Learning with Hierarchical-Deep Models

@article{Salakhutdinov2013LearningWH,
  title={Learning with Hierarchical-Deep Models},
  author={R. Salakhutdinov and J. Tenenbaum and A. Torralba},
  journal={IEEE Transactions on Pattern Analysis and Machine Intelligence},
  year={2013},
  volume={35},
  pages={1958-1971}
}
We introduce HD (or “Hierarchical-Deep”) models, a new compositional learning architecture that integrates deep learning models with structured hierarchical Bayesian (HB) models. Specifically, we show how we can learn a hierarchical Dirichlet process (HDP) prior over the activities of the top-level features in a deep Boltzmann machine (DBM). This compound HDP-DBM model learns to learn novel concepts from very few training example by learning low-level generic features, high-level features that… Expand
197 Citations
One-shot learning by inverting a compositional causal process
  • 169
  • PDF
Zero-Shot Learning via Class-Conditioned Deep Generative Models
  • 93
  • PDF
Deep Boltzmann Machines based vehicle recognition
  • 6
Combining Deep Universal Features, Semantic Attributes, and Hierarchical Classification for Zero-Shot Learning
  • 1
  • PDF
Learning Robust Visual-Semantic Embeddings
  • 95
  • PDF
...
1
2
3
4
5
...

References

SHOWING 1-10 OF 75 REFERENCES
Convolutional deep belief networks for scalable unsupervised learning of hierarchical representations
  • 2,333
  • PDF
Deep Boltzmann Machines
  • 1,837
  • PDF
A Fast Learning Algorithm for Deep Belief Nets
  • 11,871
  • PDF
Sparse Feature Learning for Deep Belief Networks
  • 750
  • PDF
Exploring Strategies for Training Deep Neural Networks
  • 927
  • PDF
One shot learning of simple visual concepts
  • 473
  • PDF
Learning Multiple Layers of Features from Tiny Images
  • 11,364
  • PDF
Acoustic Modeling Using Deep Belief Networks
  • 1,554
  • PDF
Implicit Mixtures of Restricted Boltzmann Machines
  • 85
  • PDF
...
1
2
3
4
5
...