We investigate the use of Genetic Programming (GP) as a convolutional predictor for missing pixels in images. The training phase is performed by sweeping a sliding window over an image, where the pixels on the border represent the inputs of a GP tree. The output of the tree is taken as the predicted value for the central pixel. We consider two topologies for the sliding window, namely the Moore and the Von Neumann neighborhood. The best GP tree scoring the lowest prediction error over the training set is then used to predict the pixels in the test set. We experimentally assess our approach through two experiments. In the first one, we train a GP tree over a subset of 1000 complete images from the MNIST dataset. The results show that GP can learn the distribution of the pixels with respect to a simple baseline predictor, with no significant differences observed between the two neighborhoods. In the second experiment, we train a GP convolutional predictor on two degraded images, removing around 20% of their pixels. In this case, we observe that the Moore neighborhood works better, although the Von Neumann neighborhood allows for a larger training set.
Jakobovic, D., Manzoni, L., Mariot, L., Picek, S., Castelli, M. (2021). CoInGP: convolutional inpainting with genetic programming. In GECCO 2021 - Proceedings of the 2021 Genetic and Evolutionary Computation Conference (pp.795-803). Association for Computing Machinery, Inc [10.1145/3449639.3459346].
CoInGP: convolutional inpainting with genetic programming
Manzoni, Luca;Mariot, Luca;Castelli, Mauro
2021
Abstract
We investigate the use of Genetic Programming (GP) as a convolutional predictor for missing pixels in images. The training phase is performed by sweeping a sliding window over an image, where the pixels on the border represent the inputs of a GP tree. The output of the tree is taken as the predicted value for the central pixel. We consider two topologies for the sliding window, namely the Moore and the Von Neumann neighborhood. The best GP tree scoring the lowest prediction error over the training set is then used to predict the pixels in the test set. We experimentally assess our approach through two experiments. In the first one, we train a GP tree over a subset of 1000 complete images from the MNIST dataset. The results show that GP can learn the distribution of the pixels with respect to a simple baseline predictor, with no significant differences observed between the two neighborhoods. In the second experiment, we train a GP convolutional predictor on two degraded images, removing around 20% of their pixels. In this case, we observe that the Moore neighborhood works better, although the Von Neumann neighborhood allows for a larger training set.File | Dimensione | Formato | |
---|---|---|---|
Jacobovic-2021-GECCO-AAM.pdf
accesso aperto
Tipologia di allegato:
Author’s Accepted Manuscript, AAM (Post-print)
Licenza:
Altro
Dimensione
981.42 kB
Formato
Adobe PDF
|
981.42 kB | Adobe PDF | Visualizza/Apri |
Jacobovic-2021-GECCO-VoR.pdf
Solo gestori archivio
Tipologia di allegato:
Publisher’s Version (Version of Record, VoR)
Licenza:
Tutti i diritti riservati
Dimensione
716.44 kB
Formato
Adobe PDF
|
716.44 kB | Adobe PDF | Visualizza/Apri Richiedi una copia |
I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.