Nearest Neighbours Graph Variational AutoEncoder
Graphs are versatile structures for the representation of many real-world data. Deep Learning on graphs is currently able to solve a wide range of problems with excellent results. However, both the generation of graphs and the handling of large graphs still remain open challenges. This work aims to...
| Main Authors: | , , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
MDPI AG
2023-03-01
|
| Series: | Algorithms |
| Subjects: | |
| Online Access: | https://www.mdpi.com/1999-4893/16/3/143 |
| _version_ | 1827751774897831936 |
|---|---|
| author | Lorenzo Arsini Barbara Caccia Andrea Ciardiello Stefano Giagu Carlo Mancini Terracciano |
| author_facet | Lorenzo Arsini Barbara Caccia Andrea Ciardiello Stefano Giagu Carlo Mancini Terracciano |
| author_sort | Lorenzo Arsini |
| collection | DOAJ |
| description | Graphs are versatile structures for the representation of many real-world data. Deep Learning on graphs is currently able to solve a wide range of problems with excellent results. However, both the generation of graphs and the handling of large graphs still remain open challenges. This work aims to introduce techniques for generating large graphs and test the approach on a complex problem such as the calculation of dose distribution in oncological radiotherapy applications. To this end, we introduced a pooling technique (ReNN-Pool) capable of sampling nodes that are spatially uniform without computational requirements in both model training and inference. By construction, the ReNN-Pool also allows the definition of a symmetric un-pooling operation to recover the original dimensionality of the graphs. We also present a Variational AutoEncoder (VAE) for generating graphs, based on the defined pooling and un-pooling operations, which employs convolutional graph layers in both encoding and decoding phases. The performance of the model was tested on both the realistic use case of a cylindrical graph dataset for a radiotherapy application and the standard benchmark dataset sprite. Compared to other graph pooling techniques, ReNN-Pool proved to improve both performance and computational requirements. |
| first_indexed | 2024-03-11T07:02:52Z |
| format | Article |
| id | doaj.art-c2c9cb74de1545fca5d9cd0e30ad9792 |
| institution | Directory Open Access Journal |
| issn | 1999-4893 |
| language | English |
| last_indexed | 2024-03-11T07:02:52Z |
| publishDate | 2023-03-01 |
| publisher | MDPI AG |
| record_format | Article |
| series | Algorithms |
| spelling | doaj.art-c2c9cb74de1545fca5d9cd0e30ad97922023-11-17T09:09:08ZengMDPI AGAlgorithms1999-48932023-03-0116314310.3390/a16030143Nearest Neighbours Graph Variational AutoEncoderLorenzo Arsini0Barbara Caccia1Andrea Ciardiello2Stefano Giagu3Carlo Mancini Terracciano4Department of Physics, Sapienza University of Rome, 00185 Rome, ItalyIstituto Superiore di Sanità, 00161 Rome, ItalyINFN Section of Rome, 00185 Rome, ItalyDepartment of Physics, Sapienza University of Rome, 00185 Rome, ItalyDepartment of Physics, Sapienza University of Rome, 00185 Rome, ItalyGraphs are versatile structures for the representation of many real-world data. Deep Learning on graphs is currently able to solve a wide range of problems with excellent results. However, both the generation of graphs and the handling of large graphs still remain open challenges. This work aims to introduce techniques for generating large graphs and test the approach on a complex problem such as the calculation of dose distribution in oncological radiotherapy applications. To this end, we introduced a pooling technique (ReNN-Pool) capable of sampling nodes that are spatially uniform without computational requirements in both model training and inference. By construction, the ReNN-Pool also allows the definition of a symmetric un-pooling operation to recover the original dimensionality of the graphs. We also present a Variational AutoEncoder (VAE) for generating graphs, based on the defined pooling and un-pooling operations, which employs convolutional graph layers in both encoding and decoding phases. The performance of the model was tested on both the realistic use case of a cylindrical graph dataset for a radiotherapy application and the standard benchmark dataset sprite. Compared to other graph pooling techniques, ReNN-Pool proved to improve both performance and computational requirements.https://www.mdpi.com/1999-4893/16/3/143graph neural networkvariational autoencoderpoolingnearest neighbours |
| spellingShingle | Lorenzo Arsini Barbara Caccia Andrea Ciardiello Stefano Giagu Carlo Mancini Terracciano Nearest Neighbours Graph Variational AutoEncoder Algorithms graph neural network variational autoencoder pooling nearest neighbours |
| title | Nearest Neighbours Graph Variational AutoEncoder |
| title_full | Nearest Neighbours Graph Variational AutoEncoder |
| title_fullStr | Nearest Neighbours Graph Variational AutoEncoder |
| title_full_unstemmed | Nearest Neighbours Graph Variational AutoEncoder |
| title_short | Nearest Neighbours Graph Variational AutoEncoder |
| title_sort | nearest neighbours graph variational autoencoder |
| topic | graph neural network variational autoencoder pooling nearest neighbours |
| url | https://www.mdpi.com/1999-4893/16/3/143 |
| work_keys_str_mv | AT lorenzoarsini nearestneighboursgraphvariationalautoencoder AT barbaracaccia nearestneighboursgraphvariationalautoencoder AT andreaciardiello nearestneighboursgraphvariationalautoencoder AT stefanogiagu nearestneighboursgraphvariationalautoencoder AT carlomanciniterracciano nearestneighboursgraphvariationalautoencoder |