Search Results for author: Artem Chernodub

Found 9 papers, 4 papers with code

RNN Embeddings for Identifying Difficult to Understand Medical Words

1 code implementation WS 2019 Hanna Pylieva, Artem Chernodub, Natalia Grabar, Thierry Hamon

We introduce novel embeddings received from RNN - FrnnMUTE (French RNN Medical Understandability Text Embeddings) which allow to reach up to 87. 0 F1 score in identification of difficult words.

Word Embeddings

TARGER: Neural Argument Mining at Your Fingertips

1 code implementation ACL 2019 Artem Chernodub, Oleksiy Oliynyk, Philipp Heidenreich, Alex Bondarenko, Matthias Hagen, Chris Biemann, Alex Panchenko, er

We present TARGER, an open source neural argument mining framework for tagging arguments in free input texts and for keyword-based retrieval of arguments from an argument-tagged web-scale corpus.

Argument Mining

Sampling-based Gradient Regularization for Capturing Long-Term Dependencies in Recurrent Neural Networks

no code implementations24 Jun 2016 Artem Chernodub, Dimitri Nowicki

In this paper we propose a novel universal technique that makes the norm of the gradient stay in the suitable range.

Norm-preserving Orthogonal Permutation Linear Unit Activation Functions (OPLU)

no code implementations8 Apr 2016 Artem Chernodub, Dimitri Nowicki

We propose a novel activation function that implements piece-wise orthogonal non-linear mappings based on permutations.

Neurocontrol methods review

no code implementations17 Nov 2015 Artem Chernodub, Dmitry Dziuba

Methods of applying neural networks to control plants are considered.

Cannot find the paper you are looking for? You can Submit a new open access paper.