A Data Science Central Community
just an idea.
as part of course which I am attending, today teacher taught SOM (self organizing maps)
he used colors as one of the example to explain it.
The thought came as follows:
Bring a color expert and a domain expert together.
look at all the data variables, reduce it,.............
for each variable decide a color, the range of values/shades for that color can be the values of the variable.
normalize based on the color scale which can be used to fit the required values and variables.
the sum of color for all the variables should give a color in such a way that it explains the clusters and distance between two vectors.
something different from cosine, manhattan, euclidean..............
apply MC / MCMC a transition matrix of notes can be created and using this technique a music can be generated. matrix is generated using existing music/songs