Soft Rotation Equivariant Convolutional Neural Networks
| Autor: | Jose Costa Pereira, Jaime S. Cardoso, Eduardo Castro |
|---|---|
| Rok vydání: | 2020 |
| Předmět: |
Computer science
Orientation (computer vision) Generalization business.industry Feature extraction Pattern recognition 02 engineering and technology 010501 environmental sciences 01 natural sciences Convolutional neural network Regularization (mathematics) Convolution Test set 0202 electrical engineering electronic engineering information engineering Feature (machine learning) Equivariant map 020201 artificial intelligence & image processing Artificial intelligence business Rotation (mathematics) 0105 earth and related environmental sciences |
| Zdroj: | IJCNN Web of Science |
| DOI: | 10.1109/ijcnn48605.2020.9206640 |
| Popis: | A key to the generalization ability of Convolutional Neural Networks (CNNs) is the idea that patterns that appear in one region of the image have a high probability of appearing in other regions. This notion is also true for other spatial relationships, such as orientation. Motivated by the fact that in the early layers of CNNs distinct filters often encode for the same feature at different angles, we propose to incorporate the rotation equivariant prior in these models. In this work, different regularization strategies that capture the notion of approximate equivariance were designed and quantitatively evaluated in their ability to generate rotation-equivariant models and their effect on the model’s capacity to generalize to unseen data. Some of these strategies consistently lead to higher test set accuracies when compared to a baseline model, on classification tasks. We conclude that the rotation equivariance prior should be adopted in the general setting when modeling visual data. |
| Databáze: | OpenAIRE |
| Externí odkaz: |
|