Bayesian Nonparametrics

Previous Topic Next Topic
 
classic Classic list List threaded Threaded
12 messages Options
Reply | Threaded
Open this post in threaded view
|

Bayesian Nonparametrics

Martin T
Hi,

I recently started working on a package on Bayesian Nonparametrics. Would anyone be interested in joining?
The current focus methods like: Dirichlet Process Mixture Models with different inference approaches, Hirarchical Dirichlet Process Mixture Models with different inference approaches, Indian Buffet Process Model and several extensions of those models. (E.g. models based on Pitman–Yor process)
I would be happy in someone shares my interest in those models and would like to contribute to the package.

The current code base is probably too messed up for a decent API and as I will not have the time to polish it, it will likely stay that way.

best,
Martin

--
You received this message because you are subscribed to the Google Groups "julia-stats" group.
To unsubscribe from this group and stop receiving emails from it, send an email to [hidden email].
For more options, visit https://groups.google.com/d/optout.
Reply | Threaded
Open this post in threaded view
|

Re: Bayesian Nonparametrics

Alex Williams
Yes, I'd be interested. I probably don't have the bandwidth to make substantial contributions, but I can help test out what you write at the very least. Send me the link to the Github repo.

Thanks and good luck!

-- Alex


On Tue, Sep 15, 2015 at 7:43 AM, Martin T <[hidden email]> wrote:
Hi,

I recently started working on a package on Bayesian Nonparametrics. Would anyone be interested in joining?
The current focus methods like: Dirichlet Process Mixture Models with different inference approaches, Hirarchical Dirichlet Process Mixture Models with different inference approaches, Indian Buffet Process Model and several extensions of those models. (E.g. models based on Pitman–Yor process)
I would be happy in someone shares my interest in those models and would like to contribute to the package.

The current code base is probably too messed up for a decent API and as I will not have the time to polish it, it will likely stay that way.

best,
Martin

--
You received this message because you are subscribed to the Google Groups "julia-stats" group.
To unsubscribe from this group and stop receiving emails from it, send an email to [hidden email].
For more options, visit https://groups.google.com/d/optout.

--
You received this message because you are subscribed to the Google Groups "julia-stats" group.
To unsubscribe from this group and stop receiving emails from it, send an email to [hidden email].
For more options, visit https://groups.google.com/d/optout.
Reply | Threaded
Open this post in threaded view
|

Re: Bayesian Nonparametrics

Adham Beyki
In reply to this post by Martin T
I'd be interested too cause I have been working on a package for Bayesian Mixture models on and off too. I have adopted Yee Whye Teh's approach in his MATLAB code for HDP and coded the Gibbs sampler for BMM, DPM, LDA and HDP.

Best,
Adham

--
You received this message because you are subscribed to the Google Groups "julia-stats" group.
To unsubscribe from this group and stop receiving emails from it, send an email to [hidden email].
For more options, visit https://groups.google.com/d/optout.
Reply | Threaded
Open this post in threaded view
|

Re: Bayesian Nonparametrics

Martin T
In reply to this post by Alex Williams
This sounds great.
Maybe if you are interested, you could also provide some ideas on interesting examples.

I will clean up the repo a bit and send you the link afterwards.

best,
Martin

On Tuesday, September 15, 2015 at 5:31:34 PM UTC+2, Alex Williams wrote:
Yes, I'd be interested. I probably don't have the bandwidth to make substantial contributions, but I can help test out what you write at the very least. Send me the link to the Github repo.

Thanks and good luck!

-- Alex


On Tue, Sep 15, 2015 at 7:43 AM, Martin T <<a href="javascript:" target="_blank" gdf-obfuscated-mailto="NuMpUMM_AgAJ" rel="nofollow" onmousedown="this.href=&#39;javascript:&#39;;return true;" onclick="this.href=&#39;javascript:&#39;;return true;">trapp....@...> wrote:
Hi,

I recently started working on a package on Bayesian Nonparametrics. Would anyone be interested in joining?
The current focus methods like: Dirichlet Process Mixture Models with different inference approaches, Hirarchical Dirichlet Process Mixture Models with different inference approaches, Indian Buffet Process Model and several extensions of those models. (E.g. models based on Pitman–Yor process)
I would be happy in someone shares my interest in those models and would like to contribute to the package.

The current code base is probably too messed up for a decent API and as I will not have the time to polish it, it will likely stay that way.

best,
Martin

--
You received this message because you are subscribed to the Google Groups "julia-stats" group.
To unsubscribe from this group and stop receiving emails from it, send an email to <a href="javascript:" target="_blank" gdf-obfuscated-mailto="NuMpUMM_AgAJ" rel="nofollow" onmousedown="this.href=&#39;javascript:&#39;;return true;" onclick="this.href=&#39;javascript:&#39;;return true;">julia-stats...@googlegroups.com.
For more options, visit <a href="https://groups.google.com/d/optout" target="_blank" rel="nofollow" onmousedown="this.href=&#39;https://groups.google.com/d/optout&#39;;return true;" onclick="this.href=&#39;https://groups.google.com/d/optout&#39;;return true;">https://groups.google.com/d/optout.

--
You received this message because you are subscribed to the Google Groups "julia-stats" group.
To unsubscribe from this group and stop receiving emails from it, send an email to [hidden email].
For more options, visit https://groups.google.com/d/optout.
Reply | Threaded
Open this post in threaded view
|

Re: Bayesian Nonparametrics

Martin T
In reply to this post by Adham Beyki
Thats great!
The initial stuff I wrote was also similar to the code by Yee Whye Teh.
I'll clean up the repo a bit and send you the link. It would be great if you could take a look at the rough structure I follow, and feel free to integrate your stuff..
As I just started, there will be only the DP Mixture Model some dataset generators and maybe a HDP with Gibbs.

best,
Martin

On Wednesday, September 16, 2015 at 4:42:03 AM UTC+2, Adham Beyki wrote:
I'd be interested too cause I have been working on a package for Bayesian Mixture models on and off too. I have adopted Yee Whye Teh's approach in his MATLAB code for HDP and coded the Gibbs sampler for BMM, DPM, LDA and HDP.

Best,
Adham

--
You received this message because you are subscribed to the Google Groups "julia-stats" group.
To unsubscribe from this group and stop receiving emails from it, send an email to [hidden email].
For more options, visit https://groups.google.com/d/optout.
Reply | Threaded
Open this post in threaded view
|

Re: Bayesian Nonparametrics

Vincent Grunert
In reply to this post by Martin T
Same here. I'd be happy to help too.

On Tuesday, September 15, 2015 at 4:43:42 PM UTC+2, Martin T wrote:
Hi,

I recently started working on a package on Bayesian Nonparametrics. Would anyone be interested in joining?
The current focus methods like: Dirichlet Process Mixture Models with different inference approaches, Hirarchical Dirichlet Process Mixture Models with different inference approaches, Indian Buffet Process Model and several extensions of those models. (E.g. models based on Pitman–Yor process)
I would be happy in someone shares my interest in those models and would like to contribute to the package.

The current code base is probably too messed up for a decent API and as I will not have the time to polish it, it will likely stay that way.

best,
Martin

--
You received this message because you are subscribed to the Google Groups "julia-stats" group.
To unsubscribe from this group and stop receiving emails from it, send an email to [hidden email].
For more options, visit https://groups.google.com/d/optout.
Reply | Threaded
Open this post in threaded view
|

Re: Bayesian Nonparametrics

Matthew Pearce
In reply to this post by Martin T
Hello

I'm just getting into Julia, but my intended use is in this area. I have written samplers for DPMMs and IBP based models in other languages.

Can't see a link to the repo in this thread, but it would be interesting to follow and perhaps contribute at a later date.

Matthew

On Tuesday, September 15, 2015 at 3:43:42 PM UTC+1, Martin T wrote:
Hi,

I recently started working on a package on Bayesian Nonparametrics. Would anyone be interested in joining?
The current focus methods like: Dirichlet Process Mixture Models with different inference approaches, Hirarchical Dirichlet Process Mixture Models with different inference approaches, Indian Buffet Process Model and several extensions of those models. (E.g. models based on Pitman–Yor process)
I would be happy in someone shares my interest in those models and would like to contribute to the package.

The current code base is probably too messed up for a decent API and as I will not have the time to polish it, it will likely stay that way.

best,
Martin

--
You received this message because you are subscribed to the Google Groups "julia-stats" group.
To unsubscribe from this group and stop receiving emails from it, send an email to [hidden email].
For more options, visit https://groups.google.com/d/optout.
Reply | Threaded
Open this post in threaded view
|

Re: Bayesian Nonparametrics

Martin T
Hi,
I tried to clean up the some of the stuff I coded. I'll have to move some more codes to the following repo in the next weeks.
But if anyone is interested you can find the repo under:
https://github.com/trappmartin/npBayes.jl

best,
Martin


On Friday, September 18, 2015 at 5:12:20 PM UTC+2, Matthew Pearce wrote:
Hello

I'm just getting into Julia, but my intended use is in this area. I have written samplers for DPMMs and IBP based models in other languages.

Can't see a link to the repo in this thread, but it would be interesting to follow and perhaps contribute at a later date.

Matthew

On Tuesday, September 15, 2015 at 3:43:42 PM UTC+1, Martin T wrote:
Hi,

I recently started working on a package on Bayesian Nonparametrics. Would anyone be interested in joining?
The current focus methods like: Dirichlet Process Mixture Models with different inference approaches, Hirarchical Dirichlet Process Mixture Models with different inference approaches, Indian Buffet Process Model and several extensions of those models. (E.g. models based on Pitman–Yor process)
I would be happy in someone shares my interest in those models and would like to contribute to the package.

The current code base is probably too messed up for a decent API and as I will not have the time to polish it, it will likely stay that way.

best,
Martin

--
You received this message because you are subscribed to the Google Groups "julia-stats" group.
To unsubscribe from this group and stop receiving emails from it, send an email to [hidden email].
For more options, visit https://groups.google.com/d/optout.
Reply | Threaded
Open this post in threaded view
|

new package: BNP.jl

Martin T
Dear all,

I moved the repository and refactored lots of code to hopefully provide a first usable package on Bayesian Nonparamtrics.
The package will be presented at the NIPS Workshop on Bayesian Nonparametrics: The Next Generation.

New repo: https://github.com/trappmartin/BNP.jl
Docs: http://bnpjl.readthedocs.org/en/latest/

I'll continue cleaning up the code and producting more documentation as well as demos and further codes of BNP models.

Best,
Martin

--
You received this message because you are subscribed to the Google Groups "julia-stats" group.
To unsubscribe from this group and stop receiving emails from it, send an email to [hidden email].
For more options, visit https://groups.google.com/d/optout.
Reply | Threaded
Open this post in threaded view
|

Re: new package: BNP.jl

Charles Novaes de Santana
Dear Martin,

Thanks for sharing BNP.jl. It looks very interesting!!

I am not familiar enough with Bayesian statistics, but I recently I started to be interested in the use of Approximate Bayesian Computation (ABC) to compare models outputs to real data. We found an interesting package in R that helps us in this task: https://cran.r-project.org/web/packages/abc/index.html

Is BNP.jl somehow related to this (ABC)?

Thanks for your attention,

Charles

On 3 November 2015 at 13:59, Martin T <[hidden email]> wrote:
Dear all,

I moved the repository and refactored lots of code to hopefully provide a first usable package on Bayesian Nonparamtrics.
The package will be presented at the NIPS Workshop on Bayesian Nonparametrics: The Next Generation.

New repo: https://github.com/trappmartin/BNP.jl
Docs: http://bnpjl.readthedocs.org/en/latest/

I'll continue cleaning up the code and producting more documentation as well as demos and further codes of BNP models.

Best,
Martin

--
You received this message because you are subscribed to the Google Groups "julia-stats" group.
To unsubscribe from this group and stop receiving emails from it, send an email to [hidden email].
For more options, visit https://groups.google.com/d/optout.



--
Um axé! :)

--
Charles Novaes de Santana, PhD
http://www.imedea.uib-csic.es/~charles

--
You received this message because you are subscribed to the Google Groups "julia-stats" group.
To unsubscribe from this group and stop receiving emails from it, send an email to [hidden email].
For more options, visit https://groups.google.com/d/optout.
Reply | Threaded
Open this post in threaded view
|

Re: new package: BNP.jl

Jonathan Malmaud
Bayesian inference seems to be one of the popular applications of Julia (alongside computational neuroscience), so I suspect a package that does a solid job of nonparametric bayes would be highly impactful. 

On Wednesday, November 4, 2015 at 8:55:37 AM UTC-6, Charles Santana wrote:
Dear Martin,

Thanks for sharing BNP.jl. It looks very interesting!!

I am not familiar enough with Bayesian statistics, but I recently I started to be interested in the use of Approximate Bayesian Computation (ABC) to compare models outputs to real data. We found an interesting package in R that helps us in this task: <a href="https://cran.r-project.org/web/packages/abc/index.html" target="_blank" rel="nofollow" onmousedown="this.href=&#39;https://www.google.com/url?q\75https%3A%2F%2Fcran.r-project.org%2Fweb%2Fpackages%2Fabc%2Findex.html\46sa\75D\46sntz\0751\46usg\75AFQjCNH7RzPPTsupiYOKy4rZ74EywoTjeg&#39;;return true;" onclick="this.href=&#39;https://www.google.com/url?q\75https%3A%2F%2Fcran.r-project.org%2Fweb%2Fpackages%2Fabc%2Findex.html\46sa\75D\46sntz\0751\46usg\75AFQjCNH7RzPPTsupiYOKy4rZ74EywoTjeg&#39;;return true;">https://cran.r-project.org/web/packages/abc/index.html

Is BNP.jl somehow related to this (ABC)?

Thanks for your attention,

Charles

On 3 November 2015 at 13:59, Martin T <<a href="javascript:" target="_blank" gdf-obfuscated-mailto="EQSrNS2KAwAJ" rel="nofollow" onmousedown="this.href=&#39;javascript:&#39;;return true;" onclick="this.href=&#39;javascript:&#39;;return true;">trapp....@...> wrote:
Dear all,

I moved the repository and refactored lots of code to hopefully provide a first usable package on Bayesian Nonparamtrics.
The package will be presented at the NIPS Workshop on Bayesian Nonparametrics: The Next Generation.

New repo: <a href="https://github.com/trappmartin/BNP.jl" target="_blank" rel="nofollow" onmousedown="this.href=&#39;https://www.google.com/url?q\75https%3A%2F%2Fgithub.com%2Ftrappmartin%2FBNP.jl\46sa\75D\46sntz\0751\46usg\75AFQjCNEHsfgeuQFV6SebHBNqSEbdR0Twnw&#39;;return true;" onclick="this.href=&#39;https://www.google.com/url?q\75https%3A%2F%2Fgithub.com%2Ftrappmartin%2FBNP.jl\46sa\75D\46sntz\0751\46usg\75AFQjCNEHsfgeuQFV6SebHBNqSEbdR0Twnw&#39;;return true;">https://github.com/trappmartin/BNP.jl
Docs: <a href="http://bnpjl.readthedocs.org/en/latest/" target="_blank" rel="nofollow" onmousedown="this.href=&#39;http://www.google.com/url?q\75http%3A%2F%2Fbnpjl.readthedocs.org%2Fen%2Flatest%2F\46sa\75D\46sntz\0751\46usg\75AFQjCNExGNqjItM0wxy_kW_Lz2rq4nNNMg&#39;;return true;" onclick="this.href=&#39;http://www.google.com/url?q\75http%3A%2F%2Fbnpjl.readthedocs.org%2Fen%2Flatest%2F\46sa\75D\46sntz\0751\46usg\75AFQjCNExGNqjItM0wxy_kW_Lz2rq4nNNMg&#39;;return true;">http://bnpjl.readthedocs.org/en/latest/

I'll continue cleaning up the code and producting more documentation as well as demos and further codes of BNP models.

Best,
Martin

--
You received this message because you are subscribed to the Google Groups "julia-stats" group.
To unsubscribe from this group and stop receiving emails from it, send an email to <a href="javascript:" target="_blank" gdf-obfuscated-mailto="EQSrNS2KAwAJ" rel="nofollow" onmousedown="this.href=&#39;javascript:&#39;;return true;" onclick="this.href=&#39;javascript:&#39;;return true;">julia-stats...@googlegroups.com.
For more options, visit <a href="https://groups.google.com/d/optout" target="_blank" rel="nofollow" onmousedown="this.href=&#39;https://groups.google.com/d/optout&#39;;return true;" onclick="this.href=&#39;https://groups.google.com/d/optout&#39;;return true;">https://groups.google.com/d/optout.



--
Um axé! :)

--
Charles Novaes de Santana, PhD
<a href="http://www.imedea.uib-csic.es/~charles" target="_blank" rel="nofollow" onmousedown="this.href=&#39;http://www.google.com/url?q\75http%3A%2F%2Fwww.imedea.uib-csic.es%2F~charles\46sa\75D\46sntz\0751\46usg\75AFQjCNEqj05GhlMnUboXpyk_8cU8Hh4LtA&#39;;return true;" onclick="this.href=&#39;http://www.google.com/url?q\75http%3A%2F%2Fwww.imedea.uib-csic.es%2F~charles\46sa\75D\46sntz\0751\46usg\75AFQjCNEqj05GhlMnUboXpyk_8cU8Hh4LtA&#39;;return true;">http://www.imedea.uib-csic.es/~charles

--
You received this message because you are subscribed to the Google Groups "julia-stats" group.
To unsubscribe from this group and stop receiving emails from it, send an email to [hidden email].
For more options, visit https://groups.google.com/d/optout.
Reply | Threaded
Open this post in threaded view
|

Re: new package: BNP.jl

Dustin Tran
> Is BNP.jl somehow related to this (ABC)? 

ABC is more generally a technique for Bayesian inference, and which includes as a set of models Bayesian nonparametric ones for inference. It’s simply a scheme where standard Monte Carlo techniques should be avoided because likelihood evaluation is very expensive if not impossible.

Dustin
On Nov 4, 2015, at 8:39 PM, Jonathan Malmaud <[hidden email]> wrote:

Bayesian inference seems to be one of the popular applications of Julia (alongside computational neuroscience), so I suspect a package that does a solid job of nonparametric bayes would be highly impactful. 

On Wednesday, November 4, 2015 at 8:55:37 AM UTC-6, Charles Santana wrote:
Dear Martin,

Thanks for sharing BNP.jl. It looks very interesting!!

I am not familiar enough with Bayesian statistics, but I recently I started to be interested in the use of Approximate Bayesian Computation (ABC) to compare models outputs to real data. We found an interesting package in R that helps us in this task: <a href="https://cran.r-project.org/web/packages/abc/index.html" target="_blank" rel="nofollow" onmousedown="this.href='https://www.google.com/url?q\75https%3A%2F%2Fcran.r-project.org%2Fweb%2Fpackages%2Fabc%2Findex.html\46sa\75D\46sntz\0751\46usg\75AFQjCNH7RzPPTsupiYOKy4rZ74EywoTjeg';return true;" onclick="this.href='https://www.google.com/url?q\75https%3A%2F%2Fcran.r-project.org%2Fweb%2Fpackages%2Fabc%2Findex.html\46sa\75D\46sntz\0751\46usg\75AFQjCNH7RzPPTsupiYOKy4rZ74EywoTjeg';return true;" class="">https://cran.r-project.org/web/packages/abc/index.html

Is BNP.jl somehow related to this (ABC)?

Thanks for your attention,

Charles

On 3 November 2015 at 13:59, Martin T <<a href="javascript:" target="_blank" gdf-obfuscated-mailto="EQSrNS2KAwAJ" rel="nofollow" onmousedown="this.href='javascript:';return true;" onclick="this.href='javascript:';return true;" class="">trapp....@...> wrote:
Dear all,

I moved the repository and refactored lots of code to hopefully provide a first usable package on Bayesian Nonparamtrics.
The package will be presented at the NIPS Workshop on Bayesian Nonparametrics: The Next Generation.

New repo: <a href="https://github.com/trappmartin/BNP.jl" target="_blank" rel="nofollow" onmousedown="this.href='https://www.google.com/url?q\75https%3A%2F%2Fgithub.com%2Ftrappmartin%2FBNP.jl\46sa\75D\46sntz\0751\46usg\75AFQjCNEHsfgeuQFV6SebHBNqSEbdR0Twnw';return true;" onclick="this.href='https://www.google.com/url?q\75https%3A%2F%2Fgithub.com%2Ftrappmartin%2FBNP.jl\46sa\75D\46sntz\0751\46usg\75AFQjCNEHsfgeuQFV6SebHBNqSEbdR0Twnw';return true;" class="">https://github.com/trappmartin/BNP.jl
Docs: <a href="http://bnpjl.readthedocs.org/en/latest/" target="_blank" rel="nofollow" onmousedown="this.href='http://www.google.com/url?q\75http%3A%2F%2Fbnpjl.readthedocs.org%2Fen%2Flatest%2F\46sa\75D\46sntz\0751\46usg\75AFQjCNExGNqjItM0wxy_kW_Lz2rq4nNNMg';return true;" onclick="this.href='http://www.google.com/url?q\75http%3A%2F%2Fbnpjl.readthedocs.org%2Fen%2Flatest%2F\46sa\75D\46sntz\0751\46usg\75AFQjCNExGNqjItM0wxy_kW_Lz2rq4nNNMg';return true;" class="">http://bnpjl.readthedocs.org/en/latest/

I'll continue cleaning up the code and producting more documentation as well as demos and further codes of BNP models.

Best,
Martin

--
You received this message because you are subscribed to the Google Groups "julia-stats" group.
To unsubscribe from this group and stop receiving emails from it, send an email to <a href="javascript:" target="_blank" gdf-obfuscated-mailto="EQSrNS2KAwAJ" rel="nofollow" onmousedown="this.href='javascript:';return true;" onclick="this.href='javascript:';return true;" class="">julia-stats...@googlegroups.com.
For more options, visit <a href="https://groups.google.com/d/optout" target="_blank" rel="nofollow" onmousedown="this.href='https://groups.google.com/d/optout';return true;" onclick="this.href='https://groups.google.com/d/optout';return true;" class="">https://groups.google.com/d/optout.



--
Um axé! :)

--
Charles Novaes de Santana, PhD
<a href="http://www.imedea.uib-csic.es/~charles" target="_blank" rel="nofollow" onmousedown="this.href='http://www.google.com/url?q\75http%3A%2F%2Fwww.imedea.uib-csic.es%2F~charles\46sa\75D\46sntz\0751\46usg\75AFQjCNEqj05GhlMnUboXpyk_8cU8Hh4LtA';return true;" onclick="this.href='http://www.google.com/url?q\75http%3A%2F%2Fwww.imedea.uib-csic.es%2F~charles\46sa\75D\46sntz\0751\46usg\75AFQjCNEqj05GhlMnUboXpyk_8cU8Hh4LtA';return true;" class="">http://www.imedea.uib-csic.es/~charles

--
You received this message because you are subscribed to the Google Groups "julia-stats" group.
To unsubscribe from this group and stop receiving emails from it, send an email to [hidden email].
For more options, visit https://groups.google.com/d/optout.

--
You received this message because you are subscribed to the Google Groups "julia-stats" group.
To unsubscribe from this group and stop receiving emails from it, send an email to [hidden email].
For more options, visit https://groups.google.com/d/optout.