Generating Trading Rules for Stock Markets Using Robust Genetic Network Programming and Portfolio Beta

In this paper, Robust Genetic Network Programming (R-GNP) for generating trading rules for stocks is described. R-GNP is a new evolutionary algorithm, where solutions are represented using graph structures. It has been clarified that R-GNP works well especially in dynamic environments. In the proposed hybrid model, R-GNP is applied to generating stock trading rules with variance of fitness values. The unique point is that the generalization ability of R-GNP is improved by using the robust fitness function, which consists of the fitness functions with the original data and a good number of correlated data. Generally speaking, the hybrid intelligent system consists of three steps: priority selection by the portfolio β, optimization by the Genetic Relation Algorithm (GRA), and stock trading by R-GNP. In the simulations, the trading model is trained using the stock prices of 10 brands on the Tokyo Stock Exchange, and then the generalization ability is tested. From the simulation results, it is clarified that the trading rules created by the proposed R-GNP model obtain much higher profits than the traditional methods even in the world-wide financial crisis of 2007. Hence, its effectiveness has been confirmed.

A Class Association Rule Based Classifier Using Probability Density Functions for Intrusion Detection Systems

As the number of computer systems connected to the Internet is increasing exponentially, the computer security has become a crucial problem, and many techniques for Intrusion detection have been proposed to detect network attacks efficiently. On the other hand, data mining algorithms based on Genetic Network Programming (GNP) have been proposed and applied to Intrusion detection recently. GNP is a graph-based evolutionary algorithm and can extract many important class association rules by making use of the distinguished representation ability of the graph structure. In this paper, probabilistic classification algorithms based on multi-dimensional probability distribution are proposed and combined with conventional class association rule mining of GNP, and applied to network intrusion detection for the performance evaluation. The proposed classification algorithms are based on 1) one-dimensional probability density functions and 2) a two-dimensional joint probability density function. These functions represent the distribution of normal and intrusion accesses and efficiently classify a new access data into normal, known intrusion or even unknown intrusion. The simulations using KDD99Cup database from MIT Lincoln Laboratory show some advantages of the proposed algorithms over the conventional mean and standard deviation-based method.

Immunity-based optimal estimation approach for a new real time group elevator dynamic control application for energy and time saving

Nowadays, the increasing use of group elevator control systems owing to increasing building heights makes the development of high-performance algorithms necessary in terms of time and energy saving. Although there are many studies in the literature about this topic, they are still not effective enough because they are not able to evaluate all features of system. In this paper, a new approach of immune system-based optimal estimate is studied for dynamic control of group elevator systems. The method is mainly based on estimation of optimal way by optimizing all calls with genetic, immune system and DNA computing algorithms, and it is evaluated with a fuzzy system. The system has a dynamic feature in terms of the situation of calls and the option of the most appropriate algorithm, and it also adaptively works in terms of parameters such as the number of floors and cabins. This new approach which provides both time and energy saving was carried out in real time. The experimental results comparatively demonstrate the effects of method. With dynamic and adaptive control approach in this study carried out, a significant progress on group elevator control systems has been achieved in terms of time and energy efficiency according to traditional methods.

An Explicit Memory Scheme of Genetic Network Programming

Many classical methods such as Genetic Algorithm (GA), Genetic Programming (GP), Evolutionary Strategies (ES), etc. have made significant contribution to the study of evolutionary computation. And recently, a new approach named Genetic Network Programming (GNP) has been proposed especially for solving complex problems in dynamic environments. It is based on the algorithms of classical evolutionary computation techniques and uses data structures of directed graphs which are the unique feature of GNP. Focusing on GNP’s distinguished expression ability of the graph structure, this paper proposes an enhanced architecture for standard GNP in order to improve the performance of GNP by adopting an explicit memory scheme which records and utilizes the exploited information flexibly and extensively during the evolution process of GNP. In the enhanced architecture, the important gene information of the elite individuals is extracted and accumulated in the memory during evolution. Among the accumulated information, some of them are selected and used to guide the agents. In this paper, the proposed architecture is applied to the tileworld which is an excellent benchmark for evaluating the architecture demonstrating its superiority.

Enhancing Bidding Strategy Using Genetic Network Programming in Agent-Based Multiple Round English Auction

The agent-based auction mechanism widely used in web sites and originally designed for trading goods for customers might not be the most efficient one in the future, while there is a demand of automated auction agents, which are adaptable to the dynamic auction environments. To this end, this paper discusses how to apply Genetic Network Programming (GNP) to automated auction agents in order to make a bid efficiently and effectively at each time step according to the auction environments, and Multiple Round English Auction (MREA) mechanism studied in this paper is based on multi-agent systems, which aims to help the buyer to procure profitable deals as much as possible. GNPbased agent is compared with other agents using conventional strategies in MREA. It has been found from the simulations that the proposed method could help agents to evolve their strategies generation by generation, which shows that GNP has a good performance of helping the agent to find the suitable strategy under various situations and outperform than other strategies.

Integrated Rule Mining Based on Fuzzy GNP and Probabilistic Classification for Intrusion Detection

With the increasing popularity of the Internet, network security has become a serious problem recently. How to detect intrusions effectively becomes an important component in network security. Therefore, a variety of algorithms have been devoted to this challenge. Genetic network programming is a newly developed evolutionary algorithm with directed graph gene structures, and it has been applied to data mining for intrusion detection systems providing good performances in intrusion detection. In this paper, an integrated rule mining algorithm based on fuzzy GNP and probabilistic classification is proposed. The integrated rule mining uses fuzzy class association rule mining algorithm to extract rules with different classes. Actually, it can deal with both discrete and continuous attributes in network connection data. Then, the classification is done probabilistically using different class rules. The integrated method showed excellent results by simulation experiments.

No Time Limit and Time Limit Model of Multiple Round Dutch Auction Based on Genetic Network Programming

Nowadays, Dutch auction is used widely at online auction sites. To make online Dutch auction more efficient and more intelligent, it is useful to develop an agent using evolutionary computation which will be adaptive to different auction environments. In this paper, a Genetic Network Programming (GNP) based strategy for auction agents has been proposed to do auctions in multiple round Dutch Auction environments under two types of auction models, no time limit model and time limit model. GNP is a graph-based evolutionary method extended from Genetic Algorithms (GA) and Genetic Programming (GP), which can create optimal solutions by evolution. Although the application of GNP to English auction has been done already, here, a new GNP structure is used for Dutch auction. The simulation results show that the GNP based strategy can also make the agents work well in Dutch auction and the advanced GNP structure makes the agents perform better than that in English auction.

A Novel Taxi Dispatch System Integrating a Multi-Customer Strategy and Genetic Network Programming

Taxi service usually benefits people by providing comfortable and flexible ride experiences. However, an inherent problem, the insufficient number of taxis at traffic peak, has baffled taxi service ever since it existed. This paper hereby proposes a Multi-Customer Taxi Dispatch System (MCTDS), where taxis are granted a right to take customers with different Origin-Destination (OD) pairs simultaneously, to shorten the total waiting time and traveling time. In addition, to mitigate the damage of detours, MCTDS is built based on Genetic Network Programming, a graph-based evolutionary algorithm that has shown excellent performances previously in some complicated applications. We also modify the structure of GNP to achieve an improvement in performance. In the simulation part, we demonstrate that MCTDS outperforms the conventional GNP and some heuristic taxi dispatch approaches.

A Double-Deck Elevator Systems Controller with Idle Cage Assignment Algorithm Using Genetic Network Programming

So far, many studies on Double-Deck Elevator Systems (DDES) have been done for exploring more efficient algorithms to improve the system transportation capacity, especially in a heavy traffic mode. The main idea of these algorithms is to decrease the number of stops during a round trip by grouping the passengers with the same destination as much as possible. Unlike what occurs in this mode, where all cages almost always keep moving, there is the case, where some cages become idle in a light traffic mode. Therefore, how to dispatch these idle cages, which is seldom considered in the heavy traffic mode, becomes important when developing the controller of DDES. In this paper, we propose a DDES controller with idle cage assignment algorithm embedded using Genetic Network Programming (GNP) for a light traffic mode, which is based on a timer and event-driven hybrid model. To verify the efficiency and effectiveness of the proposed method, some experiments have been done under a special down-peak pattern where passengers emerge especially at the 7th floor. Simulation results show that the proposed method improves the performance compared with the case when the cage assignment algorithm is not employed and works better than six other heuristic methods in a light traffic mode.

Evolving Asset Portfolios by Genetic Relation Algorithm

Global financial development have opened innumerable risks and opportunities for investments. A global view of the portfolio allocation through diversification brings advantages for the risk allocation in investments. In this paper, an asset allocation framework under the return, risk and liquidity considerations is proposed for short term investment using Genetic Relation Algorithm. Simulations using the stocks, bonds and currencies from relevant financial markets in USA, Europe and Asia show that the proposed framework is effective and robust. The efficacy of the proposed method is compared against the relevant constructs in finance and computational fields.

Genetic Network Programming with Estimation of Distribution Algorithms for Class Association Rule Mining in Traffic Prediction

Genetic Network Programming (GNP) is one of the evolutionary optimization algorithms, which uses directed-graph structures to represent its solutions. It has been clarified that GNP works well to find class association rules in traffic prediction systems. In this paper, a novel evolutionary paradigm named GNP with Estimation of Distribution Algorithms (GNP-EDAs) is proposed and used to find important class association rules in traffic prediction systems. In GNP-EDAs, a probabilistic model replaces crossover and mutation to enhance the evolution. The new population of individuals is produced from the probabilistic distribution estimated from the selected elite individuals of the previous generation. The probabilistic information on the connections and transitions of GNP-EDAs is extracted from its population to construct the probabilistic model. In this paper, two methods are described to build the probabilistic model for producing the offspring. In addition, a classification mechanism is introduced to estimate the traffic prediction based on the extracted class association rules. We compared GNPEDAs with the conventional GNP and the simulation results showed that GNP-EDAs can extract the class association rules more effectively, when the number of the candidate class association rules increase. And the classification accuracy of the proposed method shows good results in traffic prediction systems.

Traffic Flow Prediction with Genetic Network Programming (GNP)

In this paper, a method for traffic flow prediction has been proposed to obtain prediction rules from the past traffic data using Genetic Network Programming (GNP). GNP is an evolutionary approach which can evolve itself and find the optimal solutions. It has been clarified that GNP works well especially in dynamic environments since GNP is consisted of directed graph structures, creates quite compact programs and has an implicit memory function. In this paper, GNP is applied to create a traffic flow prediction model. And we proposed the spatial adjacency model for the prediction and two kinds of models forN-step prediction. Additionally, the adaptive penalty functions are adopted for the fitness function in order to alleviate the infeasible solutions containing loops in the training process. Furthermore, the sharing function is also used to avoid the premature convergence.

Combination of Two Evolutionary Methods for Mining Association Rules in Large and Dense Databases

Among several methods of extracting association rules that have been reported, a new evolutionary method named Genetic Network Programming (GNP) has also shown its effectiveness for small databases in the sense that they have a relatively small number of attributes. However, this conventional GNP method is not be able to deal with large databases with a huge number of attributes, because its search space becomes very large, causing bad performance at running time. The aim of this paper is to propose a new method to extract association rules from large and dense databases with a huge amount of attributes through the combination of conventional GNP based mining method and a specially designed genetic algorithm (GA). Each of these evolutionary methods works in its own processing level and they are highly synchronized to act as one system.

Our strategy consists in the division of a large and dense database into many small databases. These small databases are considered as individuals and form a population. Then the conventional GNP based mining method is applied to extract association rules for each of these individuals. Finally, the population is evolved through several generations using GA with special genetic operators considering the acquired information. Two complementary processing levels are defined: Global Level and Local Level, each with its own independent tasks and processes. In the Global Level mainly GA process is carried out, whereas in the Local Level, conventional GNP based mining method is carried out in parallel and they generate their own local pools of association rules. Several special genetic operations for GA in the Global Level are proposed and the performance of each of them and their combination is shown and compared.

In our simulations, the conventional GNP based mining method and our proposed method are compared using a real world large and dense database with a huge amount of attributes. The results show that extending the conventional GNP based mining method using GA allows to extract association rules from large and dense databases directly and more efficiently than the conventional GNP method.

Genetic Network Programming with Rule Accumulation and its Application to Tile-World Problem

Genetic Network Programming (GNP) is an evolutionary algorithm derived from GA and GP. Directed graph structure, reusability of nodes, and implicit memory function enable GNP to deal with complex problems in dynamic environments efficiently and effectively, as many paper demonstrated. This paper proposed a new method to optimize GNP by extracting and using rules. The basic idea of GNP with Rule Accumulation (GNP with RA) is to extract rules with higher fitness values from the elite individuals and store them in the pool every generation. A rule is defined as a sequence of successive judgment results and a processing node, which represent the good experiences of the past behavior. As a result, the rule pool serves as an experience set of GNP obtained in the evolutionary process. By extracting the rules during the evolutionary period and then matching them with the situations of the environment, we could, for example, guide agents' behavior properly and get better performance of the agents. In this paper, we apply GNP with RA to the problem of determining agents' behaviors in the Tile-world environment in order to evaluate its effectiveness. The simulation results demonstrate that GNP with RA could have better performances than the conventional GNP both in the average fitness value and its stability.

A Study of Double-Deck Elevator Systems Using Genetic Network Programming with Reinforcement Learning

In order to increase the transportation capability of elevator group systems in high-rise buildings without adding elevator installation space, double-deck elevator systems (DDES) is developed as one of the next generation elevator group control systems. Artificial intelligence (AI) technologies have been employed to find some efficient solutions in the elevator group control systems during the late 20th century. Genetic Network Programming (GNP), a new evolutionary computation method, has been employed as the elevator group control system controller in some studies of recent years. Moreover, reinforcement learning (RL) has been also found to be useful for more improvements of elevator group control performances when it is combined with GNP. In this paper, we proposed a new approach of DDES using GNP with RL, and did some experiments on a simulated elevator group control system of a typical office building to evaluate its applicability and efficiency. Simulation results show that the DDES using GNP with RL performs better than the one without RL in regular and down-peak time, while both of them outperforms a conventional approach and a heuristic approach in all three traffic patterns.