Pinned

1. language-recognition language-recognition Public

   This project shows how to train a language-recognizer from scratch that is able to distinguish between German and English, robustly.

   Jupyter Notebook 11 6

2. dense-hmm dense-hmm Public

   DenseHMM is a modification of Hidden Markov Models (HMMs) that allows to learn dense vector representations of both the hidden states and the observables via gradient-descent methods.

   Python 2 1

3. second-moment-loss second-moment-loss Public

   The second-moment loss (SML) is a novel training objective for dropout-based regression networks that yields improved uncertainty estimates.

   Jupyter Notebook 5 1