Browsing by Subject "Prediction"

Now showing 1 - 20 of 25

Open Access
Adaptive prediction and vector quantization based very low bit rate video codec
(1993) Ulukuş, Şennur
A very low bit rate video codec (coder/decoder) based on motion compensation, adaptive prediction, vector quantization (VQ) and entropy coding, and a new prediction scheme based on Gibbs random field (GRF) model are presented. The codec is specifically designed for the video-phone application for which the main constraint is to transmit the coded bit stream via the existing telephone lines. Proposed codec can operate in the transmission bit rate interval ranging from 8 to 32 Kbits/s which is defined as the very low bit rates for video coding. Four different coding strategies are adapted to the system, and depending on the characteristics of the image data in the block one of these coding methods is chosen by the coder. Linear prediction is implemented in the codec, and the performances of the two prediction schemes are compared at several transmission bit rates. The need for any prediction is also questioned, by implementing the same codec structure without prediction and comparing the performances of the codecs with prediction and without prediction. It is proved that the presented codec can be used in transmitting the video signal via the existing telephone network for the video-phone applications. Also, it is observed that the codec with GRF model based non-linear prediction has a better performance compared to the codec with linear prediction.
Open Access
Artificial neural network and decision tree–based models for prediction and validation of in vitro organogenesis of two hydrophytes—Hemianthus callitrichoides and Riccia fluitans
(Springer, 2023-08-02) Özcan, Esra; Atar, Hasan Hüseyin; Ali, Seyid Amjad; Aasim, Muhammad
The application of plant tissue culture protocols for aquatic plants has been widely adopted in recent years to produce cost-effective plants for aquarium industry. In vitro regeneration protocol for the two different hydrophytes Hemianthus callitrichoides (Cuba) and Riccia fluitans were optimized for appropriate basal medium, sucrose, agar, and plant growth regulator concentration. The MS No:3B and SH + MSVit basal medium yielded a maximum clump diameter of 5.53 cm for H. callitrichoides and 3.65 cm for R. fluitans. The application of 20 g/L sucrose was found appropriate for yielding larger clumps in both species. Solidification of the medium with 1 g/L agar was optimized for inducing larger clumps with rooting for both species. Provision of basal medium with any concentration of 6-benzylaminopurine (BAP) and α-naphthaleneacetic acid (NAA) was found detrimental for inducing larger clumps for both species. The largest clumps of H. callitrichoides (5.51 cm) and R. fluitans (4.59 cm) were obtained on basal medium without any plant growth regulators. The attained data was also predicted and validated by employing multilayer perceptron (MLP), random forest (RF), and extreme gradient boosting (XGBoost) algorithms. The performance of the models was tested with three different performance metrics, namely, coefficient of regression (R2), means square error (MSE), and mean absolute error (MAE). Results revealed that MLP and RF models performed better than the XGBoost model. The protocols developed in this study have shown promising outcomes and the findings can irrefutably assist to produce H. callitrichoides and R. fluitans on a large scale for the local aquarium industry.
Open Access
Audio–visual predictive processing in the perception of humans and robots
(Springer Science and Business Media B.V., 2023-04-05) Sarıgül, B.; Urgen, Burcu A.
Recent work in cognitive science suggests that our expectations affect visual perception. With the rise of artificial agents in human life in the last few decades, one important question is whether our expectations about non-human agents such as humanoid robots affect how we perceive them. In the present study, we addressed this question in an audio–visual context. Participants reported whether a voice embedded in a noise belonged to a human or a robot. Prior to this judgment, they were presented with a human or a robot image that served as a cue and allowed them to form an expectation about the category of the voice that would follow. This cue was either congruent or incongruent with the category of the voice. Our results show that participants were faster and more accurate when the auditory target was preceded by a congruent cue than an incongruent cue. This was true regardless of the human-likeness of the robot. Overall, these results suggest that our expectations affect how we perceive non-human agents and shed light on future work in robot design.
Open Access
Behavioral and neural investigation on the effects of prior information on biological motion perception
(2023-07) Elmas, Hüseyin Orkun
The capacity to understand the actions of others, a cognitive phenomenon known as biological motion perception, is crucial for humans. Recent research demonstrates that biological motion is processed distinctively compared to the motions of inanimate objects. A dedicated brain network for processing biological motion and actions has been uncovered through fMRI studies. M/EEG studies have revealed time windows within which biological motion processing occurs. Despite these findings, a comprehensive understanding of the fundamental mechanisms driving biological motion perception, especially the effects of top-down processes, and the temporal dimension of these effects still remain unexplored. Recent evidence in visual perception suggests that prior knowledge and expectations affect visual perception; however, the generalizability of these effects to socially important stimuli, such as biological motion, is still unknown. This study aims to illuminate the effects of prior information on the behavioral and neural mechanisms of biological motion perception. To this end, we conducted a series of behavioral experiments and an EEG experiment to investigate the effects of prior information on biological motion perception. Through our behavioral experiments, we found that prior information influences the individuation process of biological motion, albeit conditionally. Specifically, this influence is observed only when the cue carries information about the type of action in the biological motion stimuli, and the reliability of the cue is high, at 75%. Our EEG experiment demonstrated that correct and incorrect prior information affects the temporal dimension of biological motion perception, suggesting an early effect of prior information during biological motion processing. More-over, a comparison of the temporal generalization matrices suggested that correct prior information accelerates biological motion perception by accelerating the for-mation of related representations in the brain relative to the neutral condition. Additionally, the temporal generalization analysis results illustrate a sequence in representations within brain activity: the representation of location information precedes the representation of action type of biological motion. These results suggest that computational models, developed to model the underlying mechanisms of biological motion perception, should consider the implications of predictive processes and their temporal dimension. Furthermore, these findings support the applicability of predictive models to not only low-level stimuli but also to higher-level stimuli.
Open Access
Çağrı merkezi metin madenciliği yaklaşımı
(IEEE, 2017-05) Yiğit, İ. O.; Ateş, A. F.; Güvercin, Mehmet; Ferhatosmanoğlu, Hakan; Gedik, Buğra
Günümüzde çağrı merkezlerindeki görüşme kayıtlarının sesten metne dönüştürülebilmesi görüşme kaydı metinleri üzerinde metin madenciliği yöntemlerinin uygulanmasını mümkün kılmaktadır. Bu çalışma kapsamında görüşme kaydı metinleri kullanarak görüşmenin içeriğinin duygu yönünden (olumlu/olumsuz) değerlendirilmesi, müşteri memnuniyetinin ve müşteri temsilcisi performansının ölçülmesi amaçlanmaktadır. Yapılan çalışmada görüşme kaydı metinlerinden metin madenciliği yöntemleri ile yeni özellikler çıkarılmıştır. Metinlerden elde edilen özelliklerden yararlanılarak sınıflandırma ve regresyon yöntemleriyle görüşme kayıtlarının içeriklerinin değerlendirilmesini sağlayacak tahmin modelleri oluşturulmuştur. Bu çalışma sonucunda ortaya çıkarılan tahmin modellerinin Türk Telekom bünyesindeki çağrı merkezlerinde kullanılması hedeflenmektedir.
Open Access
Deep learning-based QoE prediction for streaming services in mobile networks
(IEEE, 2022-11-15) Huang, Gan; Erçetin, Özgür; Gökcesu, Hakan; Kalem, Gökhan
Video streaming accounts for the most of the global Internet traffic and providing a high user Quality of Experience (QoE) is considered an essential target for mobile network operators (MNOs). QoE strongly depends on network Quality of Service (QoS) parameters. In this work, we use real-world network traces obtained from a major cellular operator in Turkey to establish a mapping from network side parameters to the user QoE. To this end, we use a model-aided deep learning method for first predicting channel path loss, and then, employ this prediction for predicting video streaming MOS. The experimental results demonstrate that the proposed model-aided deep learning model can guarantee higher prediction accuracy compared to predictions only relying on mathematical models. We also demonstrate that even though a trained model cannot be directly transferred from one geographical area to another, they significantly reduce the volume of required training when used for prediction in a new area.
Open Access
Does the appearance of an agent affect how we perceive his/her voice? Audio-visual predictive processes in human-robot interaction
(IEEE Computer Society, 2020) Sarıgül, Büşra; Saltık, İmge; Hokelek, Batuhan; Ürgen, Burcu A.
Robots increasingly become part of our lives. How we perceive and predict their behavior has been an important issue in HRI. To address this issue, we adapted a well-established prediction paradigm from cognitive science for HRI. Participants listened a greeting phrase that sounds either human-like or robotic. They indicated whether the voice belongs to a human or a robot as fast as possible with a key press. Each voice was preceded with a human or robot image (a human-like robot or a mechanical robot) to cue the participant about the upcoming voice. The image was either congruent or incongruent with the sound stimulus. Our findings show that people reacted faster to robotic sounds in congruent trials than incongruent trials, suggesting the role of predictive processes in robot perception. In sum, our study provides insights about how robots should be designed, and suggests that designing robots that do not violate our expectations may result in a more efficient interaction between humans and robots.
Open Access
An eager regression method based on best feature projections
(Springer, Berlin, Heidelberg, 2001) Aydın, Tolga; Güvenir, H. Altay
This paper describes a machine learning method, called Regression by Selecting Best Feature Projections (RSBFP). In the training phase, RSBFP projects the training data on each feature dimension and aims to find the predictive power of each feature attribute by constructing simple linear regression lines, one per each continuous feature and number of categories per each categorical feature. Because, although the predictive power of a continuous feature is constant, it varies for each distinct value of categorical features. Then the simple linear regression lines are sorted according to their predictive power. In the querying phase of learning, the best linear regression line and thus the best feature projection are selected to make predictions. © Springer-Verlag Berlin Heidelberg 2001.
Open Access
End-to-end hybrid architectures for effective sequential data prediction
(2023-08) Aydın, Mustafa Enes
We investigate nonlinear prediction in an online setting and introduce two hybrid models that effectively mitigate, via end-to-end architectures, the need for hand-designed features and manual model selection issues of conventional nonlinear prediction/regression methods. Particularly, we first use an enhanced recurrent neural network (LSTM) to extract features from sequential signals, while pre-serving the state information, i.e., the history, and soft gradient boosted decision trees (sGBDT) to produce the final output. The connection is in an end-to-end fashion and we jointly optimize the whole architecture using stochastic gradient descent. Secondly, we again use recursive structures (LSTM) for automatic fea-ture extraction out of raw data but accompany it with a traditional linear time series model (SARIMAX) to deal with the intricacies of the sequential data, e.g., seasonality. The unification of the models is again in a joint manner; it is through a single state space and we optimize the entire architecture using particle filter-ing. The proposed frameworks are generic so that one can use other recurrent architectures, e.g., GRUs, and differentiable machine learning algorithms as well as time series models that have state space representations in lieu of the specific models presented. We demonstrate the learning behavior of the models on syn-thetic data and the significant performance improvements over the conventional methods and the disjoint counterparts over various real life datasets, with which we also show the generic nature of the frameworks. Furthermore, we openly share the source code of the proposed methods to facilitate further research.
Open Access
Estimating the chance of success and suggestion for treatment in IVF
(2013) Mısırlı, Gizem
In medicine, the chance of success for a treatment is important for decision making for the doctor and the patient. This thesis focuses on the domain of In Vitro Fertilization (IVF), where there are two issues: the first one is the decision on whether or not go with the treatment procedure, the second one is the selection of the proper treatment protocol for the patient. It is important for both the doctor and the couple to have some idea about the chance of success of the treatment after the initial evaluation. If the chance of success is low, the patient couple may decide not to proceed with this stressful and expensive treatment. Once a decision for treatment is made, the next issue for the doctors is the choice of the treatment protocol which is the most suitable for the couple. Our first aim is to develop techniques to estimate the chance of success and determine the factors that affect the success in IVF treatment. So, we employ ranking algorithms to estimate the chance of success. The ranking methods used are RIMARC (Ranking Instances by Maximizing the Area under the ROC Curve), SVMlight (Support Vector Machine Ranking Algorithm) and RIkNN (Ranking Instances using k Nearest Neighbour). All of these three algorithms learn a model to rank the instances based on their score values. RIMARC is a method for ranking instances by maximizing the area under the ROC curve. SVMlight is an implementation of Support Vector Machine for ranking instances. RIkNN is a k Nearest Neighbour (kNN) based algorithm that is developed for ranking instances based on similarity metric. We also used RIwkNN, which is the version of RIkNN where the features are assigned weights by experts in the domain. These algorithms are compared on the basis of the AUC of 10-fold stratified cross-validation. Moreover, these ranking algorithms are modified as a classification algorithm and compared on the basis of the accuracy of 10-fold stratified cross-validation. As a by-product, the RIMARC algorithm learns the factors that affect the success in IVF treatment. It calculates feature weights and creates rules that are in a human readable form and easy to interpret. After a decision for a treatment is made, the second aim is to determine which treatment protocol is the most suitable for the couple. In IVF treatment, many different types of drugs and dosages are used, however, which drug and the dosage are the most suitable for the given patient is not certain. Doctors generally make their decision based on their past experiences and the results of research published all over the world. To the best of our knowledge, there are no methods for learning a model that can be used to suggest the best feature values to increase the chance that the class label to be the desired one. We will refer to such a system as Suggestion System. To help doctors in making decision on the selection of the suitable treatment protocols, we present three suggestion systems that are based on well-known machine learning techniques. We will call the suggestion systems developed as a part of this work as NSNS (Nearest Successful Neighbour Based Suggestion), kNNS (k Nearest Neighbour Based Suggestion) and DTS (Decision Tree Based Suggestion). We also implemented the weighted version of NSNS using feature weights that are produced by the RIMARC algorithm. Moreover, we propose performance metrics for the evaluation of the suggestion algorithms. We introduce four evaluation metrics namely; pessimistic metric (mp), optimistic metric (mo), validated optimistic metric (mvo) and validated pessimistic metric (mvp) to test the correctness of the algorithms. In order to help doctors to utilize developed algorithms, we develop a decision support system, called RAST (Risk Analysis and Suggestion for Treatment). This system is actively being used in the IVF center at Etlik Z¨ubeyde Hanım Woman’s Health and Teaching Hospital.
Open Access
Expression of IFITM1 in chronic myeloid leukemia patients
(Elsevier, 2005) Akyerli, C. B.; Beksac, M.; Holko, M.; Frevel, M.; Dalva, K.; Özbek, U.; Soydan, E.; Özcan, M.; Özet, G.; İlhan, O.; Gürman, G.; Akan, H.; Williams, B. R. G.; Özçelik, T.
We investigated the peripheral blood gene expression profile of interferon induced transmembrane protein 1 (IFITM1) in sixty chronic myeloid leukemia (CML) patients classified according to new prognostic score (NPS). IFITM1 is a component of a multimeric complex involved in the trunsduction of antiproliferative and cell adhesion signals. Expression level of IFITM1 was found significantly different between the high- and low-risk groups (P = 9.7976 × 10-11) by real-time reverse transcription polymerase chain reaction (RT-PCR). Higher IFITM1 expression correlated with improved survival (P = 0.01). These results indicate that IFITM1 expression profiling could be used for molecular classification of CML, which may also predict survival.
Open Access
A hybrid framework for sequential data prediction with end-to-end optimization
(Elsevier, 2022-08-08) Aydin, M.E.; Kozat, Süleyman S.
We investigate nonlinear prediction in an online setting and introduce a hybrid model that effectively mitigates, via an end-to-end architecture, the need for hand-designed features and manual model selection issues of conventional nonlinear prediction/regression methods. In particular, we use recursive structures to extract features from sequential signals, while preserving the state information, i.e., the history, and boosted decision trees to produce the final output. The connection is in an end-to-end fashion and we jointly optimize the whole architecture using stochastic gradient descent, for which we also provide the backward pass update equations. In particular, we employ a recurrent neural network (LSTM) for adaptive feature extraction from sequential data and a gradient boosting machinery (soft GBDT) for effective supervised regression. Our framework is generic so that one can use other deep learning architectures for feature extraction (such as RNNs and GRUs) and machine learning algorithms for decision making as long as they are differentiable. We demonstrate the learning behavior of our algorithm on synthetic data and the significant performance improvements over the conventional methods over various real life datasets. Furthermore, we openly share the source code of the proposed method to facilitate further research. © 2022 Elsevier Inc.
Open Access
A large-scale sentiment analysis for Yahoo! Answers
(ACM, 2012) Küçüktunç, O.; Cambazoğlu, B. B.; Weber, I.; Ferhatosmanoğlu, Hakan
Sentiment extraction from online web documents has recently been an active research topic due to its potential use in commercial applications. By sentiment analysis, we refer to the problem of assigning a quantitative positive/negative mood to a short bit of text. Most studies in this area are limited to the identification of sentiments and do not investigate the interplay between sentiments and other factors. In this work, we use a sentiment extraction tool to investigate the influence of factors such as gender, age, education level, the topic at hand, or even the time of the day on sentiments in the context of a large online question answering site. We start our analysis by looking at direct correlations, e.g., we observe more positive sentiments on weekends, very neutral ones in the Science & Mathematics topic, a trend for younger people to express stronger sentiments, or people in military bases to ask the most neutral questions. We then extend this basic analysis by investigating how properties of the (asker, answerer) pair affect the sentiment present in the answer. Among other things, we observe a dependence on the pairing of some inferred attributes estimated by a user's ZIP code. We also show that the best answers differ in their sentiments from other answers, e.g., in the Business & Finance topic, best answers tend to have a more neutral sentiment than other answers. Finally, we report results for the task of predicting the attitude that a question will provoke in answers. We believe that understanding factors influencing the mood of users is not only interesting from a sociological point of view, but also has applications in advertising, recommendation, and search. Copyright 2012 ACM.
Open Access
Online learning under adverse settings
(2015-05) Özkan, Hüseyin
We present novel solutions for contemporary real life applications that generate data at unforeseen rates in unpredictable forms including non-stationarity, corruptions, missing/mixed attributes and high dimensionality. In particular, we introduce novel algorithms for online learning, where the observations are received sequentially and processed only once without being stored, under adverse settings: i) no or limited assumptions can be made about the data source, ii) the observations can be corrupted and iii) the data is to be processed at extremely fast rates. The introduced algorithms are highly effective and efficient with strong mathematical guarantees; and are shown, through the presented comprehensive real life experiments, to significantly outperform the competitors under such adverse conditions. We develop a novel highly dynamical ensemble method without any stochastic assumptions on the data source. The presented method is asymptotically guaranteed to perform as well as, i.e., competitive against, the best expert in the ensemble, where the competitor, i.e., the best expert, itself is also specifically designed to continuously improve over time in a completely data adaptive manner. In addition, our algorithm achieves a significantly superior modeling power (hence, a significantly superior prediction performance) through a hierarchical and self-organizing approach while mitigating over training issues by combining (taking finite unions of) low-complexity methods. On the contrary, the state-of-the-art ensemble techniques are heavily dependent on static and unstructured expert ensembles. In this regard, we rigorously solve the resulting issues such as the over sensitivity to source statistics as well as the incompatibility between the modeling power and the computational load/precision. Our results uniformly hold for every possible input stream in the deterministic sense regardless of the stationary or non-stationary source statistics. Furthermore, we directly address the data corruptions by developing novel versatile imputation methods and thoroughly demonstrate that the anomaly detection -in addition to being stand alone an important learning problem- is extremely effective for corruption detection/imputation purposes. To that end, as the first time in the literature, we develop the online implementation of the Neyman-Pearson characterization for anomalies in stationary or non-stationary fast streaming temporal data. The introduced anomaly detection algorithm maximizes the detection power at a specified controllable constant false alarm rate with no parameter tuning in a truly online manner. Our algorithms can process any streaming data at extremely fast rates without requiring a training phase or a priori information while bearing strong performance guarantees. Through extensive experiments over real/synthetic benchmark data sets, we also show that our algorithms significantly outperform the state-of-the-art as well as the most recently proposed techniques in the literature with remarkable adaptation capabilities to non-stationarity.
Open Access
Predicting optimal facility location without customer locations
(ACM, 2017-08) Yilmaz, Emre; Elbaşı, Sanem; Ferhatosmanoğlu, Hakan
Deriving meaningful insights from location data helps businesses make better decisions. One critical decision made by a business is choosing a location for its new facility. Optimal location queries ask for a location to build a new facility that optimizes an objective function. Most of the existing works on optimal location queries propose solutions to return best location when the set of existing facilities and the set of customers are given. However, most businesses do not know the locations of their customers. In this paper, we introduce a new problem setting for optimal location queries by removing the assumption that the customer locations are known. We propose an optimal location predictor which accepts partial information about customer locations and returns a location for the new facility. The predictor generates synthetic customer locations by using given partial information and it runs optimal location queries with generated location data. Experiments with real data show that the predictor can find the optimal location when sufficient information is provided. © 2017 Copyright held by the owner/author(s).
Open Access
Predicting personality traits with semantic structures and LSTM-based neural networks
(Elsevier, 2022-10) Kosan, Muhammed Ali; Karacan, Hacer; Ürgen, Burcu Ayşen
There is a need to obtain more information about target audiences in many areas such as law enforcement agencies, institutions, human resources, and advertising agencies. In this context, in addition to the information provided by individuals, their personal characteristics are also important. In particular, the predictability of personality traits of individuals is seen as a major parameter in making decisions about individuals. Textual and media data in social media, where people produce the most data, can provide clues about people's personal lives, characteristics, and personalities. Each social media environment may contain different assets and structures. Therefore, it is important to make a structural analysis according to the social media platform. There is also a need for a labelled dataset to develop a model that can predict personality traits from social media data. In this study, first, a personality dataset was created which was retrieved from Twitter and labelled with IBM Personality Insight. Then the unstructured data were transformed into meaningful and processable data, LSTM-based prediction models were created with the structural analysis, and evaluations were made on both our dataset and PAN-2015-EN. © 2022 THE AUTHORS
Open Access
Prediction of failure of commercial banks in Turkey
(1996) Yağlı, Bülent
The aim of this study is failure prediction in Turkish Banking Sector. The results of four prediction models are compared to find out the most efficient one. The models used in this study are: Discriminant Analysis, Logit Analysis, Factor-Logistic Analysis and Alternative Accounting Measures for Prediction. According to the results of this study. Discriminant Analysis has the best predictive ability. Logit Analysis, Beaver’s Method and Factor-Logistic Analysis are ranked after the Discriminant Analysis from best to worst predictive ability.
Open Access
Prediction, classification and recommendation in e-health via contextual partitioning
(IEEE, 2021-07-19) Qureshi, Muhammad Anjum
In this paper, we propose a multipurpose contextual partitioning based estimation algorithm. Exploiting the similarities between contexts (side information: such as age, Gender etc.,) related to patient data in healthcare repository or database, multidimensional spheres are generated over Euclidean space. Then, conditional first and second order characteristics are predicted using sample-based mean and covariance. These conditional statistics of particular patient data subset (sphere) serve the following purposes: i) Prediction for missing values (conditional mean), ii) Partitioned principal components for better classification (conditional covariance) and iii) Recommendation for medical Test or physician (conditional covariance). The proposed approach uniformly partitions the context space into spheres, and then, for each sphere estimates the conditional mean and covariance using only the data (excluding the context data) in the selected sphere. Hence, providing three in one solution i.e., Prediction, Classification and Recommendation for healthcare data using conditional probabilistic characteristics. The overall error is decomposed into estimation and approximation errors. In a particular sphere, estimation error is dependent on the number of instances, while approximation error is dependent on the dissimilarity of instances.
Open Access
Pressure-induced interlinking of carbon nanotubes
(American Physical Society, 2000) Yildirim, T.; Gülseren, O.; Kiliç, Ç.; Çıracı, Salim
We predict new forms of carbon consisting of one- and two-dimensional networks of interlinked single-wall carbon nanotubes, some of which are energetically more stable than van der Waals packing of the nanotubes on a hexagonal lattice. These interlinked nanotubes are further transformed with higher applied external pressures to more dense and complicated stable structures, in which curvature-induced carbon sp3 rehybridizations are formed. We also discuss the energetics of the bond formation between nanotubes and the electronic properties of these predicted novel structures.
Open Access
Reduced density matrix approach to phononic dissipation in friction
(2000) Özpineci, A.; Leitner, D. M.; Çıracı, Salim
Understanding mechanisms for energy dissipation from nanoparticles in contact with large samples is a central problem in describing friction microscopically. Calculation of the reduced density matrix appears to be the most suitable method to study such systems that are coupled to a large environment. In this paper, the time evolution of the reduced density matrix has been evaluated for an arbitrary system coupled to a heat reservoir. The formalism is then applied to study the vibrational relaxation following the stick-slip motion of an asperity on a surface. The frequency and temperature dependence of the relaxation time is also determined. Predictions of the reduced density matrix are compared with those obtained by using the Golden Rule approach.