jamiebullock
commented
10 years ago Firstly, apologies for the lack of coherent documentation—I've been focussing so far on getting the objects in place.
What you describe above is called multi-dimensional regression.
To do this with ml.mlp, it needs to be in "regression mode". This means that rather than giving you a single class label as an output, it gives you a continuous floating point value—so if the input is "half way" between class 1 and class 2, it will give you an output of 1.5. To put the object in regression mode, set the attribute "mode" to "1", e.g. by sending the message "mode 1"
Secondly you need to set the "num_outputs" to whatever you want "M" to be in your N to M mapping. So it you want to control a synth with 7 parameters, you would send the message "num_outputs 7".
The rest follows the logic of all the other ml.* objects, you just "add" with your "M" dimensional target vector prepended to your "N" dimensional input vector. So for a 7-dimensional target vector (1 2 3 4 5 6 7) and a 3 dimensional input vector (10 20 30), you would do:
''' add 1 2 3 4 5 6 7 10 20 30 '''
When you are done adding, you "train"
Then
''' classify 10 20 30 '''
Should output
''' 1 2 3 4 5 6 7 '''
batchku
what i imagine for MLP is a system where you associate N-dim input vectors with M-dim output vectors. during training, one would associate a number of samples of N-dimensional vectors to produce a certain M-dim output vector.
this is what wekinator does.
as example would be to use the 3 axis accel data and control a synth that has 7 parameters.
is this how this object works?
do you use "input_activation_function" to provide the object with desired OUTPUT vectors?
or basically: how do you provide ml.mlp with the desired M-dim output vectors that your provided examples of N-dim input vectors should map to?