Deep Learning for accelerometer image data of x, y, z axis

Hi All,

I want to predict different diseases by using the images of the acceleration signal instead of using the numerical data itself. The data constitute of different walking activities collected with acceleometer sensor and labled with diseases type. Any algorithm, model or method suggestions for predictions will be highly appreciated. Both supervised and unsupervised learning could be applied. Please advise me in detail from implementation point of view. Thanks in advance.

Can you upload a sample data set to GIT so that we can help you?

@QuantScientist I am not allowed to do so due to data sharing polices. The data is based on the plots of the acceleration data (x,y,z). Each plot is also labeled with different diseases type. I will consider those plots as images for training. Please advise me if you have any idea. I will appreciate your time and effort.

If the classification is Binary, then take a look at my example here:

The data is numeric (floating point) so it may be regarded as your accelerometer data.

Thank you so much @QuantScientist. But my classification is not binary and I am having 4 different type of classifications and want to train the models based on the images (plots of acceleration data).