“FIN Status Report 2017-4-17”版本间的差异

来自cslt Wiki
跳转至： 导航、 搜索
第4行： 第4行：
 
|-
 
|-
 
| Yang Wang||
 
| Yang Wang||
−
debug dfp and do some experiments.<br>
+
1. Modify the model architecture to share parameters when extracting features.<br>
−
can control the performance on the training dataset by setting the goal,
+
The training process seems to be right.<br>
−
but it performs poorly on the testing dataset
+
  
 
||  
 
||  
−
improve dfp
+
1. Investigate the impact of kinds of regularization.<br>
 +
2. Online learning
 +
 
  
 
|-
 
|-

2017年4月20日 (四) 04:20的版本

People Last Week This Week
Yang Wang

1. Modify the model architecture to share parameters when extracting features.
The training process seems to be right.

1. Investigate the impact of kinds of regularization.
2. Online learning


Yuxin Zhang

Help check code of BullTrader.
Discuss dfp model and run tests.

Help improve the dfp model and run more tests.

Yaodong Wang explore the future contract and organize the data for nips Continue organizing the data
Tongzheng Ren Implement volume, run the test. Prepare the baseline.
取自“http://index.cslt.org/mediawiki/index.php?title=FIN_Status_Report_2017-4-17&oldid=26732”