Incremental Quasi-Newton Methods with Faster Superlinear Convergence Rates

no code implementations • 4 Feb 2024 • Zhuanghua Liu, Luo Luo, Bryan Kian Hsiang Low

The recently proposed incremental quasi-Newton method is based on BFGS update and achieves a local superlinear convergence rate that is dependent on the condition number of the problem.

Decentralized Sum-of-Nonconvex Optimization

no code implementations • 4 Feb 2024 • Zhuanghua Liu, Bryan Kian Hsiang Low

However, the convergence rate of the PMGT-SVRG algorithm has a linear dependency on the condition number, which is undesirable for the ill-conditioned problem.

Towards Sharper First-Order Adversary with Quantized Gradients

1 code implementation • 1 Feb 2020 • Zhuanghua Liu, Ivor W. Tsang

However, in state-of-the-art first-order attacks, adversarial examples with sign gradients retain the sign information of each gradient component but discard the relative magnitude between components.

Adversarial Robustness Quantization