Fikret S. Gürgen

Collection and Analysis of a Parkinson Speech Dataset With Multiple Types of Sound Recordings

There has been an increased interest in speech pattern analysis applications of Parkinsonism for building predictive telediagnosis and telemonitoring models. For this purpose, we have collected a wide variety of voice samples, including sustained vowels, words, and sentences compiled from a set of speaking exercises for people with Parkinsons disease. There are two main issues in learning from such a dataset that consists of multiple speech recordings per subject: 1) How predictive these various types, e.g., sustained vowels versus words, of voice samples are in Parkinsons disease (PD) diagnosis? 2) How well the central tendency and dispersion metrics serve as representatives of all sample recordings of a subject? In this paper, investigating our Parkinson dataset using well-known machine learning tools, as reported in the literature, sustained vowels are found to carry more PD-discriminative information. We have also found that rather than using each voice recording of each subject as an independent data sample, representing the samples of a subject with central tendency and dispersion metrics improves generalization of the predictive model.

Computer and Information Sciences - ISCIS 2005

Invited Speakers.- Keeping Viruses Under Control.- Online Auctions: Notes on Theory, Practice, and the Role of Agents.- Computer Networks.- A Unified Approach to Survivability of Connection-Oriented Networks.- SCTP Based Framework for Mobile Web Agent.- An Agent-Based Scheme for Efficient Multicast Application in Mobile Networks.- An Enhanced One Way Function Tree Rekey Protocol Based on Chinese Remainder Theorem.- Admission Control for Multicast Routing with Quality of Service in Ad Hoc Networks.- An Efficient On-line Job Admission Control Scheme to Guarantee Deadlines for QoS-Demanding Applications.- A Methodology of Resilient MPLS/VPN Path Management Under Multiple Link Failures.- Sensor and Satellite Networks.- Comparison of Hyper-DAG Based Task Mapping and Scheduling Heuristics for Wireless Sensor Networks.- A Markov-Based Model to Analyze the Temporal Evolution and Lifetime of a Sensor Network.- Power-Efficient Seamless Publishing and Subscribing in Wireless Sensor Networks.- Group-Oriented Channel Protection for Mobile Devices in Digital Multimedia Broadcasting.- IP Traffic Load Distribution in NGEO Broadband Satellite Networks - (Invited Paper).- Cross-Layer Management of Radio Resources in an Interactive DVB-RCS-Based Satellite Network-(Invited Paper).- Aggressive Back off Strategy in Congestion Management Algorithm for DBS-RCS - (Invited Paper).- TCP-Peach++: Enhancement of TCP-Peach+ for Satellite IP Networks with Asymmetrical Bandwidth and Persistent Fades-(Invited Paper).- Security and Cryptography.- Automatic Translation of Serial to Distributed Code Using CORBA Event Channels.- Fault Tolerant and Robust Mutual Exclusion Protocol for Synchronous Distributed Systems.- Exact Best-Case End-to-End Response Time Analysis for Hard Real-Time Distributed Systems.- A Formal Policy Specification Language for an 802.11 WLAN with Enhanced Security Network.- A Generic Policy-Conflict Handling Model.- A Truly Random Number Generator Based on a Continuous-Time Chaotic Oscillator for Applications in Cryptography.- A New Cryptanalytic Time-Memory Trade-Off for Stream Ciphers.- SVM Approach with a Genetic Algorithm for Network Intrusion Detection.- Performance Evaluation.- Modeling Access Control Lists with Discrete-Time Quasi Birth-Death Processes.- Stochastic Bounds on Partial Ordering: Application to Memory Overflows Due to Bursty Arrivals.- QoS Evaluation Method in Multimedia Applications Using a Fuzzy Genetic Rule-Based System.- Impact of Setup Message Processing and Optical Switch Configuration Times on the Performance of IP over Optical Burst Switching Networks.- Characterizing Gnutella Network Properties for Peer-to-Peer Network Simulation.- Computing Communities in Large Networks Using Random Walks.- Fame as an Effect of the Memory Size.- Keeping Viruses Under Control.- Distributed Evaluation Using Multi-agents.- Classification of Volatile Organic Compounds with Incremental SVMs and RBF Networks.- E-Commerce and Web Services.- Agent Based Dynamic Execution of BPEL Documents.- A Fair Multimedia Exchange Protocol.- A Pervasive Environment for Location-Aware and Semantic Matching Based Information Gathering.- A Web Service Platform for Web-Accessible Archaeological Databases.- A WSDL Extension for Performance-Enabled Description of Web Services.- A Novel Authorization Mechanism for Service-Oriented Virtual Organization.- Metrics, Methodology, and Tool for Performance-Considered Web Service Composition.- Brazilian Software Process Reference Model and Assessment Method.- Multiagent Systems.- A Secure Communication Framework for Mobile Agents.- A Novel Algorithm for the Coordination of Multiple Mobile Robots.- Multiagent Elite Search Strategy for Combinatorial Optimization Problems.- Managing Theories of Trust in Agent Based Systems.- Applying Semantic Capability Matching into Directory Service Structures of Multi Agent Systems.- Self-organizing Distribution of Agents over Hosts.- Machine Learning.- Evolutionary Design of Group Communication Schedules for Interconnection Networks.- Memetic Algorithms for Nurse Rostering.- Discretizing Continuous Attributes Using Information Theory.- System Identification Using Genetic Programming and Gene Expression Programming.- ARKAQ-Learning: Autonomous State Space Segmentation and Policy Generation.- Signature Verification Using Conic Section Function Neural Network.- Fusion of Rule-Based and Sample-Based Classifiers - Probabilistic Approach.- Construction of a Learning Automaton for Cycle Detection in Noisy Data Sequences.- Information Retrieval and Natural Language Processing.- A New Trend Heuristic Time-Variant Fuzzy Time Series Method for Forecasting Enrollments.- Using GARCH-GRNN Model to Forecast Financial Time Series.- Boosting Classifiers for Music Genre Classification.- Discriminating Biased Web Manipulations in Terms of Link Oriented Measures.- ORF-NT: An Object-Based Image Retrieval Framework Using Neighborhood Trees.- Text Categorization with Class-Based and Corpus-Based Keyword Selection.- Aligning Turkish and English Parallel Texts for Statistical Machine Translation.- The Effect of Windowing in Word Sense Disambiguation.- Pronunciation Disambiguation in Turkish.- Image and Speech Processing.- Acoustic Flow and Its Applications.- A DCOM-Based Turkish Speech Recognition System: TREN - Turkish Recognition ENgine.- Speaker Recognition in Unknown Mismatched Conditions Using Augmented PCA.- Real Time Isolated Turkish Sign Language Recognition from Video Using Hidden Markov Models with Global Features.- An Animation System for Fracturing of Rigid Objects.- 2D Shape Tracking Using Algebraic Curve Spaces.- A Multi-camera Vision System for Real-Time Tracking of Parcels Moving on a Conveyor Belt.- Selection and Extraction of Patch Descriptors for 3D Face Recognition.- Implementation of a Video Streaming System Using Scalable Extension of H.264.- Blotch Detection and Removal for Archive Video Restoration.- Performance Study of an Image Restoration Algorithm for Bursty Mobile Satellite Channels.- Algorithms and Database Systems.- Polymorphic Compression.- Efficient Adaptive Data Compression Using Fano Binary Search Trees.- Word-Based Fixed and Flexible List Compression.- Effective Early Termination Techniques for Text Similarity Join Operator.- Multimodal Video Database Modeling, Querying and Browsing.- Semantic Load Shedding for Prioritized Continuous Queries over Data Streams.- Probabilistic Point Queries over Network-Based Movements.- Effective Clustering by Iterative Approach.- Recursive Lists of Clusters: A Dynamic Data Structure for Range Queries in Metric Spaces.- Incremental Clustering Using a Core-Based Approach.- Indexing of Sequences of Sets for Efficient Exact and Similar Subsequence Matching.- An Investigation of the Course-Section Assignment Problem.- Crympix: Cryptographic Multiprecision Library.- Optimal Control for Real-Time Feedback Rate-Monotonic Schedulers.- Graphical User Interface Development on the Basis of Data Flows Specification.- Theory of Computing.- Generalizing Redundancy Elimination in Checking Sequences.- A Computable Version of Dinis Theorem for Topological Spaces.- Improved Simulation of Quantum Random Walks.- An Alternative Proof That Exact Inference Problem in Bayesian Belief Networks Is NP-Hard.- Recovering the Lattice of Repetitive Sub-functions.- Epilogue.- Erol Gelenbes Career and Contributions.

Adaptive anti-spam filtering for agglutinative languages: a special case for Turkish

We propose anti-spare filtering methods for agglutinative languages in general and for Turkish in particular. The methods are dynamic and are based on Artificial Neural Networks (ANN) and Bayesian Networks. The developed algorithms are user-specific and adapt themselves with the characteristics of the incoming e-mails. The algorithms have two main components. The first one deals with the morphology of the words and the second one classifies the e-mails by using the roots of the words extracted by the morphological analysis. Two ANN structures, single layer perceptron and multi-layer perceptron, are considered and the inputs to the networks are determined using binary model and probabilistic model. Similarly, for Bayesian classification, three different approaches are employed: binary model, probabilistic model, and advanced probabilistic model. In the experiments, a total of 750 e-mails (410 spare and 340 normal) were used and a success rate of about 90% was achieved.

Ensemble of SVMs for incremental learning

Support Vector Machines (SVMs) have been successfully applied to solve a large number of classification and regression problems. However, SVMs suffer from the catastrophic forgetting phenomenon, which results in loss of previously learned information. Learn++ have recently been introduced as an incremental learning algorithm. The strength of Learn++ lies in its ability to learn new data without forgetting previously acquired knowledge and without requiring access to any of the previously seen data, even when the new data introduce new classes. To address thecatastrophic forgetting problem and to add the incremental learning capability to SVMs, we propose using an ensemble of SVMs trained with Learn++. Simulation results on real-world and benchmark datasets suggest that the proposed approach is promising.

A feature selection method based on kernel canonical correlation analysis and the minimum Redundancy-Maximum Relevance filter method

Highlights? We propose a feature selection method based on a recently popular mRMR criterion, which we called KCCAmRMR. ? Our method is based on finding the unique information that a candidate variable possesses about the target variable. ? We propose using correlated functions explored by KCCA instead of using the features themselves as inputs to mRMR. In this paper, we propose a feature selection method based on a recently popular minimum Redundancy-Maximum Relevance (mRMR) criterion, which we called Kernel Canonical Correlation Analysis basedmRMR (KCCAmRMR) based on the idea of finding the unique information, i.e. information that is distinct from the set of already selected variables, that a candidate variable possesses about the target variable. In simplest terms, for this purpose, we propose using correlated functions explored by KCCA instead of using the features themselves as inputs to mRMR. We demonstrate the usefulness of our method on both toy and benchmark datasets.

Performance evaluation of evolutionary heuristics in dynamic environments

In recent years, there has been a growing interest in applying genetic algorithms to dynamic optimization problems. In this study, we present an extensive performance evaluation and comparison of 13 leading evolutionary algorithms with different characteristics on a common platform by using the moving peaks benchmark and by varying a set of problem parameters including shift length, change frequency, correlation value and number of peaks in the landscape. In order to compare solution quality or the efficiency of algorithms, the results are reported in terms of both offline error metric and dissimilarity factor, our novel comparison metric presented in this paper, which is based on signal similarity. Computational effort of each algorithm is reported in terms of average number of fitness evaluations and the average execution time. Our experimental evaluation indicates that the hybrid methods outperform the related work with respect to quality of solutions for various parameters of the given benchmark problem. Specifically, hybrid methods provide up to 24% improvement with respect to offline error and up to 30% improvement with respect to dissimilarity factor by requiring more computational effort than other methods.

IUGR detection by ultrasonographic examinations using neural networks

This article presents a study that supports a computer-based diagnostic approach to detection of intrauterine growth retardation (IUGR). As an aid to clinical decisions, fetuses that are truly growth retarded and at risk for increased morbidity and mortality should be differentiated from those who have reached their genetic growth potential and are not at increased risk. A wide variety of mathematical formulas (or composite tables) have been proposed for the estimation of fetal weight from ultrasonographic measurements. For these formulas, the timing of the examinations to estimate fetal weight has become controversial due to the poor correlation of early results with the outcomes several weeks later, and also the technical difficulty and poor reproduction of late results. Among the attempts to improve accuracy, one may use more accurate estimated fetal-weight formulas or a single biometric parameter to identify growth abnormalities. This study confirms the following results: 1) in the ultrasound examination the prediction using multiple parameters is better than the prediction using a single parameter; 2) the experiments also show that multiple examinations give a better insight for the diagnosis of IUGR than does a single examination; 3) a neural net is a very helpful tool for correlating many variables.

Fuzzy Support Vector Machines for ECG Arrhythmia Detection

Besides cardiovascular diseases, heart attacks are the main cause of death around the world. Pre-monitoring or pre-diagnostic helps to prevent heart attacks and strokes. ECG plays a key role in this regard. In recent studies, SVM with different kernel functions and parameter values are applied for classification on ECG data. The classification model of SVM can be improved by assigning membership values for inputs. SVM combined with fuzzy theory, FSVM, is exercised on UCI Arrhythmia Database. Five different membership functions are defined. It is shown that the accuracy of classification can be improved by defining appropriate membership functions. ANFIS is used in order to interpret the resulting classification model. The ANFIS model of the ECG data is compared to and found consistent with the medical knowledge.

Neural-network-based decision making in diagnostic applications

In this article the NN approach for medical decision making was applied for three specific examples. The first example was decision making with single-valued data for IUGR detection. The second example was decision making with double-valued data in prediction of ovulation. The third example was the use of independent NN modules and consensus theory for prediction of ovulation time. The NN approach has superiority over classical statistical approaches for decision making with medical data for the following reasons: 1. It is distribution-free. 2. It captures correlative features and does not need any specific consideration for mutual test dependence. 3. It provides weighted reliability of various tests. 4. It produces fast, accurate results. The statistical decision approach will probably outperform the NN approach in making decisions when an accurate distribution model is provided. However, the NN is proposed as a useful tool to help physicians in decision making and diagnosis of certain symptoms. The capability and performance of this tool has generally been proven in combining mutually dependent medical tests.

Performance evaluation of digital audio watermarking techniques designed in time, frequency and cepstrum domains

This study investigates the performances of a variety of audio watermarking techniques designed in time, frequency and cepstrum domains. A framework of comparison is performed with respect to bit error rate (BER), objective and subjective perceptual quality, computational complexity and robustness to signal processing such as low-pass filtering, requantization and MPEG Layer 3 (MP3) compression. It is observed that the cepstrum domain technique is superior to other techniques in terms of almost all criteria. Additionally, a new watermark embedding technique is introduced that combines frequency hopping spread spectrum (FHSS) and frequency masking (FM) techniques. It is experimentally concluded that the proposed technique performs fairly well and is robust to MP3 compression.