Turker Ince

Evolutionary artificial neural networks by multi-dimensional particle swarm optimization

In this paper, we propose a novel technique for the automatic design of Artificial Neural Networks (ANNs) by evolving to the optimal network configuration(s) within an architecture space. It is entirely based on a multi-dimensional Particle Swarm Optimization (MD PSO) technique, which re-forms the native structure of swarm particles in such a way that they can make inter-dimensional passes with a dedicated dimensional PSO process. Therefore, in a multidimensional search space where the optimum dimension is unknown, swarm particles can seek both positional and dimensional optima. This eventually removes the necessity of setting a fixed dimension a priori, which is a common drawback for the family of swarm optimizers. With the proper encoding of the network configurations and parameters into particles, MD PSO can then seek the positional optimum in the error space and the dimensional optimum in the architecture space. The optimum dimension converged at the end of a MD PSO process corresponds to a unique ANN configuration where the network parameters (connections, weights and biases) can then be resolved from the positional optimum reached on that dimension. In addition to this, the proposed technique generates a ranked list of network configurations, from the best to the worst. This is indeed a crucial piece of information, indicating what potential configurations can be alternatives to the best one, and which configurations should not be used at all for a particular problem. In this study, the architecture space is defined over feed-forward, fully-connected ANNs so as to use the conventional techniques such as back-propagation and some other evolutionary methods in this field. The proposed technique is applied over the most challenging synthetic problems to test its optimality on evolving networks and over the benchmark problems to test its generalization capability as well as to make comparative evaluations with the several competing techniques. The experimental results show that the MD PSO evolves to optimum or near-optimum networks in general and has a superior generalization capability. Furthermore, the MD PSO naturally favors a low-dimension solution when it exhibits a competitive performance with a high dimension counterpart and such a native tendency eventually yields the evolution process to the compact network configurations in the architecture space rather than the complex ones, as long as the optimality prevails.

A Generic and Robust System for Automated Patient-Specific Classification of ECG Signals

This paper presents a generic and patient-specific classification system designed for robust and accurate detection of ECG heartbeat patterns. The proposed feature extraction process utilizes morphological wavelet transform features, which are projected onto a lower dimensional feature space using principal component analysis, and temporal features from the ECG data. For the pattern recognition unit, feedforward and fully connected artificial neural networks, which are optimally designed for each patient by the proposed multidimensional particle swarm optimization technique, are employed. By using relatively small common and patient-specific training data, the proposed classification system can adapt to significant interpatient variations in ECG patterns by training the optimal network structure, and thus, achieves higher accuracy over larger datasets. The classification experiments over a benchmark database demonstrate that the proposed system achieves such average accuracies and sensitivities better than most of the current state-of-the-art algorithms for detection of ventricular ectopic beats (VEBs) and supra-VEBs (SVEBs). Over the entire database, the average accuracy-sensitivity performances of the proposed system for VEB and SVEB detections are 98.3%-84.6% and 97.4%-63.5%, respectively. Finally, due to its parameter-invariant nature, the proposed system is highly generic, and thus, applicable to any ECG dataset.

Atmospheric disturbances that generate intermittent turbulence in nocturnal boundary layers

Using the unprecedented observational facilities deployed duringthe 1999 Cooperative Atmosphere-Surface Exchange Study (CASES-99),we found three distinct turbulent events on the night of 18October 1999. These events resulted from a density current,solitary wave, and internal gravity wave, respectively. Our studyfocuses on the turbulence intermittency generated by the solitarywave and internal gravity wave, and intermittent turbulenceepisodes associated with pressure change and wind direction shiftsadjacent to the ground. Both the solitary and internal gravitywaves propagated horizontally and downward. During the passage ofboth the solitary and internal gravity waves, local thermal andshear instabilities were generated as cold air was pushed abovewarm air and wind gusts reached to the ground. These thermal andshear instabilities triggered turbulent mixing events. Inaddition, strong vertical acceleration associated with thesolitary wave led to large non-hydrostatic pressure perturbationsthat were positively correlated with temperature. The directionaldifference between the propagation of the internal gravity waveand the ambient flow led to lateral rolls. These episodic studiesdemonstrate that non-local disturbances are responsible for localthermal and shear instabilities, leading to intermittentturbulence in nocturnal boundary layers. The origin of thesenon-local disturbances needs to be understood to improve mesoscalenumerical model performance.

Intermittent Turbulence Associated with a Density Current Passage in the Stable Boundary Layer

Using the unprecedented observational capabilities deployed duringthe Cooperative Atmosphere-Surface Exchange Study-99 (CASES-99),we found three distinct turbulence events on the night of 18October 1999, each of which was associated with differentphenomena: a density current, solitary waves, and downwardpropagating waves from a low-level jet. In this study, we focus onthe first event, the density current and its associatedintermittent turbulence. As the cold density current propagatedthrough the CASES-99 site, eddy motions in the upper part of thedensity current led to periodic overturning of the stratifiedflow, local thermal instability and a downward diffusion ofturbulent mixing. Propagation of the density current induced asecondary circulation. The descending motion following the head ofthe density current resulted in strong stratification, a sharpreduction in the turbulence, and a sudden increase in the windspeed. As the wind surge propagated toward the surface, shearinstability generated upward diffusion of turbulent mixing. Wedemonstrate in detail that the height and sequence of the localthermal and shear instabilities associated with the dynamics ofthe density current are responsible for the apparent intermittentturbulence.

Fractional Particle Swarm Optimization in Multidimensional Search Space

In this paper, we propose two novel techniques, which successfully address several major problems in the field of particle swarm optimization (PSO) and promise a significant breakthrough over complex multimodal optimization problems at high dimensions. The first one, which is the so-called multidimensional (MD) PSO, re-forms the native structure of swarm particles in such a way that they can make interdimensional passes with a dedicated dimensional PSO process. Therefore, in an MD search space, where the optimum dimension is unknown, swarm particles can seek both positional and dimensional optima. This eventually removes the necessity of setting a fixed dimension a priori, which is a common drawback for the family of swarm optimizers. Nevertheless, MD PSO is still susceptible to premature convergences due to lack of divergence. Among many PSO variants in the literature, none yields a robust solution, particularly over multimodal complex problems at high dimensions. To address this problem, we propose the fractional global best formation (FGBF) technique, which basically collects all the best dimensional components and fractionally creates an artificial global best (aGB) particle that has the potential to be a better ¿guide¿ than the PSOs native gbest particle. This way, the potential diversity that is present among the dimensions of swarm particles can be efficiently used within the aGB particle. We investigated both individual and mutual applications of the proposed techniques over the following two well-known domains: 1) nonlinear function minimization and 2) data clustering. An extensive set of experiments shows that in both application domains, MD PSO with FGBF exhibits an impressive speed gain and converges to the global optima at the true dimension regardless of the search space dimension, swarm size, and the complexity of the problem.

Real-Time Patient-Specific ECG Classification by 1-D Convolutional Neural Networks

Goal: This paper presents a fast and accurate patient-specific electrocardiogram (ECG) classification and monitoring system. Methods: An adaptive implementation of 1-D convolutional neural networks (CNNs) is inherently used to fuse the two major blocks of the ECG classification into a single learning body: feature extraction and classification. Therefore, for each patient, an individual and simple CNN will be trained by using relatively small common and patient-specific training data, and thus, such patient-specific feature extraction ability can further improve the classification performance. Since this also negates the necessity to extract hand-crafted manual features, once a dedicated CNN is trained for a particular patient, it can solely be used to classify possibly long ECG data stream in a fast and accurate manner or alternatively, such a solution can conveniently be used for real-time ECG monitoring and early alert system on a light-weight wearable device. Results: The results over the MIT-BIH arrhythmia benchmark database demonstrate that the proposed solution achieves a superior classification performance than most of the state-of-the-art methods for the detection of ventricular ectopic beats and supraventricular ectopic beats. Conclusion: Besides the speed and computational efficiency achieved, once a dedicated CNN is trained for an individual patient, it can solely be used to classify his/her long ECG records such as Holter registers in a fast and accurate manner. Significance: Due to its simple and parameter invariant nature, the proposed system is highly generic, and, thus, applicable to any ECG dataset.

Real-Time Motor Fault Detection by 1-D Convolutional Neural Networks

Early detection of the motor faults is essential and artificial neural networks are widely used for this purpose. The typical systems usually encapsulate two distinct blocks: feature extraction and classification. Such fixed and hand-crafted features may be a suboptimal choice and require a significant computational cost that will prevent their usage for real-time applications. In this paper, we propose a fast and accurate motor condition monitoring and early fault-detection system using 1-D convolutional neural networks that has an inherent adaptive design to fuse the feature extraction and classification phases of the motor fault detection into a single learning body. The proposed approach is directly applicable to the raw data (signal), and, thus, eliminates the need for a separate feature extraction algorithm resulting in more efficient systems in terms of both speed and hardware. Experimental results obtained using real motor data demonstrate the effectiveness of the proposed method for real-time motor condition monitoring.

Multidimensional Particle Swarm Optimization for Machine Learning and Pattern Recognition

For many engineering problems we require optimization processes with dynamic adaptation as we aim to establish the dimension of the search space where the optimum solution resides and develop robust techniques to avoid the local optima usually associated with multimodal problems. This book explores multidimensional particle swarm optimization, a technique developed by the authors that addresses these requirements in a well-defined algorithmic approach. After an introduction to the key optimization techniques, the authors introduce their unified framework and demonstrate its advantages in challenging application domains, focusing on the state of the art of multidimensional extensions such as global convergence in particle swarm optimization, dynamic data clustering, evolutionary neural networks, biomedical applications and personalized ECG classification, content-based image classification and retrieval, and evolutionary feature synthesis. The content is characterized by strong practical considerations, and the book is supported with fully documented source code for all applications presented, as well as many sample datasets. The book will be of benefit to researchers and practitioners working in the areas of machine intelligence, signal processing, pattern recognition, and data mining, or using principles from these areas in their application domains. It may also be used as a reference text for graduate courses on swarm optimization, data clustering and classification, content-based multimedia search, and biomedical signal processing applications.

Early forest fire detection using radio-acoustic sounding system.

Automated early fire detection systems have recently received a significant amount of attention due to their importance in protecting the global environment. Some emergent technologies such as ground-based, satellite-based remote sensing and distributed sensor networks systems have been used to detect forest fires in the early stages. In this study, a radio-acoustic sounding system with fine space and time resolution capabilities for continuous monitoring and early detection of forest fires is proposed. Simulations show that remote thermal mapping of a particular forest region by the proposed system could be a potential solution to the problem of early detection of forest fires.

Automated patient-specific classification of premature ventricular contractions

In this paper, we present an automated patient-specific electrocardiogram (ECG) beat classifier designed for accurate detection of premature ventricular contractions (PVCs). In the proposed feature extraction scheme, the principal component analysis (PCA) is applied to the dyadic wavelet transform (DWT) of the ECG signal to extract morphological ECG features, which are then combined with the temporal features to form a resultant efficient feature vector. For the classification scheme, we selected the feed-forward artificial neural networks (ANNs) optimally designed by the multi-dimensional particle swarm optimization (MD-PSO) technique, which evolves the structure and weights of the network specifically for each patient. Training data for the ANN classifier include both global (total of 150 representative beats randomly sampled from each class in selected training files) and local (the first 5 min of a patients ECG recording) training patterns. Simulation results using 40 files in the MIT/BIH arrhythmia database achieved high average accuracy of 97% for differentiating normal, PVC, and other beats.