机器学习算法

基于Tensorflow以及Keras开源机器学习框架,我深入研究机器学习方法例如LSTM以及其他相关方法在时间序列方面的应用。我对他们进行比较,并且提出新的更加高效的预测方法。

股票市场预测

股票市场是一个非线性、动态性极强的复杂系统。通过对某个股票的历史信息的分析,从而预测具有一定可信性的未来股值。我正在努力提高模型的表现,从而达到帮助投资者作出更加可靠的决策。

电池状态评估

为了建立准确的评估电池电量状态以及健康状态的模型,我们使用多维时间序列作为输入建立电池电压、电流、温度同电池电量状态之间的联系。

项目进度简要介绍

在左侧图中,MLA、STOCK和SOC分别指神经网络算法、股票市场预测以及电池电量状态预测。

细节

机器学习算法

80%

“关键词: 时间序列预测,LSTM, RNN"

  • Invent a new structure of multiscale RNN
  • Compare the performance of the new method with LSTM and others
  • Test with different kind of toy data
  • Good performance in the time series forcasting

股票市场预测

75%

“关键词: 时间序列预测模型,趋势预测,情感分析"

  • Develop the stock price prediction model to track the trend of market
  • Consider multivariate related informations
  • Good accuracy

电池电量状态预测

50%

“关键词: 多维变量时间序列预测,电量状态,Simulink"

  • Build the SOC estimation model based on machine learning algorithm
  • Collect battery running simulation data in different circumstance
  • Consider voltage, current, temparature and so on

Xin Chen

Yuxuan Sui

Yadong Zhang

Chenye Zou

Team Leader.

Xin Chen

Associate Professor

Housing Market and Pridictive Analytics.

Yuxuan Sui

Research Student

Time Series Prediction Method, Financial Market Dynamics and Trading Algorithm.

Yadong Zhang

Research Student

Predictive Models of Battery State of Charge and Fast Charging Strategy.

Chenye Zou

Undergraduate Student

10 Jan 2018

Coming Soon...

Coming Soon...

15 Likes

02 Comments

10 Jan 2018

Coming Soon...

Coming Soon...

15 Likes

02 Comments

10 Jan 2018

Coming Soon...

Coming Soon...

15 Likes

02 Comments

10 Jan 2018

Coming Soon...

Coming Soon...

15 Likes

02 Comments