Robust Unsupervised Learning via L-statistic Minimization

Authors: Andreas Maurer, Daniela Angela Parletta, Andrea Paudice, Massimiliano Pontil

ICML 2021 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental Numerical experiments with KMEANS clustering and principal subspace analysis demonstrate the effectiveness of our approach.
Researcher Affiliation Academia 1Istituto Italiano di Tecnologia, Genoa, Italy 2University of Genoa, Genoa, Italy 3University of Milan, Milan, Italy 4University College London, London, UK.
Pseudocode Yes Algorithm 1
Open Source Code No The paper does not provide any explicit statement about releasing code or a link to a code repository for the methodology described.
Open Datasets Yes We used the Fashion-MNIST dataset which consists of about 70000 28 28 images of various types of clothes splitted in a training set of 60000 images and a test set of 10000 images.
Dataset Splits No The paper mentions a training set of 60000 images and a test set of 10000 images, but it does not specify a separate validation set or its size.
Hardware Specification Yes All experiments have been run on an standard laptop equipped with an Intel i9 with 8 cores each working at 2,4 GHz and 16 GB of RAM DDR4 working at 2,6 GHz.
Software Dependencies No The paper states, 'For KMEANS++ we used the sklearn implementation', but it does not provide any specific version numbers for sklearn or any other software dependencies.
Experiment Setup Yes For both RKM and KMEANS++ we T = 10 and r = 30. We initialized RKM with uniform centers and set ζ = 0.75, the same ζ is used for SD. ... We run the algorithms with T = 50, r = 30, M = 4000, k = 2 and ζ in the range [0.4, 1].