Damiant's Capstone Blog

So, moving throughout this project over the last couple months I couldn’t have asked for a better team. The team has done a great job and pulled it’s weight. The most difficult challenge right now is trying to get the model to stop predicting with specific bias. Currently, the model is biasing electronic music over the other genres most of the time. There is some rock bias too, some genres are doing great when it comes to prediction, but other genres aren’t doing as good. For example, Hip hop and R&B often get confused with each other. I am still trying to work out how to work through this an hopefully will make the model better in the last quarter. I had some initial doubts due to such a busy schedule but I feel as though we are on the right track to completing in time. The most important things I have learned throughout this project is that team work makes things a whole lot easier, you don’t have to depend on doing everything on your own.

My favorite technology in this project has to be hands down Tenserflow/Keras. It’s ability to build and train complex deep learning models for image classification is incredibly powerful. Im suprised that models can learn patterns from raw data like spectrograms and achieve good accuracy. While it is great to work with sometimes it spits out errors I don’t understand. While im not working on the data preprocessing side, I would say that would be my least favorite part since it is time consuming

I was reading up Robert Martin’s book on Clean Code. He described many different scenarios regarding why clean code is so important. You don’t want to work on a codebase that is just all over the place. Imagine having to build a codebase from scratch, and you don’t even know what the code means. That would be an awful thing to be task with. I’ve been asked what is one thing that I would like to start doing more? Well, trying to style my code is what I’m going to be focusing on, and not just that making sure to implement code comments, so other people understand what the code means. I personally do always write comments but there not always the most descriptive so I need to work on that. I also have to start looking for bad code smells. Usually, if it works I leave it alone, which shouldn’t be the case.

Hello! to everyone reading. Well, the fall term has come to an end. This will be my last post till Winter term, which honestly isn’t that far away. Version 2 of our CNN was completed today and I think that we did a great job as a group. I ended up coding us a spectrogram classifier, so instead of relying only on MFCCs, we can also feed our neural network spectrogram images to learn. I can honestly say that this class has been a blast and a change of pace. I thoroughly enjoyed how my capstone course was as real as it would be on the job. I’m on a loaded schedule for the winter semester but that’s nothing new. Work is going great; I am always learning new things. Other than that there isn’t much else going on. See you guys Winter semester!

Hi,

My name is Damiant and my timezone is PST, california. My journey with OSU started in 2022 when I transfered over from a Community College. I got started with computers as a child, the first computer I ever used was a windows xp operating system, and second was linux based. What got me started with computers was trying to script for a game called runescape back in the day. So I loved it ever since. I work as an IT specialist and my favorite technologies are atlassian products, and several open source softwares. I can’t wait for exploration in UI/UX design and choosing my technologies.

Welcome to blogs.oregonstate.edu. This is your first post. Edit or delete it, then start blogging!