Hi all, I'm doing some tests with several datasets and most of them seemed to be working fine. Somehow, I stumbled upon the leaf dataset ( https://archive.ics.uci.edu/ml/datasets/Leaf), which does not seem to be working well for. However, the dataset works fine with other algorithms (e.g. Logistic Regression L-BFGS) Therefore, I suspect this is due to some sort of malformed data format. I'm right now looking into that.
Furthermore, I am thinking of starting with the D3 visualization on the parameter setting stage. Should we be moving forward with that idea? Finally, I would like to remind that, we haven't decided a date for code review. Should we do that? Thank you On Wed, Jul 22, 2015 at 11:13 PM, Thushan Ganegedara <thu...@gmail.com> wrote: > Hi, > > Apologies about the late reply. > > Notes of the Demonstration > > Time duration: approximately 30 mins > > The demonstration was to demonstrate the implemented deeplearning feature > of WSO2-ML. The demo started first explaining the dataset used (i.e. > MNIST). The dataset is a CSV file with approximately 30000 rows and 784 > features. > > Next the dataset was loaded to WSO2-ml. Here a concern was raised > regarding selecting the type of data in the Preprocessing Phase (i.e. > Categorical vs Numerical) The suggestion was that there should be a UI > feature to change the data type for all the variables at once (very useful > for large amounts of features). > > Next the deeplearning algorithm for MNIST dataset was demonstrated and was > able to achieve an appx 95% accuracy. Regarding the deeplearning > algorithms, H-2-O doesn't seem to have different deeplearning algorithms at > the moment, but a general deep network + classifier (probably autoencoder). > So the idea was to ask H-2-O team whether they are planning to implement > different networks in the future. > > Also, it was suggested to add a visualization feature in parameter setting > stage to provide a summarized visualization of the network to the user. > > Furthermore, another suggestion was to test the deep network on real world > datasets and see how it performs. For this datasets from Kaggle will be > used. > > > About progress. > > I'm currently testing the algorithm against different datasets. and I'll > provide a detailed report on that in the recent future. > > Thank you > > > > On Wed, Jul 22, 2015 at 2:00 PM, Nirmal Fernando <nir...@wso2.com> wrote: > >> @Thushan how are you progressing? Could you please send the notes of our >> last review? >> >> On Thu, Jul 16, 2015 at 10:43 AM, CD Athuraliya <chathur...@wso2.com> >> wrote: >> >>> >>> >>> On Mon, Jul 13, 2015 at 11:12 AM, Thushan Ganegedara <thu...@gmail.com> >>> wrote: >>> >>>> Hello CD, >>>> >>>> Yes, it seems to be working fine now. But why does it show the axes in >>>> meters? Is this a d3 specific thing? >>>> >>> >>> I think *m* stands for *Milli* here. >>> >>>> >>>> On Mon, Jul 13, 2015 at 3:17 PM, Thushan Ganegedara <thu...@gmail.com> >>>> wrote: >>>> >>>>> Hi all, >>>>> >>>>> Thank you very much for pointing out. I'll get the latest update and >>>>> see. >>>>> >>>>> On Mon, Jul 13, 2015 at 3:03 PM, CD Athuraliya <chathur...@wso2.com> >>>>> wrote: >>>>> >>>>>> Hi Thushan, >>>>>> >>>>>> That method has been updated. Please get the latest. You might have >>>>>> to define your own case depending on predicted values. >>>>>> >>>>>> CD Athuraliya >>>>>> Sent from my mobile device >>>>>> On Jul 13, 2015 10:24 AM, "Nirmal Fernando" <nir...@wso2.com> wrote: >>>>>> >>>>>>> Great work Thushan! On the UI issues, @CD could help you. AFAIK >>>>>>> actual keeps the pointer to the actual label and predicted is the >>>>>>> probability and predictedLabel is after rounding it using a threshold. >>>>>>> >>>>>>> On Mon, Jul 13, 2015 at 7:14 AM, Thushan Ganegedara < >>>>>>> thu...@gmail.com> wrote: >>>>>>> >>>>>>>> Hi all, >>>>>>>> >>>>>>>> I have integrated H-2-O deeplearning to WSO2-ml successfully. >>>>>>>> Following are the stats on 2 tests conducted (screenshots attached). >>>>>>>> >>>>>>>> Iris dataset - 93.62% Accuracy >>>>>>>> MNIST (Small) dataset - 94.94% Accuracy >>>>>>>> >>>>>>>> However, there were few unusual issues that I had to spend lot of >>>>>>>> time to identify. >>>>>>>> >>>>>>>> *FrameSplitter does not work for any value other than 0.5. Any >>>>>>>> value other than 0.5, the following error is returned* >>>>>>>> (Frame splitter is used to split trainingData to train and valid >>>>>>>> sets) >>>>>>>> barrier onExCompletion for >>>>>>>> hex.deeplearning.DeepLearning$DeepLearningDriver@25e994ae >>>>>>>> java.lang.RuntimeException: java.lang.RuntimeException: >>>>>>>> java.lang.NullPointerException >>>>>>>> at >>>>>>>> hex.deeplearning.DeepLearning$DeepLearningDriver.trainModel(DeepLearning.java:382) >>>>>>>> >>>>>>>> *DeepLearningModel.score(double[] vec) method doesn't work. * >>>>>>>> The predictions obtained with score(Frame f) and score(double[] v) >>>>>>>> is shown below. >>>>>>>> >>>>>>>> *Actual, score(Frame f), score(double[] v)* >>>>>>>> 0.0, 0.0, 1.0 >>>>>>>> 1.0, 1.0, 2.0 >>>>>>>> 2.0, 2.0, 2.0 >>>>>>>> 2.0, 1.0, 2.0 >>>>>>>> 1.0, 1.0, 2.0 >>>>>>>> >>>>>>>> As you can see, score(double[] v) is quite poor. >>>>>>>> >>>>>>>> After fixing above issues, everything seems to be working fine at >>>>>>>> the moment. >>>>>>>> >>>>>>>> However, the I've a concern regarding the following method in >>>>>>>> view-model.jag -> function >>>>>>>> drawPredictedVsActualChart(testResultDataPointsSample) >>>>>>>> >>>>>>>> var actual = testResultDataPointsSample[i].predictedVsActual.actual; >>>>>>>> var predicted = >>>>>>>> testResultDataPointsSample[i].predictedVsActual.predicted; >>>>>>>> var labeledPredicted = labelPredicted(predicted, 0.5); >>>>>>>> >>>>>>>> if(actual == labeledPredicted) { >>>>>>>> predictedVsActualPoint[2] = 'Correct'; >>>>>>>> } >>>>>>>> else { >>>>>>>> predictedVsActualPoint[2] = 'Incorrect'; >>>>>>>> } >>>>>>>> >>>>>>>> why does it compare the *actual and labeledPredicted* where it >>>>>>>> should be comparing *actual and predicted*? >>>>>>>> >>>>>>>> Also, the *Actual vs Predicted graph for MNIST show the axis in >>>>>>>> "Meters" *(mnist.png) which doesn't make sense. I'm still looking >>>>>>>> into this. >>>>>>>> >>>>>>>> Thank you >>>>>>>> >>>>>>>> >>>>>>>> >>>>>>>> -- >>>>>>>> Regards, >>>>>>>> >>>>>>>> Thushan Ganegedara >>>>>>>> School of IT >>>>>>>> University of Sydney, Australia >>>>>>>> >>>>>>> >>>>>>> >>>>>>> >>>>>>> -- >>>>>>> >>>>>>> Thanks & regards, >>>>>>> Nirmal >>>>>>> >>>>>>> Associate Technical Lead - Data Technologies Team, WSO2 Inc. >>>>>>> Mobile: +94715779733 >>>>>>> Blog: http://nirmalfdo.blogspot.com/ >>>>>>> >>>>>>> >>>>>>> >>>>> >>>>> >>>>> -- >>>>> Regards, >>>>> >>>>> Thushan Ganegedara >>>>> School of IT >>>>> University of Sydney, Australia >>>>> >>>> >>>> >>>> >>>> -- >>>> Regards, >>>> >>>> Thushan Ganegedara >>>> School of IT >>>> University of Sydney, Australia >>>> >>> >>> >>> >>> -- >>> *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 >>> <http://cdathuraliya.tumblr.com/> >>> >> >> >> >> -- >> >> Thanks & regards, >> Nirmal >> >> Associate Technical Lead - Data Technologies Team, WSO2 Inc. >> Mobile: +94715779733 >> Blog: http://nirmalfdo.blogspot.com/ >> >> >> > > > -- > Regards, > > Thushan Ganegedara > School of IT > University of Sydney, Australia > -- Regards, Thushan Ganegedara School of IT University of Sydney, Australia
_______________________________________________ Dev mailing list Dev@wso2.org http://wso2.org/cgi-bin/mailman/listinfo/dev