Doubly Accelerated Methods for Faster CCA and Generalized Eigendecomposition

Authors: Zeyuan Allen-Zhu, Yuanzhi Li

ICML 2017 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Theoretical We study k-Gen EV, the problem of finding the top k generalized eigenvectors, and k-CCA, the problem of finding the top k vectors in canonicalcorrelation analysis. We propose algorithms Lazy EV and Lazy CCA to solve the two problems with running times linearly dependent on the input size and on k. Furthermore, our algorithms are doubly-accelerated: our running times depend only on the square root of the matrix condition number, and on the square root of the eigengap. This is the first such result for both k Gen EV or k-CCA. We also provide the first gapfree results, which provide running times that depend on 1/ ε rather than the eigengap.
Researcher Affiliation Collaboration 1Microsoft Research 2Princeton University. Correspondence to: Zeyuan Allen-Zhu <zeyuan@csail.mit.edu>, Yuanzhi Li <yuanzhil@cs.princeton.edu>.
Pseudocode Yes Algorithm 1 Appx PCA (A, M, δ , ε, p) ... Algorithm 2 Lazy EV(A, M, k, δ , εpca, p)
Open Source Code No The paper does not include an unambiguous statement or a direct link to the source code for the methodology described.
Open Datasets No This paper is theoretical and does not mention using or training on any specific publicly available datasets.
Dataset Splits No This paper is theoretical and does not mention any dataset splits for validation.
Hardware Specification No This is a theoretical paper and does not mention any specific hardware used for experiments.
Software Dependencies No This is a theoretical paper and does not mention specific software dependencies with version numbers.
Experiment Setup No This is a theoretical paper and does not describe any experimental setup details such as hyperparameters or training configurations.