RapidLib issueshttps://gitlab.doc.gold.ac.uk/rapid-mix/RapidLib/-/issues2017-05-02T14:07:05Zhttps://gitlab.doc.gold.ac.uk/rapid-mix/RapidLib/-/issues/32Build Essentia with Emscripten2017-05-02T14:07:05ZMichael ZbyszyńskiBuild Essentia with EmscriptenWe'd like Essentia as part of our Rapid API. It looks like it could be built with Emscripten, but we're starting from square one.We'd like Essentia as part of our Rapid API. It looks like it could be built with Emscripten, but we're starting from square one.Future APIMichael ZbyszyńskiMichael Zbyszyńskihttps://gitlab.doc.gold.ac.uk/rapid-mix/RapidLib/-/issues/29JSON API inRanges, inBases, outRange, outBase2017-05-02T14:07:05ZMichael ZbyszyńskiJSON API inRanges, inBases, outRange, outBaseI had originally had the API giving inMaxes, etc. This should be changed so the regression object takes Ranges and Bases, which are calculated elsewhere.I had originally had the API giving inMaxes, etc. This should be changed so the regression object takes Ranges and Bases, which are calculated elsewhere.API v0.1Michael ZbyszyńskiMichael Zbyszyńskihttps://gitlab.doc.gold.ac.uk/rapid-mix/RapidLib/-/issues/28Align wekinator JSON and C++ JSON2017-09-20T10:05:07ZMichael ZbyszyńskiAlign wekinator JSON and C++ JSONI ended up having to write slightly different JSON files from Wekinator and C++. This needs to be fixed.I ended up having to write slightly different JSON files from Wekinator and C++. This needs to be fixed.API v0.2Michael ZbyszyńskiMichael Zbyszyńskihttps://gitlab.doc.gold.ac.uk/rapid-mix/RapidLib/-/issues/27Implement .initialize()2017-05-02T14:07:05ZMichael ZbyszyńskiImplement .initialize()There should be a method to reset trained models.
Feature request from @rebeccaf There should be a method to reset trained models.
Feature request from @rebeccaf API v0.1Michael ZbyszyńskiMichael Zbyszyńskihttps://gitlab.doc.gold.ac.uk/rapid-mix/RapidLib/-/issues/25Check why ofx example needed input normalization2017-09-20T10:05:07ZMichael ZbyszyńskiCheck why ofx example needed input normalizationAPI v0.1Michael ZbyszyńskiMichael Zbyszyńskihttps://gitlab.doc.gold.ac.uk/rapid-mix/RapidLib/-/issues/24Expose activation function for output node2017-08-30T12:43:24ZMichael ZbyszyńskiExpose activation function for output nodeIf we want to use our Perceptron as a classifier, we might want the output node to use a sigmoid activation function.
Easy to add... worthwhile?If we want to use our Perceptron as a classifier, we might want the output node to use a sigmoid activation function.
Easy to add... worthwhile?Future APIMick GriersonMick Griersonhttps://gitlab.doc.gold.ac.uk/rapid-mix/RapidLib/-/issues/23Re-implement arguments for Regression() and Classification()2017-09-20T10:05:07ZMichael ZbyszyńskiRe-implement arguments for Regression() and Classification()I broke these while moving this code to C++.I broke these while moving this code to C++.API v0.2Michael ZbyszyńskiMichael Zbyszyńskihttps://gitlab.doc.gold.ac.uk/rapid-mix/RapidLib/-/issues/20Make ofx example just like RapidAPI_0062017-09-20T10:05:07ZMichael ZbyszyńskiMake ofx example just like RapidAPI_006Write this:
http://live.codecircle.com/d/daRnfnruuWaxdYLDn
in ofx.Write this:
http://live.codecircle.com/d/daRnfnruuWaxdYLDn
in ofx.API v0.1Michael ZbyszyńskiMichael Zbyszyńskihttps://gitlab.doc.gold.ac.uk/rapid-mix/RapidLib/-/issues/19Move JS facade into C++2017-09-20T10:05:07ZMichael ZbyszyńskiMove JS facade into C++As much as possible, code needs to be moved out of JS and into C++As much as possible, code needs to be moved out of JS and into C++API v0.1Michael ZbyszyńskiMichael Zbyszyńskihttps://gitlab.doc.gold.ac.uk/rapid-mix/RapidLib/-/issues/17Implement train method in KNN.2017-09-20T10:05:07ZMichael ZbyszyńskiImplement train method in KNN.Currently, training of a knn is wrapped around the addNeighbour function.
1) Implement "train" for real
2) Implement some kind of clear function
Should train clear first, then add? Or just add? How to present these two modes in th...Currently, training of a knn is wrapped around the addNeighbour function.
1) Implement "train" for real
2) Implement some kind of clear function
Should train clear first, then add? Or just add? How to present these two modes in the API?API v0.1Michael ZbyszyńskiMichael Zbyszyńskihttps://gitlab.doc.gold.ac.uk/rapid-mix/RapidLib/-/issues/14Unit tests for C++ code2017-09-07T10:41:32ZMichael ZbyszyńskiUnit tests for C++ codeThis would be good practice.This would be good practice.API v0.2Michael ZbyszyńskiMichael Zbyszyńskihttps://gitlab.doc.gold.ac.uk/rapid-mix/RapidLib/-/issues/13C++ write JSON2017-09-20T10:05:08ZMichael ZbyszyńskiC++ write JSONFuture APIhttps://gitlab.doc.gold.ac.uk/rapid-mix/RapidLib/-/issues/11Set in and out mins and maxes when training.2017-09-20T10:05:08ZMichael ZbyszyńskiSet in and out mins and maxes when training.Right now, the neural network class expects to know the range of inputs and outputs it is expecting when created. Make this be set at the beginning of training.
This could happen in JS, but would be better on the C++ side.Right now, the neural network class expects to know the range of inputs and outputs it is expecting when created. Make this be set at the beginning of training.
This could happen in JS, but would be better on the C++ side.API v0.1Michael ZbyszyńskiMichael Zbyszyńskihttps://gitlab.doc.gold.ac.uk/rapid-mix/RapidLib/-/issues/7Support multiple hidden layers2017-09-20T10:05:10ZMichael ZbyszyńskiSupport multiple hidden layersThe trainer hasn't been coded to handle more than one hidden layer. Fix that.The trainer hasn't been coded to handle more than one hidden layer. Fix that.API v0.1https://gitlab.doc.gold.ac.uk/rapid-mix/RapidLib/-/issues/5Improve neural network training2017-12-12T14:15:00ZMichael ZbyszyńskiImprove neural network trainingI've written neural network training, but it's not the same as Wekinator. Figure out why.
I've written neural network training, but it's not the same as Wekinator. Figure out why.
Future APIMichael ZbyszyńskiMichael Zbyszyńskihttps://gitlab.doc.gold.ac.uk/rapid-mix/RapidLib/-/issues/3C++ read JSON2017-09-20T10:05:10ZMichael ZbyszyńskiC++ read JSONCurrently, Wekinator outputs JSON. We have classes in JS that read the JSON and feed it into classes generated by emscripten. But, there is no native C++ JSON input.Currently, Wekinator outputs JSON. We have classes in JS that read the JSON and feed it into classes generated by emscripten. But, there is no native C++ JSON input.API v0.1Michael ZbyszyńskiMichael Zbyszyńskihttps://gitlab.doc.gold.ac.uk/rapid-mix/RapidLib/-/issues/2C++ documentation2017-09-20T10:05:10ZMichael ZbyszyńskiC++ documentationAll C++ CPI calls should be documented using DOxygen.All C++ CPI calls should be documented using DOxygen.API v0.1Michael ZbyszyńskiMichael Zbyszyński