Scalable Sequential Spectral Clustering

Authors: Yeqing Li, Junzhou Huang, Wei Liu

AAAI 2016 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental Extensive experiments carried out on large datasets demonstrate that the proposed sequential SC algorithm is up to a thousand times faster than the state-of-the-arts.
Researcher Affiliation Collaboration 1University of Texas at Arlington, Texas, USA 2Didi Research, Bejing, China
Pseudocode Yes Algorithm 1 Sequential K-Means (Seq KM), Algorithm 2 Sequential Singular Value Decomposition (SSVD), Algorithm 3 Sequential Spectral Clustering (Seq SC)
Open Source Code No No explicit statement or link for the open-source code of the proposed method was found.
Open Datasets Yes MNIST1. This dataset consists of 70,000 images of handwritten digits from 0 to 9. ... Cov Type. This dataset consists of 581,012 for predicting the forest cover type from cartographic variables. ... MNIST8m2. This data set consists of 8,100,000 images of handwritten digits from 0 to 9 (Loosli, Canu, and Bottou 2007).
Dataset Splits No The paper mentions evaluating performance and comparing algorithms but does not explicitly provide training/validation/test dataset splits with percentages or sample counts, nor does it refer to predefined splits with citations for reproducibility.
Hardware Specification Yes All our experiments are conducted on a desktop computer with a 3.0GHz Intel Pentium 4 CPU and 1GB RAM, Mat Lab 7.14 (32bit).
Software Dependencies Yes Mat Lab 7.14 (32bit)
Experiment Setup Yes For parameters of our algorithm, we fix m = 200 and construct 5-NN anchor graph. For all experiments, we fix each block to contain 5000 data points.