“ASR:2015-08-17”版本间的差异

来自cslt Wiki
跳转至: 导航搜索
Speech Processing
Lr讨论 | 贡献
financial group
 
第129行: 第129行:
  
 
===strategy===
 
===strategy===
* rule optimize model
+
* rule optimize model (hold)
 
:* Adaboost method  
 
:* Adaboost method  
 
* technology  index
 
* technology  index
:* survey the current technology index
+
:* survey the current technology index and test several type technology index
 
* NN
 
* NN
 
:* RNN model using Theano
 
:* RNN model using Theano
 
+
*
 
===display platform===
 
===display platform===
 
* set up the test platform in our grid
 
* set up the test platform in our grid

2015年8月21日 (五) 06:06的最后版本

Speech Processing

AM development

Environment

  • grid-14 is on repairation
  • prepare to buy a server

RNN AM

  • train monophone RNN --zhiyuan
  • train using large dataset--mengyuan
  • write code to tune learning rate--zhiyong

Mic-Array

  • hold
  • compute EER with kaldi

====Data selection unsupervised learning

  • hold
  • acoustic feature based submodular using Pingan dataset --zhiyong
  • write code to speed up --zhiyong


RNN-DAE(Deep based Auto-Encode-RNN)

  • hold
  • deliver to mengyuan, xuewei

Speaker ID

  • DNN-based sid --Lantian


Ivector&Dvector based ASR

  • hold --Tian Lan
  • Cluster the speakers to speaker-classes, then using the distance or the posterior-probability as the metric
  • dark-konowlege using i-vector
  • train on wsj(testbase dev93+evl92)
  • --hold

language vector

  • hold
  • train using language vector with the dataset of 1400h_CN + 100h_EN--mengyuan
  • hold
  • write a paper--zhiyuan
  • RNN language vector
  • train as a paper--xuewei

rectifier

  • hold
  • rectifier RNN --zhiyuan
  • hold


multi-GPU=

  • multi-stream training --Sheng Su

Text Processing

RNN LM

  • character-lm rnn(hold)
  • lstm+rnn
  • check the lstm-rnnlm code about how to Initialize and update learning rate.(hold)

Neural Based Document Classification

  • (hold)

RNN Rank Task

  • Paper: RNN Rank Net.
  • (hold)

Graph RNN

  • Entity path embeded to entity.
  • (hold)

RNN Word Segment

  • Set bound to word segment.
  • (hold)

Seq to Seq(09-15)

  • Review papers.
  • Reproduce baseline. (08-03 <--> 08-17)

Order representation

  • Nested Dropout
  • semi-linear --> neural based auto-encoder.
  • modify the objective function(hold)

Balance Representation

  • Find error signal

Recommendation

  • Reproduce baseline.
  • LDA matrix dissovle.
  • LDA (Text classification & Recommendation System) --> AAAI

RNN based QA

  • Read Source Code.
  • Attention based QA.
  • (hold)

RNN Poem Process

  • Seq based BP.
  • (hold)

Text Group Intern Project

Buddhist Process

  • (hold)

RNN Poem Process

  • Done by Haichao yu & Chaoyuan zuo Mentor : Tianyi Luo.

RNN Document Vector

  • (hold)

Image Baseline

  • Demo Release.
  • Paper Report.
  • Read CNN Paper.

Text Intuitive Idea

Trace Learning

  • (Hold)

Match RNN

  • (Hold)

financial group

tonglian platform

  • learn the platform
  • arma,ar,boosting tree is done

strategy

  • rule optimize model (hold)
  • Adaboost method
  • technology index
  • survey the current technology index and test several type technology index
  • NN
  • RNN model using Theano

display platform

  • set up the test platform in our grid