[Enhancement] Code for computing Kernel matrixes?

[WurmD]

Hi Loris,

Would you upload the code used to create the kernel matrixes?

The code so far is really well commented :), but when adapting it to apply it to the HDA dataset, I probably did something wrong, and I think it may be in the Kernel matrix computation. (I adapted this code to the HDA dataset and got lousy results. I then adapted our old MultiView code to the HDA dataset and got great results)

[lorisbaz]

Hi Dario,

The code is supposed to work with any set of kernel matrices that you provide.

Note that the code published by Figueira (that is you) et al. for re-identification was optimized for the task. Here is the paper: http://www.lorisbazzani.info/papers/proceedings/Figueiraetal_AVSS13.pdf

I would expect that you obtained different results for your task using this general version of the code. Have a look at the original ICML paper where we tested this new framework.

I would suggest to publish your multi-view code for re-identification, if someone wants to compare with your paper.

Thanks.