KR20040089791A - Neural network system of predicting economic indicator - Google Patents

Abstract

PURPOSE: A neural system estimating economical indicators is provided to estimate the economical indicators in real-time by realizing a neural network engine as hardware and learning economical indicator information such as stock information fast through connection to a PC or a mobile communication terminal connected to an Internet server. CONSTITUTION: An interface(11) connects the PC or the mobile communication terminal connected to the Internet with the neural network system. The neural network engine(13) advances the learning by using the information transmitted from the PC or the mobile communication terminal. A memory(15) stores the information transmitted from the PC or the mobile communication terminal and the neural network engine. An information organizer(12) stores the information transmitted from the PC or the mobile communication terminal to the memory and transmits the information stored in the memory to the neural network engine. A register initializer(17) sets an initial setting value for the learning/estimation of the neural network engine. A power manager(16) operates a neural network system by using the power supplied from the PC or the mobile communication terminal. A memory controller(14) controls the memory.

Description

NEURAL NETWORK SYSTEM OF PREDICTING ECONOMIC INDICATOR}

The present invention relates to a neural network system for predicting economic indicators, and more particularly to a neural network system for predicting economic indicators by learning economic indicators continuously provided from a PC or a mobile communication terminal.

The economic index market including stocks is a sample of the capitalist market economy, and the direction of development is determined by domestic and international economic and financial market trends and investment sentiment. Investors trade prices of stocks and futures in accordance with the law of supply and demand in the marketplace, suggesting price directions of various economic indicators. With the advent of the digital media, including the Internet, the stock trading of individual investors using the Internet has recently been activated, and information barriers and investment timing problems existing between institutional investors and individual investors have almost disappeared. As a result, the importance of forecast models related to economic indicators centered on stocks is increasing, and the existing moving average line-based technical analysis methods, which are already common, can no longer provide differentiated information to investors in competitive trading markets. have.

Therefore, the economic index index prediction model has been developed by the neural network engine implemented by software. In this way, the neural network engine implemented by software is mainly used for learning method with good convergence characteristics and easy implementation.

The neural network is a representative model. The neural network is trained by modifying the connection strength of each synapse in the direction of reducing the error signal between the output signal and the target value generated in each neuron of the output layer. The learning is repeated until the output value is similar to the target value, and when the learning is completed, the output corresponding to the input is output as the learning. Therefore, the neural network can get quite accurate and fast output after the learning is completed, but the continuously changing economic index such as stock index, futures index, exchange rate, etc. has the disadvantage of requiring a long learning time for further learning. .

Therefore, the neural network engine implemented by software has a high predictive hit rate, but it is not available during the trading period of the economic index market where the index changes continuously due to the slow learning time of the neural network engine. There is a problem that is used for index prediction after the next day based on the index data of the market on the day.

SUMMARY OF THE INVENTION The present invention has been made to solve the above problems, and implements a neural network engine used in hardware for prediction of continuously changing economic indicators, and a stock price continuously connected to a PC or a mobile communication terminal connected to an Internet server. An object of the present invention is to provide a neural network system that quickly learns economic indicator information such as information and predicts economic indicators in real time.

1 is a schematic block diagram of a neural network system implemented in hardware according to an embodiment of the present invention.

2 is a view showing a state in which the neural network system of FIG. 1 is connected to a computer.

3 is a view showing a state in which the neural network system of Figure 1 connected to a notebook.

4 is a view showing a state in which the neural network system of FIG. 1 is connected.

* Description of the symbols for the main parts of the drawings *

10 neural network system 11 interface unit

12: Information Organizer 13: Neural Network Engine

14: memory control unit 15: memory device

16: power management unit 17: register initializer

18: program port 20: computer

30: notebook 40: PDA

In order to achieve the above object, the present invention provides a neural network system for predicting the economic index, the interface unit for connecting the neural network system with a PC or mobile communication terminal connected to the Internet; A neural network engine for continuously learning by using the information transmitted from the PC or the mobile communication terminal; A memory device for storing information transmitted from the PC or the mobile communication terminal and information transmitted from the neural network engine; An information organizer for storing the information transmitted from the PC or the mobile communication terminal in the memory device and transmitting the information stored in the memory device to the neural network engine; A register initializer for setting initial settings for learning and predicting in the neural network engine; A power management unit for driving the neural network system using the power supplied from the PC or the mobile communication terminal; And a memory controller for controlling the memory device.

In the neural network system, the interface unit may be connected to the PC or the mobile communication terminal using any one of USB, infrared communication, IEEE 1394, RS-232, and parallel cable.

In addition, in the neural network system, the neural network engine learns information transmitted from the PC or mobile communication terminal using neural network algorithm, and the neural network algorithm includes back propagation, hop field, Kohonen, perceptron, multilayer perceptron, and Boltzmann. Machine, self-organizing map or neocognitrone algorithm.

In the neural network system, the information is an economic index.

In addition, the neural network system predicts the economic indicators by any one of seconds, minutes, days, weeks, months, and years.

Further, in the neural network system, the neural network engine processes a technical analysis algorithm, and the technical analysis algorithm is a neural network algorithm, moving average line, MACD, daily chart, stochastic, momentum, bollinger band or eliot wave theory.

Furthermore, the neural network system further includes a program port for connecting the neural network system with an external device and updating a program.

EMBODIMENT OF THE INVENTION Hereinafter, the Example of this invention is described in detail, referring drawings.

1 is a schematic block diagram of a neural network system implemented in hardware according to an embodiment of the present invention.

In FIG. 1, the neural network system 10 includes an interface unit 11, an information organizer 12, a neural network engine 13, a memory control unit 14, a memory device 15, a power management unit 16, and a register initializer. And a program port 18.

The interface unit 11 connects a neural network system 10 with a PC (not shown) or a mobile communication terminal (not shown) connected to the Internet. The interface unit 11 normally uses a PC (not shown) or mobile communication using USB. Connect to a terminal (not shown).

The neural network engine 13 continues to learn using information transmitted from a PC or a mobile communication terminal connected to the Internet.

That is, the neural network engine 13 performs initial learning with initial economic indicators of a predetermined period transmitted from the PC or the mobile communication terminal. For example, share price information (eg, highest price, lowest price, closing price, trading volume, etc.) of a company from January 1999 to the present, provided from an Internet server (not shown) connected using a PC or a mobile communication terminal. Do the initial learning with. Thereafter, the neural network engine 13 can predict the stock price in seconds, minutes, days, weeks, months, or years by learning real-time stock price information provided from the Internet server (not shown).

As such, the neural network engine 13 has a back propagation, hopfield, hopone, kohonen, perceptron, multilayer perceptron, boltzmann machine, self-organizing map. Learning is performed using a neural network algorithm such as Self-Organizing Map or Neocognitron.

In addition, the neural network engine 13 may simultaneously process various technical analysis algorithms such as neural network algorithm, moving average line, MACD, daily chart, stochastic, momentum, Bollinger band, and Elliott wave theory.

The memory device 15 stores information transmitted from a PC or a mobile communication terminal (eg, stock price information) and information transmitted from the neural network engine 13.

The data organizer 12 stores information transmitted from a PC or a mobile communication terminal in the memory device and transmits the information stored in the memory device 15 to the neural network engine 13.

The register initializer 17 sets initial setting values for learning and predicting economic indicators in the neural network engine 13. At this time, the initial set values include stocks, futures, options, exchange rates, and the like. The initial set values are set using the register initializer 17, so that the economic indicators corresponding to the initial set values can be learned and predicted. do.

The memory controller 14 controls that the information transmitted from the data organizer 12 or the neural network engine 13 is stored in the memory device 14.

The power management unit 16 drives the neural network system 10 using the power supplied from the PC or the mobile communication terminal. By using the power supplied from the PC or the mobile communication terminal, a separate power supply is unnecessary. do.

In addition, the program port 18 is connected to the neural network system 10 and an external device (not shown), so that the program provided from the external device is updated in the neural network engine 13.

2 is a diagram illustrating a state in which the neural network system of FIG. 1 is connected to a computer, and FIG. 3 is a diagram illustrating a state in which the neural network system of FIG. 1 is connected to a notebook, and FIG. 4 is a state in which the neural network system of FIG. 1 is connected. Figure showing.

In FIG. 2, the neural network system 10 is connected to a computer 20 connected to an internet server (eg, a securities company server) using a USB, and provides economic indicator information such as stock price information and the like from the computer 20. Will receive.

In FIG. 3, the neural network system 10 is connected to a notebook computer 30 connected to an Internet server (eg, a securities company server) using a USB, and provides economic indicator information such as stock price information from the notebook computer 30. Will receive.

In Fig. 4, the neural network system 10 is connected to a mobile communication terminal (e.g., PDA) 40 connected to an Internet server (e.g., a securities company server) by using USB, and economic indicators such as stock price information and the like. You will be informed.

In this embodiment, the neural network system 10 is connected to the computer 20 or the notebook computer 30 using USB, but is not limited thereto. For example, the infrared network, IEEE1394, RS-232, parallel cable, or the like may be used. It can be connected to the computer 20, notebook 30 or the mobile communication terminal 40 by using.

In addition, although the neural network system 10 in FIG. 2 has the same form as the set-top box, the neural network system 10 may have the same form as the neural network system 10 in FIG. 3 or 4.

As described above, according to the present invention, by implementing the neural network engine used for the prediction of the continuously changing economic index in hardware, the stock price information and the like continuously provided by connecting to the PC or mobile communication terminal connected to the Internet server. You can learn economic indicator information quickly, and can predict economic indicators in real time.

In the above, the present invention has been described and described with respect to certain preferred embodiments. However, the present invention is not limited to the above-described embodiments, and those skilled in the art to which the present invention pertains have various changes without departing from the spirit of the technical idea of the present invention described in the claims below. Could be done.

Claims (9)

As a neural network system for predicting economic indicators, An interface unit connecting the neural network system with a PC or a mobile communication terminal connected to the Internet; A neural network engine for continuously learning by using the information transmitted from the PC or the mobile communication terminal; A memory device for storing information transmitted from the PC or the mobile communication terminal and information transmitted from the neural network engine; An information organizer for storing the information transmitted from the PC or the mobile communication terminal in the memory device and transmitting the information stored in the memory device to the neural network engine; A register initializer for setting initial settings for learning and predicting in the neural network engine; A power management unit for driving the neural network system using the power supplied from the PC or the mobile communication terminal; And Neural network system including a memory control unit for controlling the memory device. The neural network system of claim 1, wherein the interface unit connects to the PC or the mobile communication terminal using any one of USB, infrared communication, IEEE 1394, RS-232, and parallel cable. The neural network system of claim 1, wherein the neural network engine learns information transmitted from the PC or the mobile communication terminal using a neural network algorithm. 4. The neural network system of claim 3, wherein the neural network algorithm is a backpropagation, hopfield, Kohonen, perceptron, multilayer perceptron, Boltzmann machine, self-organizing map or neocognitrone algorithm. The neural network system of claim 1, wherein the information is an economic indicator. The neural network system of claim 1, wherein the neural network system predicts the economic indicator by any one of seconds, minutes, days, weeks, months, and years. The neural network system of claim 1, wherein the neural network engine processes a technical analysis algorithm. 8. The neural network system of claim 7, wherein the technical analysis algorithm is a neural network algorithm, moving average line, MACD, daily chart, stochastic, momentum, Bollinger band or Elliot wave theory. 9. The neural network system according to any one of claims 1 to 8, wherein the neural network system further comprises a program port for connecting to an external device and updating a program.