5 ms·
Just saw your talk. Man you are one hell of a speaker. Great slides, animations and presentation. I enjoyed it. It seems like a strange thing that no one thoug
by 110011 9y ago
Just saw your talk. Man you are one hell of a speaker. Great slides, animations and presentation. I enjoyed it.
It seems like a strange thing that no one thought of doing this before! Nice that you identified a cool problem and solved it :)
- scottlegrand2 9y agoNot entirely true. Sparse x Dense Matrix is an important component of the DSSTNE deep learning framework: https://github.com/amzn/amazon-dsstne https://github.com/amzn/amazon-dsstne But this library is clearly more thorough.
- fredrikbk 9y agoFor sure, as you point out, many such kernels have been written, going at least as far back as 1967. We believe our contribution is being able to generate kernels for all the expressions. Thanks for the reference though! We’re trying to learn where sparsity may be important in neural networks.
- fredrikbk 9y agoThanks, that is really nice of you :)