Deterministic Independent Component Analysis

Authors: Ruitong Huang, Andras Gyorgy, Csaba Szepesvári

ICML 2015 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental Experimental results are reported in Section 6.
Researcher Affiliation Academia Department of Computing Science, University of Alberta, Edmonton, AB T6G2E8 Canada
Pseudocode Yes Algorithm 1 The HKICA algorithm. input x(t) for 1 t T. output An estimation of the mixing matrix A.; Algorithm 2 Deterministic ICA (DICA) input x(t) for 1 t T. output An estimation of the mixing matrix A.; Algorithm 3 Recursive version of HKICA (HKICA.R) input x(t) for 1 t T. output An estimation of the mixing matrix A.; Algorithm 4 The Decompose helper function
Open Source Code No The paper does not provide concrete access to source code for the methodology described.
Open Datasets No The paper uses synthetically generated data rather than a publicly available dataset: 'We generate a 6-dimensional BPSK signal s as follows. Let p = ( 19). We generate a {+1, 1} valued sequence q(t) uniformly at random for 1 t T, and set si(t) = q(t)i sin(pit).'
Dataset Splits No The paper generates data ('We take T = 20000 instances of the observed signal') but does not specify how this data is split into training, validation, or test sets.
Hardware Specification No The paper does not provide specific hardware details used for running its experiments.
Software Dependencies No The paper mentions using 'ITE toolbox (Szab o et al., 2012)' and 'recursive Fourier PCA algorithm of Xiao (2014)' but does not provide specific version numbers for any software dependencies.
Experiment Setup Yes We take T = 20000 instances of the observed signal on time steps t = 1, . . . , 20000. We test the noise ratio c from 0 (noise-free) to 1 (very noisy). All the algorithms are evaluated on a 150 repetitions. For each repetition, we try 3 times and report the best.