Hi, Thank you very much Chathurike.It will be really helpful for my NN Builder.
Thank you. On Fri, Jun 3, 2016 at 10:50 AM, CD Athuraliya <chathur...@wso2.com> wrote: > Hi Lakini, > > You can gain some insight for your NN builder from TensorFlow Playground > [1]. It is also open source under Apache 2.0 licence [2]. > > [1] http://playground.tensorflow.org > [2] https://github.com/tensorflow/playground > > Regards, > CD > > On Thu, Jun 2, 2016 at 8:00 AM, Upul Bandara <u...@wso2.com> wrote: > >> Yes at the moment this looks OK. >> >> Please put some effort to create a basic version of the NN builder and >> let's have a quick demo. >> >> Let me know if you have further questions or clarifications. >> >> Thanks, >> Upul >> >> On Wed, Jun 1, 2016 at 11:29 PM, Lakini Senanayaka < >> lakinisenanayak...@gmail.com> wrote: >> >>> Hi, >>> >>> Thank you very much Upul. I will submit a demo soon. >>> >>> I have drawn a sketch of NN builder front end[1]. Basically it will be >>> like this UI[2].(The content will be changed.) >>> >>> [1]Sketch of NN Builder >>> <https://docs.google.com/document/d/1c9F5xECaxuXq65RJaBO33AZZqMN0qNJsqeux3tRfUHE/edit?usp=sharing> >>> [2]https://www.draw.io/ <https://www.draw.io/> >>> >>> In [1],under "layers" users can select the type of the >>> layer-Input,Hidden or Output layer. >>> Under "node" and "Connectors"-users can drag and drop nodes and >>> connectors to the working area and they can build their NN as they wish. >>> >>> In the right hand side,there is a setting area.From that user can set >>> Optimization algorithms,iterations,learning rates,seed etc. of the neural >>> network. >>> >>> Could you please give me comments and your thoughts regrading this? >>> >>> Thank you. >>> >>> >>> >>> On Wed, Jun 1, 2016 at 2:19 PM, Upul Bandara <u...@wso2.com> wrote: >>> >>>> Hi, >>>> >>>> Good progress. >>>> >>>> The key parts of the project are designing the front-end of NN builder, >>>> communicating between the front-end and the Deeplearning4J back-end. >>>> So at this stage of the project, it is better to put your effort on >>>> above the components. Later, we can integrate what you have built with WSO2 >>>> Machine Learning server. >>>> >>>> Once you have completed a basic NN builder (ability to build a simple >>>> feedforward is enough) we would like to have a quick demo. >>>> >>>> Let me know if you have further questions or clarifications. >>>> >>>> Thanks, >>>> Upul >>>> >>>> On Wed, Jun 1, 2016 at 10:42 AM, Lakini Senanayaka < >>>> lakinisenanayak...@gmail.com> wrote: >>>> >>>>> Hi, >>>>> >>>>> I'm sorry for the late response.Thank you very much for the last >>>>> email.I have gone through JQueryUI and I have implemented some samples >>>>> using JQueryUI.I have gone through the WSO2 machine learner and I have >>>>> identified the place to insert this jaggery page(replace the >>>>> hyper-parameter page in the ML UI). >>>>> Currently I'm designing the sketch of the Deep Neural Network >>>>> Builder-the front end. >>>>> I'll send my sketch before tonight.Based on your comments I can start >>>>> implementations. >>>>> >>>>> Thank you. >>>>> >>>>> >>>>> On Wed, Jun 1, 2016 at 9:29 AM, Supun Sethunga <sup...@wso2.com> >>>>> wrote: >>>>> >>>>>> Hi Lakini, >>>>>> >>>>>> Any update on the progress? >>>>>> >>>>>> Regards, >>>>>> Supun >>>>>> >>>>>> On Fri, May 27, 2016 at 5:34 PM, Supun Sethunga <sup...@wso2.com> >>>>>> wrote: >>>>>> >>>>>>> Hi Lakini, >>>>>>> >>>>>>> Sorry for the delayed response. As the first part of the project, >>>>>>> you can start the work on the UI/drag and drop feature, as the core of >>>>>>> your >>>>>>> project is based around that. Therefore, shall we try to get a basic >>>>>>> version of drag and drop UI, by the mid-term evaluation? (No need to >>>>>>> connect the UI with the dl4j, for the first phase) So this would be >>>>>>> include: >>>>>>> >>>>>>> - A link/re-direct to go to Visual builder, upon selecting the >>>>>>> Neural Netowrks (as the algorithm) >>>>>>> - The drawing/dropping area and a panel to pick the objects to >>>>>>> be dragged (objects as in, nodes, layers, links, etc) >>>>>>> >>>>>>> This drag and drop page should replace the hyper-parameter page in >>>>>>> the ML UI. More precisely, In the Machine Learner wizard, when a user >>>>>>> picks >>>>>>> the algorithm name as Neural Network, and proceed, this drag and drop >>>>>>> page >>>>>>> should be prompted instead of the hyper-parameters page. You can create >>>>>>> the >>>>>>> overall page as a jaggery page (.jag). It would be easier for you to >>>>>>> get a >>>>>>> copy of an existing page, and modifying the content. That will preserve >>>>>>> the >>>>>>> existing styles, session handling etc. >>>>>>> >>>>>>> Please feel free to raise any question you come across during >>>>>>> implementing. >>>>>>> >>>>>>> Regards, >>>>>>> Supun >>>>>>> >>>>>>> On Tue, May 24, 2016 at 8:09 PM, Lakini Senanayaka < >>>>>>> lakinisenanayak...@gmail.com> wrote: >>>>>>> >>>>>>>> Hi, >>>>>>>> >>>>>>>> Thank you very much for the suggestions.I will be using JQueryUI. >>>>>>>> >>>>>>>> As my coding period started yesterday according to the GSoC >>>>>>>> Schedule I would like to know the workload that should be completed >>>>>>>> before the midterm evaluation. Please advise me accordingly. >>>>>>>> >>>>>>>> Thank you. >>>>>>>> >>>>>>>> >>>>>>>> On Tue, May 24, 2016 at 10:34 AM, Supun Sethunga <sup...@wso2.com> >>>>>>>> wrote: >>>>>>>> >>>>>>>>> Hi Lakini, >>>>>>>>> >>>>>>>>> Yes you can use any of them, with open source licence. But I would >>>>>>>>> prefer JQueryUI, as its a well known library, and is already been >>>>>>>>> used by >>>>>>>>> some of the wso2 products. >>>>>>>>> >>>>>>>>> Regards, >>>>>>>>> Supun >>>>>>>>> >>>>>>>>> On Tue, May 24, 2016 at 9:29 AM, Lakini Senanayaka < >>>>>>>>> lakinisenanayak...@gmail.com> wrote: >>>>>>>>> >>>>>>>>>> Hi, >>>>>>>>>> >>>>>>>>>> I have found some libraries to build drag and drop UI for our >>>>>>>>>> front end .They are JQury UI[1],Dragula[2],Draggablily[3]. >>>>>>>>>> I hope I can do the developments using these libraries. >>>>>>>>>> >>>>>>>>>> [1]jqueryui <http://jqueryui.com/> >>>>>>>>>> [2]dragula <https://bevacqua.github.io/dragula/> >>>>>>>>>> [3]draggabilly <http://draggabilly.desandro.com/> >>>>>>>>>> >>>>>>>>>> Could you please tell me if you have any other suggestions? >>>>>>>>>> >>>>>>>>>> On Sat, May 21, 2016 at 11:05 PM, Lakini Senanayaka < >>>>>>>>>> lakinisenanayak...@gmail.com> wrote: >>>>>>>>>> >>>>>>>>>>> Hi , >>>>>>>>>>> >>>>>>>>>>> This is my weekly progress update of my project. >>>>>>>>>>> >>>>>>>>>>> I have solved all the problems I have mentioned in the last >>>>>>>>>>> email.I could find a class BaseDatasetIterator which is in DL4J >>>>>>>>>>> where we >>>>>>>>>>> can find inbuilt iterators for CurvesDataSetIterator, >>>>>>>>>>> IrisDataSetIterator, MnistDataSetIterator, >>>>>>>>>>> MovingWindowBaseDataSetIterator, RawMnistDataSetIterator .There >>>>>>>>>>> is no iterator for CIFAR dataset yet. >>>>>>>>>>> >>>>>>>>>>> I have gone through the whole documentation except >>>>>>>>>>> Deeplearning4j on Spark.The latest documentation is very >>>>>>>>>>> understandable >>>>>>>>>>> than the earlier one.I have studied neural networks- Restricted >>>>>>>>>>> Boltzmann Machines, >>>>>>>>>>> Convolutional Nets (ConvNets), >>>>>>>>>>> Long Short-Term Memory Units (LSTMs), >>>>>>>>>>> Denoising Autoencoders, >>>>>>>>>>> Recurrent Nets and LSTMs, >>>>>>>>>>> Multilayer Neural Nets, >>>>>>>>>>> Deep-Belief Network, >>>>>>>>>>> Deep AutoEncoder, >>>>>>>>>>> Stacked Denoising Autoencoders >>>>>>>>>>> >>>>>>>>>>> I have run the example codes[1] and I have understood the >>>>>>>>>>> implementations.I have rerun the codes with modifying different >>>>>>>>>>> parameter >>>>>>>>>>> values like number and size of the hidden layers, learning rate, >>>>>>>>>>> momentum, >>>>>>>>>>> weight distribution and various types of regularization and checked >>>>>>>>>>> the >>>>>>>>>>> performance. >>>>>>>>>>> >>>>>>>>>>> I have implemented a convolution net[3] and trained it using LFW >>>>>>>>>>> dataset[2] and a Recurrent net[4].Still I have a problem in >>>>>>>>>>> vectorizing >>>>>>>>>>> CIFRA-10 data set. >>>>>>>>>>> >>>>>>>>>>> Currently I am researching about libraries to build drag & >>>>>>>>>>> drop UI for front end. >>>>>>>>>>> >>>>>>>>>>> [1]dl4j-0.4-examples >>>>>>>>>>> <https://github.com/deeplearning4j/dl4j-0.4-examples> >>>>>>>>>>> [2]LFW Face Dataset <http://vis-www.cs.umass.edu/lfw/> >>>>>>>>>>> [3]ConvolutionNetLFW >>>>>>>>>>> <https://github.com/Lakini/Deeplearning4Java/blob/master/src/main/java/org/deeplearning4j/examples/convolution/ConvolutionNetLFW.java> >>>>>>>>>>> [4]RecurrentNetworkExample >>>>>>>>>>> <https://github.com/Lakini/Deeplearning4Java/blob/master/src/main/java/org/deeplearning4j/examples/recurrent/basic/RecurrentNetworkExample.java> >>>>>>>>>>> >>>>>>>>>>> >>>>>>>>>>> Could you please guide me to do the next step of my project . >>>>>>>>>>> >>>>>>>>>>> Thank you. >>>>>>>>>>> >>>>>>>>>>> Sincerely, >>>>>>>>>>> *Lakini Senanayaka.* >>>>>>>>>>> >>>>>>>>>>> >>>>>>>>>> >>>>>>>>>> >>>>>>>>>> -- >>>>>>>>>> Thank you. >>>>>>>>>> >>>>>>>>>> Sincerely, >>>>>>>>>> *Lakini Senanayaka.* >>>>>>>>>> >>>>>>>>>> >>>>>>>>> >>>>>>>>> >>>>>>>>> -- >>>>>>>>> *Supun Sethunga* >>>>>>>>> Software Engineer >>>>>>>>> WSO2, Inc. >>>>>>>>> http://wso2.com/ >>>>>>>>> lean | enterprise | middleware >>>>>>>>> Mobile : +94 716546324 >>>>>>>>> >>>>>>>> >>>>>>>> >>>>>>>> >>>>>>>> -- >>>>>>>> Thank you. >>>>>>>> >>>>>>>> Sincerely, >>>>>>>> *Lakini Senanayaka.* >>>>>>>> >>>>>>>> >>>>>>> >>>>>>> >>>>>>> -- >>>>>>> *Supun Sethunga* >>>>>>> Software Engineer >>>>>>> WSO2, Inc. >>>>>>> http://wso2.com/ >>>>>>> lean | enterprise | middleware >>>>>>> Mobile : +94 716546324 >>>>>>> >>>>>> >>>>>> >>>>>> >>>>>> -- >>>>>> *Supun Sethunga* >>>>>> Software Engineer >>>>>> WSO2, Inc. >>>>>> http://wso2.com/ >>>>>> lean | enterprise | middleware >>>>>> Mobile : +94 716546324 >>>>>> >>>>> >>>>> >>>>> >>>>> -- >>>>> Thank you. >>>>> >>>>> Sincerely, >>>>> *Lakini Senanayaka.* >>>>> >>>>> >>>> >>>> >>>> -- >>>> Upul Bandara, >>>> Associate Technical Lead, WSO2, Inc., >>>> Mob: +94 715 468 345. >>>> >>> >>> >>> >>> -- >>> Thank you. >>> >>> Sincerely, >>> *Lakini Senanayaka.* >>> >>> >> >> >> -- >> Upul Bandara, >> Associate Technical Lead, WSO2, Inc., >> Mob: +94 715 468 345. >> >> _______________________________________________ >> Dev mailing list >> Dev@wso2.org >> http://wso2.org/cgi-bin/mailman/listinfo/dev >> >> > > > -- > *CD Athuraliya* > Software Engineer > WSO2, Inc. > lean . enterprise . middleware > Mobile: +94 716288847 <94716288847> > LinkedIn <http://lk.linkedin.com/in/cdathuraliya> | Twitter > <https://twitter.com/cdathuraliya> | Blog > <https://cdathuraliya.wordpress.com/> > -- Thank you. Sincerely, *Lakini Senanayaka.*
_______________________________________________ Dev mailing list Dev@wso2.org http://wso2.org/cgi-bin/mailman/listinfo/dev