Comments - Nice Generalization of the K-NN Clustering Algorithm -- Also Useful for Data Reduction - Data Science Central2019-07-21T09:39:39Zhttps://www.datasciencecentral.com/profiles/comment/feed?attachedTo=6448529%3ABlogPost%3A606241&%3Bxn_auth=noGreat article Vincent!
Have y…tag:www.datasciencecentral.com,2017-10-31:6448529:Comment:6441612017-10-31T21:37:29.475ZRafael Fantini da Costahttps://www.datasciencecentral.com/profile/RafaelFantinidaCosta
<p>Great article Vincent!<br/><br/></p>
<p>Have you compared the results compare with a Gaussian Mixture Model with spherical clusters (e.g. when the covariance matrix is constrained to r * I, where r is related to the radius of the circle and I is de identity matrix)?</p>
<p>Great article Vincent!<br/><br/></p>
<p>Have you compared the results compare with a Gaussian Mixture Model with spherical clusters (e.g. when the covariance matrix is constrained to r * I, where r is related to the radius of the circle and I is de identity matrix)?</p> Interesting approach! Have yo…tag:www.datasciencecentral.com,2017-08-16:6448529:Comment:6075282017-08-16T13:29:33.519ZPulkkinen Heikkihttps://www.datasciencecentral.com/profile/PulkkinenHeikki
<p>Interesting approach! Have you compared the performance to K-NN yet?<br/><br/>The brute force search over all data points is not the only way of making K-NN queries. A precomputed <a href="https://en.wikipedia.org/wiki/K-d_tree" target="_blank">k-d tree</a> (which can be compared to computing the cliques) can be used to find nearest neighbours in logarithmic time. The downside is that the efficiency drops in higher dimensions.</p>
<p>Interesting approach! Have you compared the performance to K-NN yet?<br/><br/>The brute force search over all data points is not the only way of making K-NN queries. A precomputed <a href="https://en.wikipedia.org/wiki/K-d_tree" target="_blank">k-d tree</a> (which can be compared to computing the cliques) can be used to find nearest neighbours in logarithmic time. The downside is that the efficiency drops in higher dimensions.</p>