GPU-Accelerated Forward-Backward algorithm with Application to Lattice-Free MMI
Abstract
We propose to express the forward-backward algorithm in terms of operations between sparse matrices in a specific semiring. This new perspective naturally leads to a GPU-friendly algorithm which is easy to implement in Julia or any programming languages with native support of semiring algebra. We use this new implementation to train a TDNN with the LF-MMI objective function and we compare the training time of our system with PyChain - a recently introduced C++/CUDA implementation of the LF-MMI loss. Our implementation is about two times faster while not having to use any approximation such as the "leaky-HMM".
- Publication:
-
arXiv e-prints
- Pub Date:
- October 2021
- DOI:
- 10.48550/arXiv.2112.00709
- arXiv:
- arXiv:2112.00709
- Bibcode:
- 2021arXiv211200709O
- Keywords:
-
- Computer Science - Distributed;
- Parallel;
- and Cluster Computing;
- Computer Science - Computation and Language
- E-Print:
- Submitted to ICASSP 2022