Vladimir Stanovov

Self-Configuring Hybrid Evolutionary Algorithm for Fuzzy Imbalanced Classification with Adaptive Instance Selection

Abstract A novel approach for instance selection in classification problems is presented. This adaptive instance selection is designed to simultaneously decrease the amount of computation resources required and increase the classification quality achieved. The approach generates new training samples during the evolutionary process and changes the training set for the algorithm. The instance selection is guided by means of changing probabilities, so that the algorithm concentrates on problematic examples which are difficult to classify. The hybrid fuzzy classification algorithm with a self-configuration procedure is used as a problem solver. The classification quality is tested upon 9 problem data sets from the KEEL repository. A special balancing strategy is used in the instance selection approach to improve the classification quality on imbalanced datasets. The results prove the usefulness of the proposed approach as compared with other classification methods.

Hybrid self-configuring evolutionary algorithm for automated design of fuzzy logic rule base

In this paper a method for fuzzy logic systems design, which implements the latest developments in this field, is presented. The main evolutionary algorithm uses the Pittsburgtype approach, and the Michigan-type one is used as a mutation operator. A self-configuring technique is used to adjust the algorithm parameters based on their success rates. The novelty here is the algorithms ability to adjust the probability using either the genetic or heuristic method for the incorporation of a new rule in the rule base. Previously, this was done voluntarily. It is demonstrated that this new algorithms flexibility does not decrease its performance although it makes it fully automated.

Streaming pulse data to the cloud with bluetooth LE or NODEMCU ESP8266

The paper describes the development of the three prototypes which enable monitoring of heart pulse sensor data on the cloud. The prototypes are based on a) Bluetooth module, b) Low Energy Bluetooth module and c) ESP8266 Wi-Fi module. The client side implementation is developed by the application of JavaScript. Node.js was used on the server side with application of node-serialport library with Bluetooth modules. In the case where ESP8266 Wi-Fi module was used, the data was transmitted directly to the cloud. The three different prototypes were compared according to the power consumption and complexity of design. The code example is provided illustrating the approach to develop cloud based graphing interface for biomedical mobile monitoring.

Prototype of group heart rate monitoring with NODEMCU ESP8266

Paper describes the development of prototype that enables monitoring of heart rate and inter beat interval for several subjects. The prototype was realized using ESP8266 hardware modules, WebSocket library, nodejs and JavaScript. System architecture is described where nodejs server acts as the signal processing and GUI code provider for clients. Signal processing algorithm was implemented in JavaScript. Application GUI is presented which can be used on mobile devices. Several important parts of the code are described which illustrate the communication between ESP8266 modules, server and clients. Developed prototype shows one of the possible realizations of group monitoring of biomedical data.

Instance Selection Approach for Self-Configuring Hybrid Fuzzy Evolutionary Algorithm for Imbalanced Datasets

We propose an instance selection technique with subsample balancing for an evolutionary classification algorithm. The technique creates subsamples of the training sample in a way to guide the learning process towards problematic areas of the search space. For unbalanced datasets, the number of instances of different classes is artificially balanced to get better classification results. We apply this technique to a self-configured hybrid evolutionary fuzzy classification algorithm. We performed tests on 4 datasets to evaluate the accuracy as well as other classification quality measures for different parameters of the active instance selection procedure. The results shown by our algorithm are comparable or even better than other algorithms on the same classification problems.

Fuzzy rule bases automated design with self-configuring evolutionary algorithm

Self-configuring evolutionary algorithm of fuzzy rule bases automated deign for solving classification problems, which combines Pittsburgh and Michigan approaches, is introduced. The evolutionary algorithm is based on the Pittsburgh approach where every individual is a rule base and the Michigan approach is used as a mutation operator. A self-configuration method is used to adjust probabilities of the usage of selection, mutation and Michigan part operators. Testing the algorithm on a number of real-world problems demonstrates its efficiency comparing to several other commonly used approaches.

Soft Island Model for Population-Based Optimization Algorithms.

Population-based optimization algorithms adopt a regular network as topologies with one set of potential solutions, which may encounter the problem of premature convergence. In order to improve the performance of optimization techniques, this paper proposes a soft island model topology. The initial population is virtually separated into several subpopulations, and the connection between individuals from subpopulations is probabilistic. The workability of the proposed model was demonstrated through its implementation to the Particle Swarm Optimization and Differential Evolution algorithms and their modifications. Experiments were conducted on benchmark functions taken from the CEC’2017 competition. The best parameters for the new topology adaptation mechanism were found. Results verify the effectiveness of the population-based algorithms with the proposed model when compared with the same algorithms without the model. It was established that by applying this topology adaptation mechanism, the population-based algorithms are able to balance their exploitation and exploration abilities during the search process.

Co-operation of Biology Related Algorithms for Solving Opinion Mining Problems by Using Different Term Weighting Schemes

Automatically generated data mining tools namely artificial neural networks, support vector machines and fuzzy rule-based classifiers, using different term weighting schemes as data pre-processing techniques for opinion mining problems are presented. Developed collective nature-inspired self-tuning meta-heuristic for solving unconstrained and constrained real- and binary-parameter optimization problems called Co-Operation of Biology Related Algorithms was used for classifiers design. Three Opinion Mining problems from DEFT’07 competition were solved by proposed classifiers. Obtained results were compared between themselves and with results obtained by methods which were proposed by other researchers. As the result, workability and usefulness of designed classifiers were established and best data processing approach for them was found.

Cooperation of Bio-inspired and Evolutionary Algorithms for Neural Network Design

A meta-heuristic called Co-Operation of Biology-Related Algorithms (COBRA) with a fuzzy controller, as well as a new algorithm based on the cooperation of Differential Evolution and Particle Swarm Optimization (DE+PSO) and developed for solving real-valued optimization problems, were applied to the design of artificial neural networks. The usefulness and workability of both meta-heuristic approaches were demonstrated on various benchmarks. The neural network’s weight coefficients represented as a string of real-valued variables are adjusted with the fuzzy controlled COBRA or with DE+PSO. Two classification problems (image and speech recognition problems) were solved with these approaches. Experiments showed that both cooperative optimization techniques demonstrate high performance and reliability in spite of the complexity of the solved optimization problems. The workability and usefulness of the proposed meta-heuristic optimization algorithms are confirmed.

Fuzzy Logic Controller Design for Tuning the Cooperation of Biology-Inspired Algorithms

Previously, a meta-heuristic approach called Co-Operation of Biology Related Algorithms or COBRA for solving real-parameter optimization problems was introduced and described. COBRA’s basic idea consists in a cooperative work of well-known bio-inspired algorithms, which were chosen due to the similarity of their schemes. COBRA’s performance was evaluated on a set of test functions and its workability was demonstrated. Thus it was established that the idea of the algorithms’ cooperative work is useful. However, it is unclear which bionic algorithms should be included in this cooperation and how many of them. Therefore, the aim of this study was to design a fuzzy logic controller for determining which bio-inspired algorithms should be included in the co-operative work for solving optimization problems using the COBRA approach. The population sizes of the bio-inspired component-algorithms were automatically changed by the obtained controller. The experimental results obtained by the two types of fuzzy-controlled COBRA are presented and their usefulness is demonstrated.