Relationship between variables

[Christopher Turnbull]

Hi all.

I am quite new to machine learning. My background is pure mathematics.

Now, I'd like to use caffe to discover a relationship between the following readings on air sensors:

Humidity, temperature, wind speed, and the actual sensor readings of gas concentrations

I want to discover an equation as to how they're linked! 

The good thing is I have access to a lot of data...

The bad thing is I am really confused- I didn't find the tutorials for caffe particularly useful for what I want to do.

Can anyone shed light on where this equation will pop out, and how I should feed my data in?

I realize this is quite an open ended question as I haven't provided specific details. But if anyone can point me in the right direction here, I'd really appreciate it

[Jan]

Well, I am not sure Deep Learning tools are fit for learning relations expressible by (simple!) equations. To describe the functional operation of a complete convolutional network with an equation should be possible in theory, while next to impossible and very ugly in practice (the equation alone would probably fill a large number of pages). If you're ok with the fact that your "equation" is basically a sequence of filtering operations, nonlinear function applications and matrix multiplications (describe by the network config, usually in a .prototxt file) with a specific set of values (the trained weights, given in a .caffemodel file), then by all means use Caffe or some other deep learning tool. Other than that no equation will pop out anywhere. And there is no symbolic computation going on, everything is numeric.

I don't know how much background you have in machine learning, but to get accustomed to the whole paradigm maybe you should first have a look at the simple things like ordinary logistic regression before diving in full into deep learning.

Jan