How to find common characteristics of the three matrices?

Question

I have three matrices A, B and C, the size of which are all 120*1000 double, where 120 represents the number of time points and 1000 represents the total number of features. For each matrix, there is a corresponding regressor matrix, the size of which are all 120*5 double. The regressor matrices only contain "1" and "0", where "1" represents there is a stimulus in this time point and "0" represents rest time points. I want to find the common characteristics of the three matrices A, B and C combined with the three regressor matrices. Then I want to train a classifier based on matrices A and B. In the end, I want to classify matrix C based on the training data. How to realize it? Thank you!

why deduct my points? Do I ask this question in a improper way? — Poppy, Jun 15 '16 at 08:19
Two questions that would need to be answered before this becomes a question that is suitable for this site: 1. What did you try? 2. What exactly do you need, things like 'common characteristics' could mean anything. Please provide (possibly small scale) examples of what exactly you are trying to achieve. — Dennis Jaheruddin, Jun 15 '16 at 08:20
@DennisJaheruddin I thought if I want to classify matrix C based on training matrices A and B, I should find out the common characteristics of matrices A, B and C,. Am I wrong? — Poppy, Jun 15 '16 at 08:24
I really don't understand anything of what you are trying to explain or achieve. You are using the machine learning toolbox or something? http://nl.mathworks.com/solutions/machine-learning/ — Bernhard, Jun 15 '16 at 08:35
@Bernhard, I want to achieve multiple subjects classification. Here each matrix represents one subject. I am not using machine learning toolbox or something. I just do not know how to realize it. — Poppy, Jun 15 '16 at 08:37
I assume A,B and C are your 3 objects? and each object has 120 observations of its feature vector (which is 1000 in length)? What is this regressor matrix, how is it related to the feature vectors? — GameOfThrows, Jun 15 '16 at 08:46
how would you define the characteristics they have in common? same amount of stimuli, stimuli and the same position or something else? how would you train that classicfier and what kind of classicfier are we talking about? — Finn, Jun 15 '16 at 08:52
@GameOfThrows, yes, A,B and C are your 3 objects and each object has 120 observations of its feature vector (which is 1000 in length). The regressor matrix is 120*5 double, where 120 represents observations, and 5 represents 5 different stimulus categories. — Poppy, Jun 15 '16 at 08:53
@GameOfThrows, this question is in the area of fMRI. regressor matrix represents the external stimuli, where "1" represents there is an external stimulus in that time point, and "0" represents there is no stimulus in that time point. — Poppy, Jun 15 '16 at 08:55
why don't you add all these details in your question and make a coherent story? — percusse, Jun 15 '16 at 08:58
@Finn, I thought I should use some strategies to select features from the 3 subjects combined with the regressor matrices. I think I will use SVM classifier to train A and B, and then test C. This is in the area of fMRI — Poppy, Jun 15 '16 at 08:59
as proposed by @percusse please add all information from the comments to the question above and please upload example files. i feel like this would really speed up the process. please also describe the strategies you have in mind — Finn, Jun 15 '16 at 09:29

score 0 · Accepted Answer · answered Jun 19 '16 at 15:54

I was hoping that someone more qualified would step in but it looks like the lack of specific info from OP side is discouraging all of them not to answer. My comments were meant as a guideline not as answer but as requested moved my comment to answer.

First of all this is nowhere near my cup of tea so handle with extreme prejudice but:

if the features/subjects are not related

then you should handle each as separate 1D function/array/vector and train your neural network classifier (one for each feature).
if the features are dependent on each-other

then you need to use all of them as an input to your neural network classifier and have network architecture with large enough amount of nodes (wights) capable of handling such amount of data.
you need to find the dependency your self only if you want to reduce the input to classifier

but as you are going for neural network you do not need to as the neural network tends to do it itself. Of coarse if you do it will reduce the needed architecture complexity.

Anyway if you really need to do it then PCA Principal Component Analysis is your way... This step is usually done for deterministic based classifiers (not neural network ones, for example based on correlation coefficients,or based on distance in any metric etc...). PCA has the advantage that you do not need to know too much about the data ... All other reduction approaches I know of usually exploit some feature of the dependency or data but for that you would need to know the properties of input in high detail which I assume is not the case.

Do you have any Matlab Code (so I can learn from the Code at the beginning) about network architecture with large enough amount of nodes (wights) capable of handling such amount of data. — Poppy, Jun 24 '16 at 08:10
@Poppy As I mentioned this is not my cup of tea ... I do not use MATLAB nor neural networks Instead I am more comfortable with deterministic approaches in C++ . For now I was able always avoid them by different means. My favorite is the correlation coefficient. The network architecture is matter of feeling/experience and testing. You choose architecture type and size ... train it and test it if the result is OK you leave it if not decide if the traing set is OK or size or type of architecture ... + trial&error until it works. Which is nowhere near my way of doing things so I avoid it if I can — Spektre, Jun 24 '16 at 08:18

How to find common characteristics of the three matrices?

1 Answers1