The universal recommender in Action(ML) 04/19/21 by Francesca Diana This article presents a new collaborative filtering technique. The universal recommender can be run as “engine” of ActionML's Harness. Read more Leave your thoughts
NER with little data? Transformers to the rescue! 12/14/20 by Thomas Timmermann See how to fine-tune a pre-trained BERT transformer on a custom NER task, beating standard bi-LSTM thanks to more learning transfer! Read more Leave your thoughts
Take control of named entity recognition with your own Keras model! 11/13/20 by Thomas Timmermann We build, train and evaluate a bidirectional LSTM-network for named entity recognition to extract information from legal texts with Keras. Read more Leave your thoughts
NER @ CLI: Custom-named entity recognition with spaCy in four lines 11/06/20 by Thomas Timmermann We show how to extract information from legal texts, training spaCy's named entity recognition model for our task on the command line. Read more Leave your thoughts
Why user-oriented development is so important – the story of tactics.ai 08/24/20 by Denis Stalz-John and Leonie Günther In this blog post, we want to give you an insight into the product development of tactics.ai. Our initial idea... Read more Leave your thoughts
Thinking AI means re-thinking data 05/28/20 by Marcel Mikl While doing AI is sexy and cool, data infrastructure is not considered any of this, even though machine learning applications rely on it. Read more Leave your thoughts
Evaluating machine learning models: The issue with test data sets 04/21/20 by Berthold Schulte By using a representative test set and various metrics, machine learning models can be evaluated and compared. Read more Leave your thoughts
From PDF data sheets to shared understanding with serverless SHACL 04/02/20 by Matthias Frank In this article we implement a generic approach to prepare formalized knowledge for a shared understanding by extracting the information contained in arbitrary PDF data sheets. Read more Leave your thoughts
Great Expectations: Validating datasets in machine learning pipelines 02/17/20 by Marcel Mikl Machine learning models don't care if an input dataset is correct, thus validating datasets beforehand is strongly advised. Read more Leave your thoughts
