On the Interpretability of Conditional Probability Estimates in the Agnostic Setting

@inproceedings{Gao2017OnTI,
  title={On the Interpretability of Conditional Probability Estimates in the Agnostic Setting},
  author={Yihan Gao and Aditya G. Parameswaran and J. Peng},
  booktitle={AISTATS},
  year={2017}
}
  • Yihan Gao, Aditya G. Parameswaran, J. Peng
  • Published in AISTATS 2017
  • Computer Science, Mathematics
  • We study the interpretability of conditional probability estimates for binary classification under the agnostic setting or scenario. Under the agnostic setting, conditional probability estimates do not necessarily reflect the true conditional probabilities. Instead, they have a certain calibration property: among all data points that the classifier has predicted P(Y = 1|X) = p, p portion of them actually have label Y = 1. For cost-sensitive decision problems, this calibration property provides… CONTINUE READING
    3 Citations
    Randomized Ablation Feature Importance
    • 1
    • PDF
    Are Graph Neural Networks Miscalibrated?
    • 3
    • PDF
    Enhancing Evolutionary Couplings with Deep Convolutional Neural Networks
    • 60
    • PDF

    References

    SHOWING 1-10 OF 28 REFERENCES
    Sparseness vs Estimating Conditional Probabilities: Some Asymptotic Results
    • 73
    • PDF
    Predicting good probabilities with supervised learning
    • 752
    • PDF
    Transforming classifier scores into accurate multiclass probability estimates
    • 744
    • PDF
    Rademacher and Gaussian Complexities: Risk Bounds and Structural Results
    • 1,795
    • Highly Influential
    • PDF
    A Uniform Convergence Bound for the Area Under the ROC Curve
    • 14
    • PDF
    Latent Dirichlet Allocation
    • 26,012
    • PDF
    Efficient Learning of Generalized Linear and Single Index Models with Isotonic Regression
    • 109
    • PDF