Total Least Squares Regression in Input Sparsity Time

Authors: Huaian Diao, Zhao Song, David Woodruff, Xin Yang

NeurIPS 2019 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental We empirically validate our algorithm on real and synthetic data sets.
Researcher Affiliation Academia Huaian Diao Northeast Normal University & KLAS of MOE hadiao@nenu.edu.cn Zhao Song University of Washington zhaosong@uw.edu David P. Woodruff Carnegie Mellon University dwoodruf@cs.cmu.edu Xin Yang University of Washington yx1992@cs.washington.edu
Pseudocode Yes Algorithm 1 Our Fast Total Least Squares Algorithm
Open Source Code Yes 1The code can be found at https://github.com/yangxinuw/total_least_squares_code.
Open Datasets Yes We also conducted experiments on real datasets from the UCI Machine Learning Repository [DKT17]. ... We have four real datasets : Airfoil Self-Noise [UCIa] in Table 2(a), Wine Quality Red wine [UCIc, CCA+09] in Table 2(b), Wine Quality White wine [UCIc, CCA+09] in Table 2(c), Insurance Company Benchmark (COIL 2000) Data Set [UCIb, PS]
Dataset Splits No The paper discusses synthetic data and real datasets from UCI, and specifies 'sample density' for sketching, but does not provide explicit training/test/validation split percentages or sample counts for the datasets used in experiments.
Hardware Specification Yes Our numerical tests are carried out on an Intel Xeon E7-8850 v2 server with 2.30GHz and 4GB RAM
Software Dependencies Yes under Matlab R2017b.
Experiment Setup Yes In the experiments, we take sample density ρ = 0.1, 0.3, 0.6, 0.9 respectively to check our performance.