Yazar "Yildirim, Serdar" seçeneğine göre listele

Listeleniyor 1 - 9 / 9
  • [ X ]
    Öğe
    A modified feature selection method based on metaheuristic algorithms for speech emotion recognition
    (Elsevier Sci Ltd, 2021) Yildirim, Serdar; Kaya, Yasin; Kılıç, Fatih
    Feature selection plays an important role to build a successful speech emotion recognition system. In this paper, a feature selection approach which modifies the initial population generation stage of metaheuristic search algorithms, is proposed. The approach is evaluated on two metaheuristic search algorithms, a nondominated sorting genetic algorithm-II (NSGA-II) and Cuckoo Search in the context of speech emotion recognition using Berlin emotional speech database (EMO-DB) and Interactive Emotional Dyadic Motion Capture (IEMOCAP) database. Results show that the presented feature selection algorithms reduce the number of features significantly and are still effective for emotion classification from speech. Specifically, in speaker-dependent experiments of the EMO-DB, recognition rates of 87.66% and 87.20% are obtained using selected features by modified Cuckoo Search and NSGA-II respectively, whereas, for the IEMOCAP database, the accuracies of 69.30% and 68.32% are obtained using SVM classifier. For the speaker-independent experiments, we achieved comparable results for both databases. Specifically, recognition rates of 76.80% and 76.82% for EMO-DB and 59.37% and 59.52% for IEMOCAP using modified NSGA-II and Cuckoo Search respectively. (C) 2020 Elsevier Ltd. All rights reserved.
  • [ X ]
    Öğe
    A novel multi population based particle swarm optimization for feature selection
    (Elsevier, 2021) Kılıç, Fatih; Kaya, Yasin; Yildirim, Serdar
    Feature selection is an integral part of any machine learning system and the success of such systems highly depends on the relevance of features with the target domain. Feature selection can be classified as NP-Hard problem since a large number of possible solutions exists especially when the feature space is high dimensional. In addition to standard feature selection algorithms, evolutionary algorithms have also yielded promising results. In this paper, a novel multi population based particle swarm optimization (MPPSO) is proposed for feature selection. In this method, multi population start with initial solutions generated by random and Relieff based initialization and searches solution space simultaneously using both populations. 26 UCI and 3 ASU datasets are used to evaluate the performance of the method. The results show that MPPSO generally achieves better average classification accuracies than the other algorithms. Specifically, for the datasets with a large number of features, MPPSO achieves the smallest number of selected features with highest classification accuracies compared to other algorithms. (c) 2021 Elsevier B.V. All rights reserved.
  • [ X ]
    Öğe
    A Real-Time Parallel Image Processing Approach on Regular PCs with Multi-Core CPUs
    (Kaunas Univ Technology, 2017) Atasoy, Huseyin; Yildirim, Esen; Yildirim, Serdar; Kutlu, Yakup
    In this paper a parallel image processing and frame rate stabilization approach is proposed. This approach works on a regular PC with a multi-core CPU. It is implemented under. NET Framework and tested on Microsoft Windows 7 operating system, performing several experiments. It is also applied to a face recognition application to increase its image processing performance successfully. Results show that, handled workload when 4 physical cores are used is approximately 5.25 times the workload handled with one core. It is also shown that the approach successfully distributes the workload on CPU cores and produces output at a stable frame rate under both steady and unsteady workloads. This approach can be used for various signal processing or multimedia applications to parallelize their tasks to increase the performance on multi-core CPUs.
  • [ X ]
    Öğe
    Analysis of functional brain connections for positive-negative emotions using phase locking value
    (Springer, 2017) Dasdemir, Yasar; Yildirim, Esen; Yildirim, Serdar
    In this study, we investigate the brain networks during positive and negative emotions for different types of stimulus (audio only, video only and audio + video) in , and bands in terms of phase locking value, a nonlinear method to study functional connectivity. Results show notable hemispheric lateralization as phase synchronization values between channels are significant and high in right hemisphere for all emotions. Left frontal electrodes are also found to have control over emotion in terms of functional connectivity. Besides significant inter-hemisphere phase locking values are observed between left and right frontal regions, specifically between left anterior frontal and right mid-frontal, inferior-frontal and anterior frontal regions; and also between left and right mid frontal regions. ANOVA analysis for stimulus types show that stimulus types are not separable for emotions having high valence. PLV values are significantly different only for negative emotions or neutral emotions between audio only/video only and audio only/audio + video stimuli. Finding no significant difference between video only and audio + video stimuli is interesting and might be interpreted as that video content is the most effective part of a stimulus.
  • [ X ]
    Öğe
    LU triangularization extreme learning machine in EEG cognitive task classification
    (Springer London Ltd, 2019) Kutlu, Yakup; Yayik, Apdullah; Yildirim, Esen; Yildirim, Serdar
    Electroencephalography (EEG) has been used as a promising tool for investigation of brain activity during cognitive processes. The aim of this study is to reveal whether EEG signals can be used for classifying cognitive processes: arithmetic tasks and text reading. A recently introduced EEG database, which is constructed from 18 healthy subjects during a slide show including 60 slides of simple arithmetic tasks and easily readable texts, is used for this purpose. Multi-order difference plot-based time-domain attributes, number of values in specified regions after scattering the sequential difference values with several degrees, are extracted. For classification, improved extreme learning machine (ELM) scheme, namely luELM, by the use of lower-upper triangularization method instead of singular value decomposition which has disadvantages when used with huge data is proposed. As a result, higher accuracy results are achieved with reduced training time for proposed luELM classifier than traditional ELM classifier for both subject-dependent and subject-independent analysis.
  • [ X ]
    Öğe
    Mining Dominance Tree of API Calls for Detecting Android Malware
    (IEEE, 2018) Alam, Shahid; Yildirim, Serdar; Hassan, Mahamat; Sogukpinar, Ibrahim
    According to the recent Symantec threat reports, Android continues to be the most targeted mobile platform, the number of new mobile malware attacks grew by 105% from 2015 to 2016, and the number of new discovered mobile malware variants grew by 54% from 2016 to 2017. A recent McAfee threat report confers that the number of malware families found in the Google play increased by 30% in 2017. There is a need to develop new techniques and methods to stop this inundation of mobile malware attacks. In this paper we propose a new technique named Droid-DomTree that mines dominance tree of API calls in an Android APK for detecting malware. We develop, a sequential model of the dominance tree of API calls and a weighing scheme for assigning weights to each node in the dominance tree for efficient feature selection. A detection rate of 94.3% was obtained with 4 classifiers.
  • [ X ]
    Öğe
    Mining nested flow of dominant APIs for detecting android malware
    (Elsevier, 2020) Alam, Shahid; Alharbi, Soltan Abed; Yildirim, Serdar
    According to the Kaspersky Lab threat report, mobile malware attacks almost doubled in 2018. A study conducted in 2018 by Accenture found malware attacks to be the most expensive to resolve. Android Operating System (OS) is the most dominating platform on mobile devices. This makes Android OS susceptible to malware attacks. We need to develop new techniques and methods to stop this influx of malware attacks. In this paper, we propose a novel technique named DroidDomTree that mines the dominance tree of API (Application programming interface) calls to find similar patterns in Android applications for detecting malware. Dominance is a transitive relation. A dominance tree of API calls highlights a strong flow of path and identifies the nesting structure of APIs and hence emphasizes the importance of certain APIs in an application. It also helps in finding modules and their interaction in an application. If a malicious module is embedded in an application, then this provides strong evidence that the application contains malware. We use these properties and develop a nested model of the dominance tree of API calls and a new scheme for assigning weights to each node in the dominance tree for efficient feature selection. During 10-fold cross-validation, with eight different classifiers using real malware Android applications, DroidDomTree achieved detection rates in the range of 98.1%-99.3% and false positive rates in the range of 1.7%-0.4%. (C) 2019 Elsevier B.V. All rights reserved.
  • [ X ]
    Öğe
    Mininng Dominance Tree of API Calls for Detecting Android Malware
    (Institute of Electrical and Electronics Engineers Inc., 2018) Alam, Shahid; Yildirim, Serdar; Hassan, Mahamat; Sogukpinar, Ibrahim
    According to the recent Symantec threat reports, Android continues to be the most targeted mobile platform, the number of new mobile malware attacks grew by 105% from 2015 to 2016, and the number of new discovered mobile malware variants grew by 54% from 2016 to 2017. A recent McAfee threat report confers that the number of malware families found in the Google play increased by 30% in 2017. There is a need to develop new techniques and methods to stop this inundation of mobile malware attacks. In this paper we propose a new technique named Droid-DomTree that mines dominance tree of API calls in an Android APK for detecting malware. We develop, a sequential model of the dominance tree of API calls and a weighing scheme for assigning weights to each node in the dominance tree for efficient feature selection. A detection rate of 94.3% was obtained with 4 classifiers. © 2018 IEEE.
  • [ X ]
    Öğe
    Music emotion recognition using convolutional long short term memory deep neural networks
    (Elsevier - Division Reed Elsevier India Pvt Ltd, 2021) Hizlisoy, Serhat; Yildirim, Serdar; Tufekci, Zekeriya
    In this paper, we propose an approach for music emotion recognition based on convolutional long short term memory deep neural network (CLDNN) architecture. In addition, we construct a new Turkish emotional music database composed of 124 Turkish traditional music excerpts with a duration of 30 s each and the performance of the proposed approach is evaluated on the constructed database. We utilize features obtained by feeding convolutional neural network (CNN) layers with log-mel filterbank energies and mel frequency cepstral coefficients (MFCCs) in addition to standard acoustic features. Classification results show that the best performance is obtained when the new feature set is combined with the standard features using the long short term memory (LSTM) + deep neural network (DNN) classi fier. The overall accuracy of 99.19% is obtained using the proposed system with 10 fold cross-validation. Specifically, 6.45 points improvement is achieved. Additionally, the results also show that the LSTM + DNN classifier yields 1.61, 1.61 and 3.23 points improvements in music emotion recognition accuracies compared to k-nearest neighbor (k-NN), support vector machine (SVM), and Random Forest classifiers, respectively. (C) 2020 Karabuk University. Publishing services by Elsevier B.V.