2013-07-26

来自cslt Wiki
2013年7月26日 (五) 08:19Wangd讨论 | 贡献的版本

跳转至: 导航搜索

Data sharing

  • LM count files still undelivered!

DNN progress

Experiments

  • Sparse DNN on the ARM board
1200-1200-1200-3536                 1200-1200-1200-3536-sparse0.3 (sparsity 1/5)
original atlas:  RT 2.3                         RT 2.3
atlas sparse:    RT 54                          RT 14  
NIST smatmat:    RT 27.3                        RT 5.98
800-800-800-2108                    800-800-800-2108-sparse0.3 (sparsity 2/5):
original atlas: RT 1.3                          RT 1.1
NIST smatmat:   RT 11.9                         RT 5.5

600-600-600-1500 
original atlas: RT 0.9
NIST smatmat:   RT 6.5


To be done:

# SuiteSparse lib
# decrease the hidden layer from 4-2
# Test accuracy on large data set

Tencent exps

GPU & CPU merge

  1. Hold

Confidence estimation

1. DNN-based confidence estimation for decoding and alignment is done. The intuitive testing seems ok.

To DO:

  1. Large scale test.
  2. CI Phone-based full path confidence estimation.

FST sub-graph integration

  1. Coding finished.
  2. Debuging and Test next week.


Embedded progress

  1. Graph search costs a lot when the graph is large. Need to improve the indexing performance.
  2. Need to integrate the Kaldi FE with pocket-sphinx decoder.