Corpus ID: 67787678

Online Sampling from Log-Concave Distributions

@inproceedings{Lee2019OnlineSF,
  title={Online Sampling from Log-Concave Distributions},
  author={Holden Lee and Oren Mangoubi and N. Vishnoi},
  booktitle={NeurIPS},
  year={2019}
}
  • Holden Lee, Oren Mangoubi, N. Vishnoi
  • Published in NeurIPS 2019
  • Computer Science, Mathematics
  • Given a sequence of convex functions $f_0, f_1, \ldots, f_T$, we study the problem of sampling from the Gibbs distribution $\pi_t \propto e^{-\sum_{k=0}^tf_k}$ for each epoch $t$ in an online manner. Interest in this problem derives from applications in machine learning, Bayesian statistics, and optimization where, rather than obtaining all the observations at once, one constantly acquires new data, and must continuously update the distribution. Our main result is an algorithm that generates… CONTINUE READING
    5 Citations

    References

    SHOWING 1-10 OF 44 REFERENCES
    Log-concave sampling: Metropolis-Hastings algorithms are fast!
    • 92
    • PDF
    Efficient Sampling from Time-Varying Log-Concave Distributions
    • 27
    • PDF
    Logarithmic regret algorithms for online convex optimization
    • 794
    • PDF
    Logistic Regression: Tight Bounds for Stochastic and Online Optimization
    • 32
    • PDF
    Coresets for Scalable Bayesian Logistic Regression
    • 101
    • PDF
    Logistic Regression: The Importance of Being Improper
    • 24
    • PDF
    Minibatch Gibbs Sampling on Large Graphical Models
    • 11
    • PDF