Hi Marco, that sounds really interesting - I just briefly read through an article and skimmed through the paper. Besides Philipps question, if I understood it correctly, Clustream’s design is an hybrid approach combining an online & offline phase for being able to deal with data evolution over time.
While the online part suits the way StreamPipes operates quite well, I’m curious if you have any idea yet on how to integrate the offline phase as part of the current processor design? Patrick > Am 27.04.2020 um 09:06 schrieb Philipp Zehnder <[email protected]>: > > Hi Marco, > > yes, Clustream is a very good for the integration. > > I was wondering what parameters the user must enter? Probably the number of > clusters. What else? Can we describe them in a way that you don't have to be > a data science expert to use the algorithm? > > I also wondered what the output event from the processor would look like? Do > you already have an idea? > > Philipp > > >> On 27. Apr 2020, at 08:55, Marco Heyden <[email protected]> wrote: >> >> Hey Patrick and Philipp, >> >> I think I will start integrating Clustream. If possible, using MOA __ >> On the long run, I'd also like to add my own research, but we will see __ >> >> Best >> Marco >> >> Am 23.04.20, 11:21 schrieb "Patrick Wiener" <[email protected]>: >> >> Hi Marco, >> >> welcome on board :) Sounds interesting and def gonna be a great add on to >> the toolbox. >> >> Ha! Philipp was quicker than me - I was about to ask the same. Do you have >> any concrete use cases in mind? >> >> Patrick >> >>> Am 23.04.2020 um 11:20 schrieb Philipp Zehnder <[email protected]>: >>> >>> >>> Hi Marco, >>> >>> welcome to the mailing list, good to have you on board. >>> >>> What kind of streaming machine learning algorithms would you like to >>> integrate? >>> >>> Cheers, >>> Philipp >>> >>>> On 23. Apr 2020, at 09:52, Marco Heyden <[email protected]> wrote: >>>> >>>> Hey guys, >>>> >>>> >>>> >>>> my name is Marco and I am a Research Scientist at FZI. >>>> >>>> My research interests are streaming algorithms, federated learning and >>>> anytime algorithms. >>>> >>>> I’m looking forward to contributing to StreamPipes. >>>> >>>> >>>> >>>> Best >>>> >>>> Marco >>>> >>> >>> >> >> >> >> > >
