Thanks Dimitry, I shall look in that in a while, after I write a tutorial for the nen recommender engine
-Thejas On Thu, Oct 2, 2014 at 12:12 PM, Dmitriy Lyubimov <[email protected]> wrote: > Matlab specific binding are largely abandoned in favor of common set of > things that tends to be more R-Like stuff. the reason is that a lot of > things (such as e.g. elementwise tensor operators) in matlab are impossible > to implement in Scala operators, or they would have undesired precedence. R > "base" bindings are much more desirable at this point to keep things > consistent. > > However, i found that just blindly following R "base" operations design is > often less than desirable as it often would require internal state to > perform well in iterations (especially sampling in Monte Carlo based > solutions). So reasonable compromise is pursued here as well -- we of > course adopt whatever make sense but we just try to keep things immediately > recognizable for people who dealt with R before. > > As it stands, the algebraic bindings are largely complete and well-rounded > (there are small things such as e.g. `1.0 rbind Matrix` or `1.0 cbind > Vector`, especially for in-core things, which may be filled in if desired, > but ot awfully a lot. In-core matrix bindings are just DSL and therefore > are fairly easy to add. Performance is the biggest issue here. Distributed > algebra bindings IMO do not need any further additions at this point. There > are few things to tweak in optimizer (e.g. it currently doesn't rewrite > something like `drmA.t.mapBlock` correctly) but it is not terribly > important to have it as opposed to having done it right. > > Extending on implementing R "base", there are huge gaps on stats side, such > as multivariate distirbutions operation support. E.g. native efficient > Inverse Wishart sampling is one example. I filled a lot of these gaps on a > private branch but unable to contribute. So stats, natively to Colt rather > than being bridged to some other library, are very desired and very > fundamental aspect as well. Without those it is quite difficult or plainly > impossible to contribute any MCMC based implementation, for example. > Another example is to have Gaussian process and gradients, along with > aquisition function support (aka "Bayesian optmization" areas), available. > This is essential for hyperparameter search. Again, i have implemented that > but that's all private as well. But at least it is all well understood. > > So there's a lot of principled and pretty fundamental work here to make > this a well rounded environment. I know it may become all very useful if > done right -- because i am already using all of it, privately. However, all > of that math is a common knowledge that is not subject to any copyright > claims. Any work toward the above is quite welcome. > > > > On Mon, Sep 29, 2014 at 11:34 AM, thejas prasad <[email protected]> > wrote: > > > Hey Ted, > > > > It seemed interesting. I was looking at Jira and also Git and It seemed > as > > though some scala bindings where already implemented.. Am I correct? > > > > I wanted to take up a task that is trivial since I am new to scala and > > even mahout. > > > > With that said I would be interested in writing more matlab bindings. > > > > Does that sound okay? > > > > -Thejas > > > > > > > On Sun, Sep 28, 2014 at 3:15 PM, Aamir Khan <[email protected]> > > wrote: > > > Hi, > > > > > > I am also new to Apache and Mahout. This thread caught my attention. > > > Can you tell what are the areas where development is required. > > > Is there any work on *Clustering*? > > > Any guidance on how to start and useful links are highly appreciated. > > > > > > Many thanks, > > > > > > > > > On Mon, Sep 29, 2014 at 1:19 AM, Ted Dunning <[email protected]> > > wrote: > > > > > > > Thejas, > > > > > > > > What were your impressions? > > > > > > > > Which parts of the system match your background and capabilities? > > > > > > > > > > > > > > > > On Sun, Sep 28, 2014 at 11:46 AM, Thejas Prasad <[email protected] > > > > > > wrote: > > > > > > > > > Hey suneel, > > > > > > > > > > I finished reading the paper. What's next? > > > > > > > > > > Sent from my iPhone > > > > > > > > > > > On Sep 26, 2014, at 7:04 PM, Suneel Marthi <[email protected]> > > wrote: > > > > > > > > > > > > See this for a start > > > > > > > > http://mahout.apache.org/users/sparkbindings/ScalaSparkBindings.pdf > > > > > > > > > > > > > > > > > >> On Fri, Sep 26, 2014 at 8:02 PM, thejas prasad < > > [email protected]> > > > > > wrote: > > > > > >> > > > > > >> what exactly in the scala math library? > > > > > >> > > > > > >> > > > > > >> > > > > > >> On Fri, Sep 26, 2014 at 1:00 AM, Ted Dunning < > > [email protected]> > > > > > >> wrote: > > > > > >> > > > > > >>> Got it! > > > > > >>> > > > > > >>> Sorry to be dense. > > > > > >>> > > > > > >>> > > > > > >>> > > > > > >>> On Thu, Sep 25, 2014 at 4:23 PM, Thejas Prasad < > > [email protected]> > > > > > >>> wrote: > > > > > >>> > > > > > >>>> Sorry I meant to say what is the best way to get started**? > > > > > >>>> > > > > > >>>> Thanks, > > > > > >>>> Thejas > > > > > >>>> Sent from my iPhone > > > > > >>>> > > > > > >>>>> On Sep 25, 2014, at 4:28 PM, Ted Dunning < > > [email protected]> > > > > > >>> wrote: > > > > > >>>>> > > > > > >>>>>> On Thu, Sep 25, 2014 at 9:35 AM, Thejas Prasad < > > > > [email protected] > > > > > >>> > > > > > >>>> wrote: > > > > > >>>>>> > > > > > >>>>>> what is the best way to get statues > > > > > >>>>> > > > > > >>>>> > > > > > >>>>> Hmmm.... > > > > > >>>>> > > > > > >>>>> I am totally confused. You must have meant something here. > > > > > >>>>> > > > > > >>>>> Regarding your next question, the place to start work is on > the > > > > scala > > > > > >>>> math > > > > > >>>>> library. > > > > > >> > > > > > > > > > > > > > >
