figtreeKCenterClustering.m

[eyuansu62]

how can i find the implementation of figtreeKCenterClustering.m

[lttam]

figtreeKCenterClustering is the 3rd-party toolbox of the farthest-point clustering. You can see the mex-files for Linux & Mac in the code, or you can reimplement the farthest-point clustering by yourself (or you may use any other clustering method).

[eyuansu62]

thanks a lot.  Could you please help me about that the tree structure you used is just an ordinary tree or it is just used to accelerate the computation or it may contains some Hierarchy structure information of data? 

…

------------------&nbsp;原始邮件&nbsp;------------------ 发件人:&nbsp;"Tam Le"<notifications@github.com&gt;; 发送时间:&nbsp;2019年11月19日(星期二) 晚上7:31 收件人:&nbsp;"lttam/TreeWasserstein"<TreeWasserstein@noreply.github.com&gt;; 抄送:&nbsp;"末冬末秋"<772468951@qq.com&gt;;"Author"<author@noreply.github.com&gt;; 主题:&nbsp;Re: [lttam/TreeWasserstein] figtreeKCenterClustering.m (#2) figtreeKCenterClustering is the 3rd-party toolbox of the farthest-point clustering. You can see the mex-files for Linux &amp; Mac in the code, or you can reimplement the farthest-point clustering by yourself (or you may use any other clustering method). — You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub, or unsubscribe.

[lttam]

For tree metric construction, we gave the details in Section 4. You may follow the Algorithm 1 for low-dimensional support data space, or may replace the partition by the clustering in case you data space is high-dimensional.

The goal is to sample tree metrics from support data space, and then uses those tree metrics as ground metrics for computing the tree-sliced-Wasserstein distance.

[eyuansu62]

thanks！i see you do not provide the code about you process the experiments in section 6.&nbsp; would you release relevant code?

…

------------------&nbsp;原始邮件&nbsp;------------------ 发件人:&nbsp;"Tam Le"<notifications@github.com&gt;; 发送时间:&nbsp;2019年11月19日(星期二) 晚上7:49 收件人:&nbsp;"lttam/TreeWasserstein"<TreeWasserstein@noreply.github.com&gt;; 抄送:&nbsp;"末冬末秋"<772468951@qq.com&gt;;"Author"<author@noreply.github.com&gt;; 主题:&nbsp;Re: [lttam/TreeWasserstein] figtreeKCenterClustering.m (#2) For tree metric construction, we gave the details in Section 4. You may follow the Algorithm 1 for low-dimensional support data space, or may replace the partition by the clustering in case you data space is high-dimensional. The goal is to sample tree metrics from support data space, and then uses those tree metrics as ground metrics for computing the tree-sliced-Wasserstein distance. — You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub, or unsubscribe.

[eyuansu62]

Oh, i see. Maybe this request is tough, and i hope you can forgive me.
best wishes!