SIBGRAPI Digital Library Archive
2023 accepted paper update form
The submission system will allow any upgrade until August 16th, 2023.
If you have any question, please contact:
thales.korting@inpe.br
Field Name
Field Value
Reference Type
Conference Paper (Conference Proceedings)
Work Title
(*)
(?)
Short Title
Don't forget to press the "+" button to enter the co-author data, if any.
v
1
st
Author
(*)
(?)
Affiliation (1
st
Author)
(*)
2
nd
Author
(*)
(?)
Affiliation (2
nd
Author)
(*)
3
rd
Author
(*)
(?)
Affiliation (3
rd
Author)
(*)
4
th
Author
(*)
(?)
Affiliation (4
th
Author)
(*)
5
th
Author
(*)
(?)
Affiliation (5
th
Author)
(*)
6
th
Author
(*)
(?)
Affiliation (6
th
Author)
(*)
EMailAddress
Keywords
(*)
(?)
Abstract
(*)
(?)
Image classification is one of the classical problems in computer vision, and CNNs (Convolutional Neural Networks) are widely used for this task. However, the choice of a CNN can vary depending on the chosen dataset. In this context, we have trainable activation functions that are crucial in CNNs and adapt to the data. One technique for constructing these functions is to write them as a linear combination of other activation functions, where the coefficients of this combination are learned during training. However, if we have a large number of activation functions to combine, the computational cost can be very high, and manually testing and choosing these functions may be impractical, depending on the number of available activation functions. To alleviate the difficulty of choosing which activation functions should be part of the linear combination, we propose two heuristics: Linear Combination Approximator by Coefficients (LCAC) and Major and Uniform Coefficient Extractor (MUCE). Our heuristics provide an efficient selection of a subset of activation functions so that their results are better or equivalent to the linear combination that uses all 34 available activation functions in our experiments (C34), considering the image classification problem. Compared to the C34 function, the LCAC function was better or equivalent in 62.5%, and the MUCE function in 87.5% of the conducted experiments.
Language
(*)
Portuguese
Spanish
English
File Name
(attach here your file)
(?)
e-Mail (login)
rogeriouff@yahoo.com.br
:: Powered by UR
Lib
Service ::