I am Hannah. When I was a kid my favorite toy was the sieve. Now I find patterns in data. This is my sandpit ποΈ for models and bots I built for fun. My latest project is a twitterbot which tweets daily parts of Kafkas diary β¨. My background is in psychology, linguistics and neuroscience π§ , hence, finding structures in minds and brains.
I love language, reading and writing and I like to play around with AI for Natural Language Generation in German. If you wanna have a look: I have trained a plain Tensorflow/Keras net (Long Short-Term Memory Recurrent Neural Net) from scratch with a novel, which, magically, happens to be able to generate lyrics in the style of German Dadaists, too π And I have fine-tuned a pretrained GPT-2 to do the same which actually writes paragraphs of the novel in the style of my favorite German author π¬
Also you'll find a pretty accurate LSTM-RNN text classifier for obfuscated sentences. Finally, I present a more straightforward predictive ML classifier trained on open data, which is able to anticipate whether a vehicle is likely going to pass or fail a vehicle emissions test in the future. This project was a recent collaboration with some great fellow data scientists π― from WiMLDS.
π« Want to chat? Collaborate? Let's talk! -->>
Twitter: @HannahBohle
LinkedIn: hannahbohle