Patryk Filipiak

Infeasibility driven evolutionary algorithm with ARIMA-based prediction mechanism

This paper proposes an improvement of evolutionary algorithms for dynamic objective functions with a prediction mechanism based on the Autoregressive Integrated Moving Average (ARIMA) model. It extends the Infeasibility Driven Evolutionary Algorithm (IDEA) that maintains a population of feasible and infeasible solutions in order to react on changing objectives faster. Combining IDEA with ARIMA leads to a more efficient evolutionary algorithm that reacts faster to the changing objectives which profits from using information coming from the prediction mechanism and remains one time instant ahead of the original algorithm. Preliminary experiments performed on popular benchmark problems confirm that the IDEA-ARIMA outperforms the original IDEA algorithm in many cases.

Usage patterns of trading rules in stock market trading strategies optimized with evolutionary methods

This paper proposes an approach to analysis of usage patterns of trading rules in stock market trading strategies. Analyzed strategies generate trading decisions based on signals produced by trading rules. Weighted sets of trading rules are used with parameters optimized using evolutionary algorithms. A novel approach to trading rule pattern discovery, inspired by association rule mining methods, is proposed. In the experiments, patterns consisting of up to 5 trading rules were discovered which appear in no less than 50% of trading experts optimized by evolutonary algorithm.

Dynamic Portfolio Optimization in Ultra-High Frequency Environment

This paper concerns the problem of portfolio optimization in the context of ultra-high frequency environment with dynamic and frequent changes in statistics of financial assets. It aims at providing Pareto fronts of optimal portfolios and updating them when estimated return rates or risks of financial assets change. The problem is defined in terms of dynamic optimization and solved online with a proposed evolutionary algorithm. Experiments concern ultra-high frequency time series coming from the London Stock Exchange Rebuilt Order Book database and the FTSE100 index.

Making IDEA-ARIMA Efficient in Dynamic Constrained Optimization Problems

A commonly used approach in Evolutionary Algorithms for Dynamic Constrained Optimization Problems forces re-evaluation of a population of individuals whenever the landscape changes. On the contrary, there are algorithms like IDEA-ARIMA that can effectively anticipate certain types of landscapes rather than react to changes which already happened and thus be one step ahead with the dynamic environment. However, the computational cost of IDEA-ARIMA and its memory consumption are barely acceptable in practical applications. This paper proposes a set of modifications aimed at making this algorithm an efficient and competitive tool by reducing the use of memory and proposing the new anticipation mechanism.

Univariate Marginal Distribution Algorithm with Markov Chain Predictor in Continuous Dynamic Environments

This paper presents an extension of the continuous Univariate Marginal Distribution Algorithm with the prediction mechanism based on a Markov chain model in order to improve the reactivity of the algorithm in continuous dynamic optimization problems. Also a population diversification into exploring, exploiting and anticipating fractions is proposed with the auto-adaptation mechanism for updating dynamically the sizes of these fractions. The proposed approach is tested on the popular benchmark functions with the recurring type of changes.

A predictive evolutionary algorithm for dynamic constrained inverse kinematics problems

This paper presents an evolutionary approach to the Inverse Kinematics problem. The Inverse Kinematics problem concerns finding the placement of a manipulator that satisfies certain conditions. In this paper apart from reaching the target point the manipulator is required to avoid a number of obstacles. The problem which we tackle is dynamic: the obstacles and the target point may be moving which necessitates the continuous update of the solution. The evolutionary algorithm used for this task is a modification of the Infeasibility Driven Evolutionary Algorithm (IDEA) augmented with a prediction mechanism based on the ARIMA model.

Parallel CHC algorithm for solving dynamic traveling salesman problem using many-core GPU

This paper presents a massively parallel evolutionary algorithm with local search mechanism dedicated to dynamic optimization. Its application for solving Dynamic Traveling Salesman Problem (DTSP) is discussed. The algorithm is designed for many-core graphics processors with the Compute Unified Device Architecture (CUDA), which is a parallel computing architecture for nVidia graphics processors. Experiments on a number of benchmark DTSP problems confirmed the efficiency of the algorithm and the parallel computing model designed.

Continuous Population-Based Incremental Learning with Mixture Probability Modeling for Dynamic Optimization Problems

This paper proposes a multimodal extension of PBIL C based on Gaussian mixture models for solving dynamic optimization problems. By tracking multiple optima, the algorithm is able to follow the changes in objective functions more efficiently than in the unimodal case. The approach was validated on a set of synthetic benchmarks including Moving Peaks, dynamization of the Rosenbrock function and compositions of functions from the IEEE CEC’2009 competition. The results obtained in the experiments proved the efficiency of the approach in solving dynamic problems with a number of competing peaks.

Infeasibility Driven Evolutionary Algorithm with Feed-Forward Prediction Strategy for Dynamic Constrained Optimization Problems

This paper proposes a modification of Infeasibility Driven Evolutionary Algorithm that applies the anticipation mechanism following Feed-forward Prediction Strategy. The presented approach allows reacting on environmental changes more rapidly by directing some individuals into the areas of most probable occurrences of future optima. Also a novel population segmentation on exploring, exploiting and anticipating fractions is introduced to assure a better diversification of individuals and thus improve the ability to track moving optima. The experiments performed on the popular benchmarks confirmed the significant improvement in Dynamic Constrained Optimization Problems when using the proposed approach.

Artificial Immune Systems for Data Classification in Planetary Gearboxes Condition Monitoring

In the paper a problem of diagnostic data classification is discussed. The classic condition monitoring approach requires two examples of machines: one in a good and one in a bad condition. From the industrial perspective such a requirement is often very difficult to fulfill, especially in the case of machines with an unique design. To overcome it, we proposed to use the Artificial Immune System (AIS) based approach to classify multidimensional diagnostic data. AIS allows to recognize a change of the machine condition based on a training phase using the dataset related to a good condition. To validate the proposed procedure and assess efficiency of the condition recognition, an extra data set from another machine (of the same type) in a bad condition was used. In the paper several key issues related to the selection of parameters have been discussed.