Capsule Networks -- A Probabilistic Perspective
arXiv:2004.03553
Abstract
'Capsule' models try to explicitly represent the poses of objects, enforcing a linear relationship between an object's pose and that of its constituent parts. This modelling assumption should lead to robustness to viewpoint changes since the sub-object/super-object relationships are invariant to the poses of the object. We describe a probabilistic generative model which encodes such capsule assumptions, clearly separating the generative parts of the model from the inference mechanisms. With a variational bound we explore the properties of the generative model independently of the approximate inference scheme, and gain insights into failures of the capsule assumptions and inference amortisation. We experimentally demonstrate the applicability of our unified objective, and demonstrate the use of test time optimisation to solve problems inherent to amortised inference in our model.
References in corpus (7)
- Categorical Reparameterization with Gumbel-Softmax
- Stochastic Backpropagation and Approximate Inference in Deep Generative Models
- The Concrete Distribution: A Continuous Relaxation of Discrete Random Variables
- Multi-Object Representation Learning with Iterative Variational Inference
- Sparse Unsupervised Capsules Generalize Better
- Stacked Capsule Autoencoders
- Revisiting Reweighted Wake-Sleep for Models with Stochastic Control Flow