Welcome to the Statistical Natural Language Processing Group at the Institute for Computational Linguistics at Heidelberg University. Our research is on the intersection of machine learning and natural language processing, with a special focus on interactive statistical learning techniques. For example, we work on interactive neural machine translation and neural question answering systems, where an artificial intelligence agent learns from human reinforcement/bandit feedback.

We organize the weekly Statistical NLP Colloquium.

group photo
StatNLP group @ Botanical Gardens (next to our department)

Latest news

Tutorial on Statistical Methods for Reproducible Machine Learning at ICML and ECML

Stefan Riezler and Michael Hagmann are invited to present a tutorial on Statistical Methods for Reproducible Machine Learning at leading machine learning conferences: ICML and ECML.

New publications at ACL 2022

Two papers from the StatNLP group are accepted at ACL 2022:

  • “Sample, Translate, Recombine: Leveraging Audio Alignments for Data Augmentation in End-to-end Speech Translation” (arxiv)
  • “Analyzing Generalization of Vision and Language Navigation to Unseen Outdoor Areas” (arxiv)
Monograph "Validity, Reliability, and Significance: Empirical Methods for NLP and Data Science" published

The monograph is published in the Synthesis Lectures on Human Language Technologies series by Morgan & Claypool Publishers.More info

New publication at EMNLP 2021

New research from the StatNLP group on Don’t Search for a Search Method – Simple Heuristics Suffice for Adversarial Text Attacks will be presented at EMNLP 2021. Please access the paper on arxiv.

New publication at INTERSPEECH 2021

New research from the StatNLP group on On-the-Fly Aligned Data Augmentation for Sequence-to-Sequence ASR has been accepted at the INTERSPEECH 2021. The paper is available on arxiv.