Can a convolution matrix be developed to perform a side view detection and identification based on ears as weak classifieds?

Let F be a field.

Consider U, the set of n times n strictly upper triangular matrices in F.

For X, Y in U, we call them similar if there exists some S, which is non-degenerate and upper triangular, such that

X= SYS^-1

My question is, how to describe U/~, ~ being the similairy relation defined as above.

I guess there should be some standard/cannonical matrices in U that classify all others.

And then I am wondering the algorithm --- how to find the equivalence class of X, or the standard form of X?

Any answer, reference, or hints are welcome.

I have a (data set) matrix of (0 1) data, which is generated from the noisy pixels. Now I want to classify it as "noisy" or "noise free".(I have labels matrix generated by the comparison of original image and noisy images). How I can do this? Please help

My research area is 'face recognition for biometric purposes'.

I have a confusion matrix (15X15) generated by SVM classifier. And I want to plot the ROC curve from the same.

When a matrix is structured (e.g. tridiagonal matrix with the same stencil) using convolution instead of direct matrix-vector multiplication remarkable improve the performance. But if the matrix is not structured, is there any algorithm (not convert sparse matrix to form like CSR or COO) to do matrix-vector multiplication to get better performance on GPU

I have built a one-dimension network and it did well to classify my target problem. The filter size was 5, 7, 9, and 11 for the convolution layers but someone said it is not normal to increase the size of the kernel while you are going deep the layers. please help me if you can.

Can someone explain as to how I can prep my data for using it with CRIS classifier in R?I have the normalized matrix files, phenotype files and gpl files.

However I'm unsure on how to obtain the "gene expression data matrix in linear format defined as .gct file, and the first colmun annotated to gene Symbol," as stated in the CRIS classifier vignette.

Any help with this would be very much appreciated. Many thanks!

I want to conduct a series of 1-dimensional (1D) convolutional transformations of sequential features (sequence profile, predicted secondary structure and solvent accessibility) to 2-dimensional (2D) matrix by outer concatenation or outer product. How will i do it?