Person:
ERDEM, Arif Tanju

Profile Picture

Email Address

Birth Date

WoSScopusGoogle ScholarORCID

Name

Job Title

First Name

Arif Tanju

Last Name

ERDEM
OrgUnit Logo
Organizational Unit

Filters

Reset filters

Settings

Publication Search Results

Now showing 1 - 10 of 21
  • Placeholder
    ArticlePublicationMetadata only
    Bispectrum estimation using a MISO autoregressive model
    (Springer International Publishing, 2016) Erdem, Tanju; Ercan, Ali Özer; Electrical & Electronics Engineering; Computer Science; ERDEM, Arif Tanju; ERCAN, Ali Özer
    Bispectra are third-order statistics that have been used extensively in analyzing nonlinear and non-Gaussian data. Bispectrum of a process can be computed as the Fourier transform of its bicumulant sequence. It is in general hard to obtain reliable bicumulant samples at high lags since they suffer from large estimation variance. This paper proposes a novel approach for estimating bispectrum from a small set of given low lag bicumulant samples. The proposed approach employs an underlying MISO system composed of stable and causal autoregressive components. We provide an algorithm to compute the parameters of such a system from the given bicumulant samples. Experimental results show that our approach is capable of representing non-polynomial spectra with a stable underlying system model, which results in better bispectrum estimation than the leading algorithm in the literature.
  • Placeholder
    Conference ObjectPublicationMetadata only
    Automatic fall detection for elderly by using features extracted from skeletal data
    (IEEE, 2013) Davari, Amir; Aydin, T; Erdem, Tanju; Computer Science; ERDEM, Arif Tanju; Davari, Amir
    Automatic detection of unusual events such as falls is very important especially for elderly people living alone. Realtime detection of these events can reduce the health risks associated with a fall. In this paper, we propose a novel method for automatic detection of fall event by using depth cameras. Depth images generated by these cameras are used in computing the skeletal data of a person. Our contribution is to use features extracted from the skeletal data to form a strong set of features which can help us achieve an increased precision at low redundancy. Our findings indicate that our features, which are derived from skeletal data, are moderately powerful for detecting unusual events such as fall.
  • Placeholder
    Conference ObjectPublicationMetadata only
    Hareketten yapı çıkarımı için görsel kullanıcı arayüzü
    (IEEE, 2014) Hüroğlu, Cengiz; Erdem, Tanju; Computer Science; ERDEM, Arif Tanju; Hüroğlu, Cengiz
    The usage of computer vision applications such as 3D reconstruction, motion tracking and augmented reality gradually increases. The first and the most important stage of these kind of applications is estimating the 3D scene model and motion information. We developed an easy-to-use user interface in order to use in these kind of applications. The user interface we developed, contains important functionalities such as robust feature detection, feature matching and extracting 3D structure from motion. We benefit from some open-source computer vision tools such as SiftGPU and Bundler and integrate them in our user interface application, so they can easily be used.
  • Placeholder
    Conference ObjectPublicationMetadata only
    Curriculum learning for face recognition
    (IEEE, 2021) Büyüktaş, Barış; Erdem, Ç. E.; Erdem, Tanju; Computer Science; ERDEM, Arif Tanju
    We present a novel curriculum learning (CL) algorithm for face recognition using convolutional neural networks. Curriculum learning is inspired by the fact that humans learn better, when the presented information is organized in a way that covers the easy concepts first, followed by more complex ones. It has been shown in the literature that that CL is also beneficial for machine learning tasks by enabling convergence to a better local minimum. In the proposed CL algorithm for face recognition, we divide the training set of face images into subsets of increasing difficulty based on the head pose angle obtained from the absolute sum of yaw, pitch and roll angles. These subsets are introduced to the deep CNN in order of increasing difficulty. Experimental results on the large-scale CASIA-WebFace-Sub dataset show that the increase in face recognition accuracy is statistically significant when CL is used, as compared to organizing the training data in random batches.
  • Placeholder
    Conference ObjectPublicationMetadata only
    A multi-sensor integrated head-mounted display setup for augmented reality applications
    (IEEE, 2015) Kermen, Ahmet; Aydın, T.; Ercan, Ali Özer; Erdem, Tanju; Electrical & Electronics Engineering; Computer Science; ERCAN, Ali Özer; ERDEM, Arif Tanju; Kermen, Ahmet
    We present an HMD based AR system comprising visual and inertial sensors. The visual sensor is a camera pair and the inertial sensors consist of an accelerometer and a gyroscope. We discuss the temporal and spatial calibration issues that relate to such a system. We introduce simple yet effective methods for estimating the time lag between the camera and the inertial sensors and for estimating the relative pose between the camera and the inertial sensors. These methods do not require a complicated setup for data collection and involve simple equations to solve. Sample results are presented to demonstrate the visual performance of the system.
  • Placeholder
    Conference ObjectPublicationMetadata only
    INTERSPEECH 2009 duygu tanıma yarışması değerlendirmesi
    (IEEE, 2010) Bozkurt, E.; Erzin, E.; Eroğlu Erdem, Ç.; Erdem, Tanju; Computer Science; ERDEM, Arif Tanju
    Bu makalede INTERSPEECH 2009 Duygu Tanıma Yarışması sonuçlarını değerlendiriyoruz. Yarışmanın sunduğu problem doğal ve duygu bakımından zengin FAU Aibo konuşma kayıtlarının beş ve iki duygu sınıfına en doğru şekilde ayrılmasıdır. Bu problemi çözmek için bürün ilintili, spektral ve SMM-temelli (sakl Markov model) öznitelikleri Gauss Bileşen Model (GBM) sınıflandırıcılar ile inceliyoruz. Spektral öznitelikler, Mel frekans kepstral katsayıların (MFKK), doru spektral frekans (DSF) katsayılarını ve bunların türevlerini içerirken, bürün öznitelikleri perde, perdenin birinci türevi ve enerjiden oluşuyor. Bürün ilintili özniteliklerin zamanla değimini tanımlayan SMM özniteliklerini, güdümsüz eğitilen SMM yapılar ile elde ediyoruz. Ayrıca, konuşmadan duygu tanıma sonuçların iyileştirmek için farklı özniteliklerin veri kaynaşımın ve farklı sınıflandırıcıların karar kaynaşımını da inceliyoruz. İki aşamalı karar kaynaşım yöntemimiz beş ve iki sınıflı problemler için sırasıyla,% 41.59 ve %67.90 başarım oranını ve tüm yarışma sonuçları arasında 2. ve 4. sırayı elde etti .
  • Placeholder
    Conference ObjectPublicationMetadata only
    Improving automatic emotion recognition from speech signals
    (International Speech Communications Association, 2009) Bozkurt, E.; Erzin, E.; Eroğlu Erdem, Ç.; Erdem, Tanju; Computer Science; ERDEM, Arif Tanju
    We present a speech signal driven emotion recognition system. Our system is trained and tested with the INTERSPEECH 2009 Emotion Challenge corpus, which includes spontaneous and emotionally rich recordings. The challenge includes classifier and feature sub-challenges with five-class and two-class classification problems. We investigate prosody related, spectral and HMM-based features for the evaluation of emotion recognition with Gaussian mixture model (GMM) based classifiers. Spectral features consist of mel-scale cepstral coefficients (MFCC), line spectral frequency (LSF) features and their derivatives, whereas prosody-related features consist of mean normalized values of pitch, first derivative of pitch and intensity. Unsupervised training of HMM structures are employed to define prosody related temporal features for the emotion recognition problem. We also investigate data fusion of different features and decision fusion of different classifiers, which are not well studied for emotion recognition framework. Experimental results of automatic emotion recognition with the INTERSPEECH 2009 Emotion Challenge corpus are presented.
  • Placeholder
    ArticlePublicationMetadata only
    More learning with less labeling for face recognition
    (Elsevier, 2023-05) Büyüktaş, Barış; Eroğlu Erdem, Ç.; Erdem, Tanju; Computer Science; ERDEM, Arif Tanju; Büyüktaş, Barış
    In this paper, we propose an improved face recognition framework where the training is started with a small set of human annotated face images and then new images are incorporated into the training set with minimum human annotation effort. In order to minimize the human annotation effort for new images, the proposed framework combines three different strategies, namely self-paced learning (SPL), active learning (AL), and minimum sparse reconstruction (MSR). As in the recently proposed ASPL framework [1], SPL is used for automatic annotation of easy images, for which the classifiers are highly confident and AL is used to request the help of an expert for annotating difficult or low-confidence images. In this work, we propose to use MSR to subsample the low-confidence images based on diversity using minimum sparse reconstruction in order to further reduce the number of images that require human annotation. Thus, the proposed framework provides an improvement over the recently proposed ASPL framework [1] by employing MSR for eliminating “similar” images from the set selected by AL for human annotation. Experimental results on two large-scale datasets, namely CASIA-WebFace-Sub and CACD show that the proposed method called ASPL-MSR can achieve similar face recognition performance by using significantly less expert-annotated data as compared to the state-of-the-art. In particular, ASPL-MSR requires manual annotation of only 36.10% and 54.10% of the data in CACD and CASIA-WebFace-Sub datasets, respectively, to achieve the same face recognition performance as the case when the whole training data is used with ground truth labels. The experimental results indicate that the number of manually annotated samples have been reduced by nearly 4% and 2% on the two datasets as compared to ASPL [1].
  • Placeholder
    Book PartPublicationMetadata only
    RANSAC-based training data selection on spectral features for emotion recognition from spontaneous speech
    (Springer International Publishing, 2011) Bozkurt, E.; Erzin, E.; Erdem, Tanju; Eroğlu Erdem, Ç.; Computer Science; ERDEM, Arif Tanju
    Training datasets containing spontaneous emotional speech are often imperfect due the ambiguities and difficulties of labeling such data by human observers. In this paper, we present a Random Sampling Consensus (RANSAC) based training approach for the problem of emotion recognition from spontaneous speech recordings. Our motivation is to insert a data cleaning process to the training phase of the Hidden Markov Models (HMMs) for the purpose of removing some suspicious instances of labels that may exist in the training dataset. Our experiments using HMMs with Mel Frequency Cepstral Coefficients (MFCC) and Line Spectral Frequency (LSF) features indicate that utilization of RANSAC in the training phase provides an improvement in the unweighted recall rates on the test set. Experimental studies performed over the FAU Aibo Emotion Corpus demonstrate that decision fusion configurations with LSF and MFCC based classifiers provide further significant performance improvements.
  • Placeholder
    ArticlePublicationMetadata only
    Formant position based weighted spectral features for emotion recognition
    (Elsevier, 2011) Bozkurt, E.; Erzin, E.; Eroğlu Erdem, Ç.; Erdem, Tanju; Computer Science; ERDEM, Arif Tanju
    In this paper, we propose novel spectrally weighted mel-frequency cepstral coefficient (WMFCC) features for emotion recognition from speech. The idea is based on the fact that formant locations carry emotion-related information, and therefore critical spectral bands around formant locations can be emphasized during the calculation of MFCC features. The spectral weighting is derived from the normalized inverse harmonic mean function of the line spectral frequency (LSF) features, which are known to be localized around formant frequencies. The above approach can be considered as an early data fusion of spectral content and formant location information. We also investigate methods for late decision fusion of unimodal classifiers. We evaluate the proposed WMFCC features together with the standard spectral and prosody features using HMM based classifiers on the spontaneous FAU Aibo emotional speech corpus. The results show that unimodal classifiers with the WMFCC features perform significantly better than the classifiers with standard spectral features. Late decision fusion of classifiers provide further significant performance improvements.