F
Florian Laws
Researcher at University of Stuttgart
Publications - 15
Citations - 616
Florian Laws is an academic researcher from University of Stuttgart. The author has contributed to research in topics: Information extraction & Active learning. The author has an hindex of 11, co-authored 15 publications receiving 535 citations.
Papers
More filters
Proceedings ArticleDOI
Estimation of Conditional Probabilities With Decision Trees and an Application to Fine-Grained POS Tagging
Helmut Schmid,Florian Laws +1 more
TL;DR: A HMM part-of-speech tagging method which is particularly suited for POS tagsets with a large number of fine-grained tags based on splitting of the POS tags into attribute vectors and decomposition of the contextual POS probabilities of the HMM into a product of attribute probabilities.
Proceedings ArticleDOI
Stopping Criteria for Active Learning of Named Entity Recognition
Florian Laws,Hinrich Schütze +1 more
TL;DR: Three different stopping criteria for active learning of named entity recognition (NER) are investigated and it is shown that one of them, gradient-based stopping, reliably stops active learning, achieves nearoptimal NER performance and needs only about 20% as much training data as exhaustive labeling.
Proceedings Article
Active Learning with Amazon Mechanical Turk
TL;DR: The utility of active learning in crowdsourcing is evaluated on two tasks, named entity recognition and sentiment detection, and it is shown that active learning outperforms random selection of annotation examples in a noisy crowdsourcing scenario.
Proceedings ArticleDOI
CloudScan - A Configuration-Free Invoice Analysis System Using Recurrent Neural Networks
TL;DR: CloudScan as mentioned in this paper uses a recurrent neural network model that can capture long range context and compare it to a baseline logistic regression model corresponding to the current CloudScan production system, achieving 0.891 and 0.887 average F1 scores respectively on seen invoice layouts.
Proceedings ArticleDOI
Attend, Copy, Parse End-to-end Information Extraction from Documents
TL;DR: This paper proposes the Attend, Copy, Parse architecture, a deep neural network model that can be trained directly on end-to-end data, bypassing the need for word-level labels, and evaluates the proposed architecture on a large diverse set of invoices, which outperform a state-of-the-art production system based on word classification.