Figure - PMC

Skip to main content

An official website of the United States government

Here's how you know

Here's how you know

Official websites use .gov
A .gov website belongs to an official government organization in the United States.

Secure .gov websites use HTTPS
A lock ( ) or https:// means you've safely connected to the .gov website. Share sensitive information only on official, secure websites.

View full-text article in PMC

. 2021 Oct 8;9(2):nwab183. doi: 10.1093/nsr/nwab183

Search in PMC
Search in PubMed
View in NLM Catalog
Add to search

© The Author(s) 2021. Published by Oxford University Press on behalf of China Science Publishing & Media Ltd.

This is an Open Access article distributed under the terms of the Creative Commons Attribution License (https://creativecommons.org/licenses/by/4.0/), which permits unrestricted reuse, distribution, and reproduction in any medium, provided the original work is properly cited.

PMC Copyright notice

Figure 4. — Clustering 1.5 million face images and 11 million HDFS log texts. (a and b) Results of 28 photos labeled ‘Jon Polito’ in clustering the first 500 images by using (a) WFC and (b) k-means respectively. There are 24 labels of the first 500 images. By setting , k-means achieves its highest F1-measure score (0.809), resulting in only 21 correctly classified images. WFC identified a cluster with 27 correct images at in a fully unsupervised way. Finer clusters with stronger similarities were also detected at the next two scales. (c) Validation scores for clustering the first 500 images. (d) Running times and evaluation scores (F1-measure, purity) of WFC, k-means and kNN-Louvain using distributed computing. Each result in (c and d) was computed by running each experiment 10 times, and error bars indicate the standard error of the mean. (e) Multi-scale clusters of HDFS logs detected by WFC, representing meaningful HDFS operations at different hierarchical levels. The width of each line is proportional to the logarithm of corresponding cluster size.

Inline graphic — Clustering 1.5 million face images and 11 million HDFS log texts. (a and b) Results of 28 photos labeled ‘Jon Polito’ in clustering the first 500 images by using (a) WFC and (b) k-means respectively. There are 24 labels of the first 500 images. By setting , k-means achieves its highest F1-measure score (0.809), resulting in only 21 correctly classified images. WFC identified a cluster with 27 correct images at in a fully unsupervised way. Finer clusters with stronger similarities were also detected at the next two scales. (c) Validation scores for clustering the first 500 images. (d) Running times and evaluation scores (F1-measure, purity) of WFC, k-means and kNN-Louvain using distributed computing. Each result in (c and d) was computed by running each experiment 10 times, and error bars indicate the standard error of the mean. (e) Multi-scale clusters of HDFS logs detected by WFC, representing meaningful HDFS operations at different hierarchical levels. The width of each line is proportional to the logarithm of corresponding cluster size.