Deep Diffusion Autoencoders
- Resource Type
- Conference
- Authors
- Dorado, Sara; Fernandez, Angela; Dorronsoro, Jose R.
- Source
- 2019 International Joint Conference on Neural Networks (IJCNN) Neural Networks (IJCNN), 2019 International Joint Conference on. :1-8 Jul, 2019
- Subject
- Computing and Processing
Power, Energy and Industry Applications
Robotics and Control Systems
Signal Processing and Analysis
Decoding
Eigenvalues and eigenfunctions
Kernel
Encoding
Markov processes
Standards
Symmetric matrices
- Language
- ISSN
- 2161-4407
Extending work by Mishne et al. [1], we propose Deep Diffusion Autoencoders (DDA) that learn an encoder-decoder map using a composite loss function that simultaneously minimizes the reconstruction error at the output layer and the distance to a Diffusion Map embedding in the bottleneck layer. These DDA are thus able to reconstruct new patterns from points in the embedding space in a way that preserves the geometry of the sample and, as a consequence, our experiments show that they may provide a powerful tool for data augmentation.