APCNN: Tackling Cclass imbalance in relation extraction through aggregated piecewise convolutional neural networks
- Smirnova, Alisa University of Fribourg
- Audiffren, Julien ENS Paris Saclay
- Cudré-Mauroux, Philippe University of Fribourg
-
01.06.2019
Published in:
- 2019 6th Swiss Conference on Data Science (SDS). - 2019, p. 63–68
convolutional neural nets
learning (artificial intelligence)
text analysis
word processing
APCNN
class imbalance
relation extraction
aggregated piecewise convolutional neural networks
distant supervision
multiinstance variant
oversampling techniques
Convolutional neural networks
English
One of the major difficulties in applying distant supervision to relation extraction is class imbalance, as the distribution of relations appearing in text is heavily skewed. This is particularly damaging for the multi-instance variant of relation extraction. In this work, we introduce a new model called Aggregated Piecewise Convolutional Neural Networks, or APCNN, to address this problem. APCNN relies on the combination of two neural networks, a novel objective function as well as oversampling techniques to tackle class imbalance. We empirically compare APCNN to state-of-the-art approaches and show that it outperforms previous multi-instance approaches on two standard datasets.
- Faculty
- Faculté des sciences et de médecine
- Department
- Département d'Informatique
- Language
-
- English
- Classification
- Applied sciences
- License
- License undefined
- Identifiers
-
- RERO DOC 327157
- DOI 10.1109/SDS.2019.000-6
- Persistent URL
- https://folia.unifr.ch/unifr/documents/308052
Statistics
Document views: 41
File downloads:
- cud_tci.pdf: 118