TY - GEN
T1 - Fully Mixed Max-Average Pooling
T2 - 6th International Conference on Intelligent Systems and Computer Vision, ISCV 2024
AU - Ait Skourt, Brahim
AU - Nikolov, Nikola S.
AU - Majda, Aicha
N1 - Publisher Copyright:
© 2024 IEEE.
PY - 2024
Y1 - 2024
N2 - As an integral part of a convolutional neural network, the pooling layers are responsible for the down-sampling operation which aims at preventing overfitting. Besides the conventional pooling methods (max and average), various methods that involve mixing max pooling and average pooling have been proposed. In this work, we propose a new mixed pooling method, called fully mixed max-average pooling (FMMAP), and evaluate its performance within a comparative study of various conventional and state-of-the-art pooling methods. FMMAP consists of fully mixing max pooling and average pooling features instead of stochastically selecting them as done in other popular mixed-pooling methods. The experimental results suggest that FMMAP outperforms the conventional pooling methods in accuracy, and while being very close (within 0.05%) to the other mixed-pooling methods accuracy-wise, it significantly outperforms them in terms of running time, being at least 1.7 times and up to 2.7 times faster.
AB - As an integral part of a convolutional neural network, the pooling layers are responsible for the down-sampling operation which aims at preventing overfitting. Besides the conventional pooling methods (max and average), various methods that involve mixing max pooling and average pooling have been proposed. In this work, we propose a new mixed pooling method, called fully mixed max-average pooling (FMMAP), and evaluate its performance within a comparative study of various conventional and state-of-the-art pooling methods. FMMAP consists of fully mixing max pooling and average pooling features instead of stochastically selecting them as done in other popular mixed-pooling methods. The experimental results suggest that FMMAP outperforms the conventional pooling methods in accuracy, and while being very close (within 0.05%) to the other mixed-pooling methods accuracy-wise, it significantly outperforms them in terms of running time, being at least 1.7 times and up to 2.7 times faster.
KW - Average Pooling
KW - Convolutional Neural Networks
KW - Max Pooling
KW - Mixed Pooling
UR - http://www.scopus.com/inward/record.url?scp=85203879417&partnerID=8YFLogxK
M3 - Conference contribution
AN - SCOPUS:85203879417
T3 - 2024 International Conference on Intelligent Systems and Computer Vision, ISCV 2024
BT - 2024 International Conference on Intelligent Systems and Computer Vision, ISCV 2024
A2 - Sabri, My Abdelouahed
A2 - Yahyaouy, Ali
A2 - el Fazazy, Khalid
A2 - Riffi, Jamal
A2 - Mahraz, Mohamed Adnane
PB - Institute of Electrical and Electronics Engineers Inc.
Y2 - 8 May 2024 through 10 May 2024
ER -