A Deeper Look at the Unsupervised Learning of Disentangled Representations in β-VAE from the Perspective of Core Object Recognition.
Harshvardhan SikkaPublished in: CoRR (2020)
Keyphrases
- unsupervised learning
- object recognition
- deep belief networks
- invariant representations
- dissociated dipoles
- computer vision
- image understanding
- natural images
- constrained search
- supervised learning
- object detection
- d objects
- image features
- object categorization
- object classification
- object description
- deep learning
- object representation
- feature selection
- template matching
- higher level
- biologically motivated
- visual object recognition
- expectation maximization
- object class
- feature matching
- visual recognition
- scene classification
- object categories
- image matching
- multiple representations
- image representation
- model selection
- semi supervised