Early Visual Concept Learning with Unsupervised Deep Learning
This addresses the problem of unsupervised visual concept learning for AI research, with incremental improvements in disentanglement methods.
The paper tackles the challenge of automated discovery of early visual concepts from raw image data by proposing an unsupervised approach for learning disentangled representations of underlying factors of variation, resulting in a variational autoencoder framework that works well across various datasets and exhibits emergent properties like zero-shot inference.
Automated discovery of early visual concepts from raw image data is a major open challenge in AI research. Addressing this problem, we propose an unsupervised approach for learning disentangled representations of the underlying factors of variation. We draw inspiration from neuroscience, and show how this can be achieved in an unsupervised generative model by applying the same learning pressures as have been suggested to act in the ventral visual stream in the brain. By enforcing redundancy reduction, encouraging statistical independence, and exposure to data with transform continuities analogous to those to which human infants are exposed, we obtain a variational autoencoder (VAE) framework capable of learning disentangled factors. Our approach makes few assumptions and works well across a wide variety of datasets. Furthermore, our solution has useful emergent properties, such as zero-shot inference and an intuitive understanding of "objectness".