1 Feature -> 5 Labels training

Question

Dear experts,
I am a new user of ML.NET and I am not clear about how to perform my needs.
My environment is the following :
windows 10, .NET Core 3.1 LTS and latest Visual Studio.

What I want is to train a ML to get 5 Labels form 1 Feature. From the tutorial I have seen the example of 1 Feature -> 1 Label training and I am not sure if it is possible to extend the Text classification for my case.

This is a simple example dataset of what I want to achieve :

1/ Feature = The cat lenght is 30cm
-> Animal=The cat / Extra= / Prop=lenght is / quantity=30 / unit=cm

2/ Feature = The dog density is 100g/cm3
-> Animal=The dog / Extra= / Prop=density is / quantity=100 / unit=g/cm3

3/ Feature = The horse number 2 speed is 45km/h
-> Animal=The horse / Extra=number 2 / Prop=speed is / quantity=45 / unit=km/h

1 Feature input -> 5 Labels to determine : Animal, Extra, Prop, quantity, unit.
I have a list of Feature and Labels into my existing dataset, then I want to apply the result of the training on whatever new Feature not in my dataset and find the 5 Labels above.

I am not able to associate more than 1 Label to the columns of my uploaded dataset when I use the Text classification scenario.
Is there a simple way to get a multiple Labels Training ? I have seen multiclass classification mentionned into your website but I have not been able to find it in the training scenario choice ?

In the mean time, I can have a predefined list of Animal (The cat, The dog, The horse, ...) which is not containing all the existing animals. This list can be used into the training to help the algorithm. Or this list can be used when we evaluate the answer to find a predefined answer to the Animal Label.
How can I give this list into the input training?

Best regards,
David.

Answer 1

Topic abandoned