Authors:
Luc Mioulet
1
;
G. Bideault
2
;
C. Chatelain
3
;
T. Paquet
2
and
S. Brunessaux
4
Affiliations:
1
Universite de Rouen and Airbus DS, France
;
2
Universite de Rouen, France
;
3
INSA Rouen, France
;
4
Airbus DS, France
Keyword(s):
Feature Combination, Recurrent Neural Network, Neural Network, Handwriting Recognition.
Related
Ontology
Subjects/Areas/Topics:
Applications
;
Artificial Intelligence
;
Biomedical Engineering
;
Biomedical Signal Processing
;
Classification
;
Computational Intelligence
;
Health Engineering and Technology Applications
;
Human-Computer Interaction
;
Methodologies and Methods
;
Multiclassifier Fusion
;
Neural Networks
;
Neurocomputing
;
Neurotechnology, Electronics and Informatics
;
Object Recognition
;
Pattern Recognition
;
Physiological Computing Systems
;
Sensor Networks
;
Signal Processing
;
Soft Computing
;
Software Engineering
;
Theory and Methods
Abstract:
In this paper we present several combination strategies using multiple BLSTM-CTC systems. Given several
feature sets our aim is to determine which strategies are the most relevant to improve on an isolated word
recognition task (the WR2 task of the ICDAR 2009 competition), using a BLSTM-CTC architecture. We
explore different combination levels: early integration (feature combination), mid level combination and late
fusion (output combinations). Our results show that several combinations outperform single feature BLSTM-CTCs.