Do you like Cards Against Humanity or messed-up humor? Try out my browser game AI Against Humanity. All the cards you see were generated by a neural network. You’re also competing against another neural network! But beware: It contains some extremely strong and offensive language 😇
Today, let’s talk about how to learn a complex computer vision problem right in the browser! Thanks to TensorFlow’s JavaScript Library, this is easy to realize. Instead of training our own model and serving it as a finished product, we will let the user collect their own data and then train the model right there, on the client machine. Absolutely no server is neccessary! (Updated 29 JAN 2019)
»
Today I’m not going to talk about a funny use of Machine Learning or a cool application you can try at home. Instead, I want to talk about a project I’ve been working on professionally over the last few months. The job involved a complex NLP pipeline, an interesting application, and lots of new experiences for me. The knowledge I gathered helped me get in the top 4% of the Google Jigsaw Toxic Comments on Kaggle, and maybe it can also help you with your projects in the future.
Since ancient times, it has been known that machines excel at math while humans are pretty good at detecting cats in pictures. But with the advent of deep learning, the boundaries have started to blur... (Updated 19 DEC 2019)
»
Let’s teach a neural network to write Bible-like texts! If you have not heard about Andrej Karpathy’s incredibly awesome blog post about Recurrent Neural Networks, I recommend you drop everything right now and head over there for a great read! All of this here is based on his work and provided code.