SC Conference - Activity Details

High Performance Artificial Neural Network: A Computational Finance Application

C. Augusto Casas  (St Thomas Aquinas College)
Posters Session
Tuesday,  05:15PM - 07:00PM
Room Rotunda Lobby
Forecasting stock prices is an essential task for investment managers. Artificial Neural Networks (ANN) have proven successful forecasting stock prices. A complex ANN model needs to be trained with terabytes of data, which requires large amounts of computational time. Investment decisions must be made immediately after receiving new data. A technology capable of training a complex ANN within seconds would be of significant value. The purpose of this research project is the development and testing of a High-Performance Artificial Neural Network (HPANN). The model built consists of a series of 500 artificial neural networks. Each ANN is simultaneously trained in parallel by a computer cluster to forecast the prices of the individual stocks in the S&P500 Index. The networks are trained with 60 years of daily historical financial data. This project is currently under development. The poster will display the ANN, computer cluster architectures and preliminary results.
   IEEE Computer Society  /  ACM     2 0   Y E A R S   -   U N L E A S H I N G   T H E   P O W E R   O F   H P C