It should be possible to create a new classifier only by giving a CSV file with symbol names (or IDs in write-math.com or Unicode). Then it should know how many output nodes are required, adjust the currently best models topology, get the latest data and train the model.
It should be possible to create a new classifier only by giving a CSV file with symbol names (or IDs in write-math.com or Unicode). Then it should know how many output nodes are required, adjust the currently best models topology, get the latest data and train the model.