Maximilian Du

Media Annotator

When reviewing video and audio media for my book, I often find myself needing a real-time annotator. So I coded one! It listens for your keystrokes anywhere on the screen, so you can be focused on your media. It uses system time, so it runs in lockstep with any media player. It exports your annotations to a simple text format that you can add to any literature review notes. And best of all, your annotations are attached to simple macros (number keys) and can be easily changed!

Code

MidiStyle: Audio Style Transfer

CS 229 Final Project

Style transfer is pretty pervasive in visual tasks, using anything from Gram Matrix methods to CycleGAN. Can we try using established vision style transfer algorithms on audio? In this project, we show that this is indeed possible. Using a spectrogram representation, we change a piano into harp, harpsichord, electric guitar, and even timpanis. We generate our own data using MIDI, and we test on a real-world piano.

Website   /   Code

Coding Basics

Literally every time I want to make a line plot in Matplotlib, or save a model in PyTorch, or load a csv, I find myself searching it up and copy/pasting. To unify all of these simple things, I'm working on a large repository of code basics for researchers. This includes Numpy, PyTorch, LaTeX, Pandas, fileloading, Matplotlib, Seaborn, and others. I'm also including easily runnable PyTorch implementations of common ML models.

Code

Thompson Sampling Simulator

CS 109 Winning Final Project

The Multi-Armed Bandit is a fascinating theoretical question, but it is also a compelling question of philosophical intensity: how do we balance exploration vs exploitation? We look at one algorithm, known as Thompson sampling. Here, we simulate ants finding a good location for a nest. We also implement Tandem Running, which allows ants to "persuade" other ants, resulting in faster convergence.

Simulation   /   Code

The Tricks, Goofs, and Whales of Machine Learning

Machine learning is the abstraction of real learning processes into computational ones. Therefore, we find that it often imitates nature. In this 90-minute talk, I explore this parallelism by looking at connections between robot-learning and animal training. How does classical conditioning map to the mathematical Bellman Backup? How do animal trainers help their animals achieve good exploration policies? What do animal superstitions tell us about domain adaptation?

This talk was originally made for Stanford Splash, but it works for any audience of high school or undergraduate college students. If you want the slide deck or a presentation, please email me.

Automatic Source Vetter

While doing research for my whale book, I found myself needing to keep track of current events and follow whale-related developments as they happen. I'm also a full-time student, which makes this highly impractical. So, I've scripted up a system that monitors a large collection of RSS newsfeeds for a list of keywords. If any such keywords are found, the relevant links are saved to a SQL database for later perusal. This script also looks at relevant YouTube channels and specialty sites. While you may not find any use in my specialty site translators, the codebase is highly modular, so you can implement your own translator. This also is useful for any CS researcher because arXiv has RSS feed links.

Code (still actively updating)