Corpus ID: 43893689

Socratic Learning: Augmenting Generative Models to Incorporate Latent Subsets in Training Data

@article{Varma2016SocraticLA,
  title={Socratic Learning: Augmenting Generative Models to Incorporate Latent Subsets in Training Data},
  author={P. Varma and Bryan He and Dan Iter and Peng Xu and R. Yu and C. D. Sa and Christopher R{\'e}},
  journal={arXiv: Learning},
  year={2016}
}
  • P. Varma, Bryan He, +4 authors Christopher Ré
  • Published 2016
  • Computer Science, Mathematics
  • arXiv: Learning
  • A challenge in training discriminative models like neural networks is obtaining enough labeled training data. Recent approaches use generative models to combine weak supervision sources, like user-defined heuristics or knowledge bases, to label training data. Prior work has explored learning accuracies for these sources even without ground truth labels, but they assume that a single accuracy parameter is sufficient to model the behavior of these sources over the entire training set. In… CONTINUE READING
    20 Citations
    Inferring Generative Model Structure with Static Analysis
    • 31
    • PDF
    Learning Dependency Structures for Weak Supervision Models
    • 19
    • PDF
    Interactive Weak Supervision: Learning Useful Heuristics for Data Labeling
    • PDF
    Fidelity-Weighted Learning
    • 33
    • PDF
    Hybridization of Active Learning and Data Programming for Labeling Large Industrial Datasets
    • 6
    • Highly Influenced
    Zero-Shot Task Transfer
    • 16
    • PDF
    Utilizing Weak Supervision to Infer Complex Objects and Situations in Autonomous Driving Data
    • 1
    Pairwise Feedback for Data Programming
    • 2
    • PDF
    DIAG-NRE: A Deep Pattern Diagnosis Framework for Distant Supervision Neural Relation Extraction
    • 1
    • PDF

    References

    SHOWING 1-10 OF 34 REFERENCES
    Learning the Structure of Generative Models without Labeled Data
    • 72
    • PDF
    Data Programming: Creating Large Training Sets, Quickly
    • 299
    • PDF
    Training Deep Neural Networks on Noisy Labels with Bootstrapping
    • 459
    • PDF
    Generative Adversarial Nets
    • 20,249
    • PDF
    Combining Generative and Discriminative Model Scores for Distant Supervision
    • 38
    • PDF
    Label-Free Supervision of Neural Networks with Physics and Domain Knowledge
    • 173
    • PDF
    Reducing Wrong Labels in Distant Supervision for Relation Extraction
    • 179
    • PDF
    Modeling Relations and Their Mentions without Labeled Text
    • 809
    • PDF