Peter released a technical report (available from arxiv) on the computation of the Tucker decomposition on large tensors: the Tucker decomposition is just a multidimensional generalization of the Singular Value Decomposition (SVD). The report includes a new algorithm designed by Peter which is more accurate than competing Matlab implementations, in the case where you have very large tensors (3 or 4 dimensional) and need external memory computations.

There exist incremental SVD algorithms. It does seem to me that a nice property of Turney’s tensor algorithm is that it can be made part of an incremental scheme efficiently.

Another challenge would be to have a serious look at parallel implementations. I think that Turney’s scheme could benefit tremendously from several processors.

Related posts (automatically generated):

No Comments »

No comments yet.

Leave a comment

Warning: When entering a long comment, please ensure that you make copy of your text prior to submitting it. If the server should fail or if you hit a bug, you might lose your work. I am not responsible for your lost effort.

To spammers: I carefully review every single post and make sure that spam gets deleted. You are wasting your time if you are manually entering spam using this form. Read my terms of use to see what I consider to be abusive.

Example: duo plus septem is '9'. The numbers are expressed in latin numerals but you should give your answers using ordinary digits.

 

« « Previous: My research process
Next: Directed research is useless » »

« Blog's main page

Powered by WordPress

© 2004-2012, (lemire at gmail dot com). This work is licensed under a Creative Commons License.