Corpus ID: 85556114

Limited-Memory BFGS with Displacement Aggregation.

@article{Berahas2019LimitedMemoryBW,
  title={Limited-Memory BFGS with Displacement Aggregation.},
  author={Albert S. Berahas and Frank E. Curtis and Baoyu Zhou},
  journal={arXiv: Optimization and Control},
  year={2019}
}
  • Albert S. Berahas, Frank E. Curtis, Baoyu Zhou
  • Published 2019
  • Mathematics
  • arXiv: Optimization and Control
  • A displacement aggregation strategy is proposed for the curvature pairs stored in a limited-memory BFGS method such that the resulting (inverse) Hessian approximations are equal to those that would be derived from a full-memory BFGS method. This means that, if a sufficiently large number of pairs are stored, then an optimization algorithm employing the limited-memory method can achieve the same theoretical convergence properties as when full-memory (inverse) Hessian approximations are stored… CONTINUE READING

    Figures from this paper

    References

    SHOWING 1-10 OF 42 REFERENCES
    Stochastic Block BFGS: Squeezing More Curvature out of Data
    • 97
    • PDF
    Updating Quasi-Newton Matrices With Limited Storage
    • 2,176
    • Highly Influential
    • PDF
    A Stochastic Quasi-Newton Method for Large-Scale Optimization
    • 269
    • PDF
    A Multi-Batch L-BFGS Method for Machine Learning
    • 53
    • PDF
    A numerical study of limited memory BFGS methods
    • 46
    Global convergence of online limited memory BFGS
    • 94
    • PDF
    BFGS with Update Skipping and Varying Memory
    • 46
    • PDF
    A Self-Correcting Variable-Metric Algorithm for Stochastic Optimization
    • F. Curtis
    • Mathematics, Computer Science
    • ICML
    • 2016
    • 24
    • PDF
    New limited memory bundle method for large-scale nonsmooth optimization
    • 96
    • PDF
    A robust multi-batch L-BFGS method for machine learning*
    • 10
    • PDF