musical keys. It was a really nice aha moment to connect the two.
And lastly, I am taking Computer Design and Architecture. It is, without a doubt, interesting to learn what is happening at the lowest levels of computing. As I type, I imagine many temporary registers are being filled and arithmetic logic units firing. I'm looking forward to a large project where we are going to be building a MIPS compiler using C code.
That's my semester so far. I have a phone interview with a big one on Tuesday, and I'm super excited!