or subscribe with
Join 3,800+ readers for one email each week.
Digests » 41
this week's favorite
We present seven myths commonly believed to be true in machine learning research, circa Feb 2019. Also available on the ArXiv in pdf form.
What do you think of when you read the phrase ‘data science’? It’s probably some combination of keywords like statistics, machine learning, deep learning, and ‘sexiest job of the 21st century’. Or maybe it’s an image of a data scientist, sitting at her computer, putting together stunning visuals from well-run A/B tests. Either way, it’s glamorous, smart, and sophisticated. This is the narrative that data science has been selling since I entered the field almost ten years ago.
In late January, DeepMind broadcasted a demonstration of their StarCraft II agent AlphaStar. In Protoss v Protoss mirrors on a map used in pro play (Catalyst LE), it successfully beat two pro players from TeamLiquid, TLO (a Zerg player) and MaNa (a Protoss player).
A slight detour from the JPS problem first. I needed a way to write tests for stuff that can only be tested in-game, so I implemented a smoke test (I’m pretty sure professional testers will disagree with naming on this, but I’m a wild child). This just implements the BWEventListener, but instead of any logic, just tests methods. The first one I was suspicious about is the Cartography.isOnTheMap. In theory, this should give me back same as bw.getBWMap.isValidPosition(). Let’s check that!
I love how the modern Web and free software are like Lego blocks you can combine in myriad ways to create new and interesting things. While working on a recent analysis ofclimate policy leadership records for 2020 presidential candidates, I decided to try mashing together a few cool technologies I’d worked with to see if I could create a simple pipeline from my preferred open source data visualization library, Hadley Wickham’s ggplot2, to interactive data visualizations people could explore right inside their Twitter timelines.