Browsing by Author "Kerestecioglu, Feza"

Now showing 1 - 6 of 6

Anomaly detection via machine learning
(Kadir Has Üniversitesi, 2023) Çevik, Mesut; Kerestecioglu, Feza; Çevik, Mesut
Retail companies monitor inventory stock levels regularly and manage stock levels based on forecasted sales to sustain their market position. The accuracy of inventory stocks is critical for retail companies to create a correct strategy. Many retail com- panies try to detect and prevent inventory record inaccuracy caused by employee or customer theft, damage or spoilage and wrong shipments. This study is aimed to detect inaccurate stocks using machine learning methods. It uses the real inven- tory stock data of Migros Ticaret A.S¸. of Turkey’s largest supermarket chains. A multiple of machine learning algorithms such as Isolation Forest (IF), Local Outlier Factor (LOF), One-Class Support Vector Machine (OCSVM) were used to detect abnormal stock values. On the other hand, generally, researchers use public data to develop methods, and it is challenging to apply machine learning algorithms to real-life data, especially in unsupervised learning. This thesis shows how to handle real-life data noises, missing values etc. The experimental findings show the perfor- mances of machine learning methods in detecting anomalies in low and high level inventory stock.
Citation Count: 0
Circular formations of non-communicating robot groups via local strategies
(Sage Publications Ltd, 2023) Kerestecioğlu, Feza; Sen, Uemit; Isikver, Cagri; Goktekin, Ahmet
Local strategies, which are based on cost minimization, to achieve circular formations of autonomous robot groups are presented. It is assumed that the group members have no communication capabilities or any means of interchanging information among themselves, and that they can only rely on their sensors, which provide relative positions of their nearby group members. It is verified on simulations that via appropriately defined cost functions arc, arc-triangle and circle formations are obtained, which can be maintained during navigation.
Citation Count: 0
Fault Tolerant Indoor Positioning Based on Federated Kalman Filter
(Springer, 2024) Kerestecioğlu, Feza; Kerestecioglu, Feza
In this article, multi-sensor indoor positioning, which is based on fusing tri-laterated position data of the target, is considered. A novel method, which is based on federated Kalman filtering and makes use of the fingerprint data, namely, federated Kalman filter with skipped covariance updating (FKF-SCU) is proposed. The data collected on two test beds are used in comparing the performances of the proposed algorithm and that of the regular federated filter. It is shown that the proposed algorithm provides fault tolerance and quick recovery, whenever signal reception from an access point is interrupted, as well as an improvement of 12.57% on the position accuracy.
Citation Count: 6
Indoor Positioning Using Federated Kalman Filter
(IEEE, 2018) Kerestecioğlu, Feza; Kerestecioglu, Feza
In this paper, the performance of a multi-sensor fusion technique, namely Federated Kalman Filter (FKF) is studied in the context of indoor positioning problem. Kalman filters having centralized and decentralized structures are widely used in outdoor positioning and navigation applications. Global Positioning System (GI'S) is the most commonly used system for outdoor positioning/navigation, which cannot be used indoors due to the signal loss. In this study, a decentralized structure for FKF is applied in indoor positioning problem by taking its outdoor navigation performance into consideration. Simulations are perl4med with distance measurements, which are assumed to be calculated by using Received Signal Strength (RSS). Results gathered via different simulations are evaluated as promising for future studies.
Multi-sensor indoor positioning
(Kadir Has Üniversitesi, 2022) AYABAKAN, TARIK; Kerestecioğlu, Feza; Kerestecioglu, Feza
In this study, multi-sensor indoor positioning methods, which fuse the tri-laterated position data of the target are considered. The lateration is based on the dis tances that are obtained using the signal strengths received from different Wi-Fi access points. A new method, which is based on federated Kalman filtering (FKF) and makes use of the fingerprint data, namely, federated Kalman filter with skipped covariance updating (FKF-SCU) is proposed for indoor positioning. After that chal lenging issue of FKF, information sharing coefficient assignment is studied and two online adaptation methods based on received signal strength indication (RSSI) and distance information gathered from APs are proposed. Lastly, FKF-SCU structure is combined with adaptive FKF configuration. The data collected on two different test beds are used to compare the performance of the proposed positioning methods to those of the regular federated and centralized filters. It is shown on the test data that these algorithms improve the position accuracy and provide fault tolerance whenever signal reception is interrupted from an access point.
Citation Count: 4
RSSI-Based Indoor Positioning via Adaptive Federated Kalman Filter
(IEEE-Inst Electrical Electronics Engineers Inc, 2022) Kerestecioğlu, Feza; Kerestecioglu, Feza
In this paper, federated Kalman filter (FKF) is applied for indoor positioning. Position information that is multi-laterated from the distance information obtained using the received signal strengths collected from several access points are processed in a FKF to estimate the position of the target. Two approaches are presented to adjust the information-sharing coefficients of FKF using online measurements. The data collected on a test bed composed of four access points are used to assess and compare the performances of the proposed algorithms. It is shown that the estimation error can be improved considerably by adjusting the information-sharing coefficients online.