Concept-Oriented Deep Learning: Generative Concept Representations

Abstract

Generative concept representations have three major advantages over discriminative ones: they can represent uncertainty, they support integration of learning and reasoning, and they are good for unsupervised and semi-supervised learning. We discuss probabilistic and generative deep learning, which generative concept representations are based on, and the use of variational autoencoders and generative adversarial networks for learning generative concept representations, particularly for concepts whose data are sequences, structured data or graphs.

References in corpus (12)

• Auto-Encoding Variational Bayes
• What Uncertainties Do We Need in Bayesian Deep Learning for Computer Vision?
• A Critical Review of Recurrent Neural Networks for Sequence Learning
• Tutorial on Variational Autoencoders
• A Hierarchical Latent Vector Model for Learning Long-Term Structure in Music
• Syntax-Directed Variational Autoencoder for Structured Data
• NetGAN: Generating Graphs via Random Walks
• On Unifying Deep Generative Models
• Comparative Study on Generative Adversarial Networks
• A Review of Learning with Deep Generative Models from Perspective of Graphical Modeling
• Concept-Oriented Deep Learning
• TreeGAN: Syntax-Aware Sequence Generation with Generative Adversarial Networks

Cited by in corpus (5)

• Probabilistic Deep Learning with Probabilistic Neural Networks and Deep Probabilistic Models
• Tiered Latent Representations and Latent Spaces for Molecular Graphs
• Probabilistic Generative Deep Learning for Molecular Design
• Tiered Graph Autoencoders with PyTorch Geometric for Molecular Graphs
• Latent Variable Modeling for Generative Concept Representations and Deep Generative Models