“Dongxu Zhang 2015-08-31”版本间的差异

来自cslt Wiki
跳转至: 导航搜索
Work done in this week
Work done in this week
第3行: 第3行:
 
* find out some interesting directions.
 
* find out some interesting directions.
 
   (1)A more bayesian prior distribution over neural networks, that we can give constraint on a hidden layer so that the hidden
 
   (1)A more bayesian prior distribution over neural networks, that we can give constraint on a hidden layer so that the hidden
    layers follows a guassian distribution with a more reasonable mean value, which may be a direction of AAAI.  
+
    layers follows a guassian distribution with a more reasonable mean value, which may be a direction of AAAI.  
 
   (2)sequential label learning, which can be a further work with Chaoyuan.
 
   (2)sequential label learning, which can be a further work with Chaoyuan.
   (3)A new topic model, the code is done, still need to speed up.  
+
   (3)Try a new topic model,which is similar to cbow with large window size, the code is done, still need to speed up.
 
   (4)unbalanced autoencoder, haven't considered in details.  
 
   (4)unbalanced autoencoder, haven't considered in details.  
 
* Discuss ideas with Tianyi on RS and finally decide a direction, which is a deep UV structure with content knowledge.
 
* Discuss ideas with Tianyi on RS and finally decide a direction, which is a deep UV structure with content knowledge.

2015年9月1日 (二) 02:46的版本

Work done in this week

  • review papers on bayesian graph, document classification.
  • find out some interesting directions.
 (1)A more bayesian prior distribution over neural networks, that we can give constraint on a hidden layer so that the hidden
    layers follows a guassian distribution with a more reasonable mean value, which may be a direction of AAAI. 
 (2)sequential label learning, which can be a further work with Chaoyuan.
 (3)Try a new topic model,which is similar to cbow with large window size, the code is done, still need to speed up.
 (4)unbalanced autoencoder, haven't considered in details. 
  • Discuss ideas with Tianyi on RS and finally decide a direction, which is a deep UV structure with content knowledge.
  • help Chaoyuan do the baseline reproduction.

Plan to do next week

  • Compare the performance with and without topic distribution constraint. Try adding constraint on different layers.