Browsing by Subject "Synthetic data"
Now showing 1 - 4 of 4
Results Per Page
Sort Options
Item Open Access Particle swarm optimization for SAGE maximization step in channel parameter estimation(IET, 2007-11) Bodur, Harun; Tunç, Celal Alp; Aktaş, Defne; Ertürk, Vakur .B.; Altıntaş, AyhanThis paper presents an application of particle swarm optimization (PSO) in space alternating generalized expectation maximization (SAGE) algorithm. SAGE algorithm is a powerful tool for estimating channel parameters like delay, angles (azimuth and elevation) of arrival and departure, Doppler frequency and polarization. To demonstrate the improvement in processing time by utilizing PSO in SAGE algorithm, the channel parameters are estimated from a synthetic data and the computational expense of SAGE algorithm with PSO is discussed. (4 pages).Item Open Access Synthetic TEC mapping with kriging and random field priors(IEEE, 2007) Sayın, I.; Arıkan, F.; Arıkan, OrhanTotal Electron Content (TEC) can be used for analyzing the variability of the ionosphere in space and time. In this study, spatial interpolation is implemented by Kriging and Random Field Priors (RFP), which are widely used in geostatistics. Performance of Kriging and RFP methods are analyzed on synthetic TEC data for different trend functions, sampling patterns, sampling numbers, variance and range values of covariance function which is used to simulate the synthetic data, by comparing the normalized errors of interpolations. In regular sampling patterns, as opposed to random sampling, the normalized average error is very close to each other for all methods and trend assumptions. The error increases with variance and decreases with range. As the number of samples increase, the normalized error also decreases.Item Open Access Synthetic TEC mapping with ordinary and universal kriging(IEEE, 2007-06) Sayın, I.; Arıkan, F.; Arıkan, OrhanSpatiotemporal variations in the ionosphere affects the HF and satellite communications and navigation systems. Total Electron Content (TEC) is an important parameter since it can be used to analyze the spatial and temporal variability of the ionosphere. In this study, the performance of the two widely used Kriging algorithms, namely Ordinary Kriging (OrK) and Universal Kriging (UnK), is compared over the synthetic data set. In order to represent various ionospheric states, such as quiet and disturbed days, spatially correlated residual synthetic TEC data with different variances is generated and added to trend functions. Synthetic data sampled with various type of sampling patterns and for a wide range of sampling point numbers. It is observed that for small sampling numbers and with higher variability, OrK gives smaller errors. As the sample number increases, UnK errors decrease faster. For smaller variances in the synthetic surfaces, UnK gives better results. For increasing variance and decreasing range values, usually, the errors increase for both OrK and UnK. © 2007 IEEE.Item Open Access Synthetic18K: Learning better representations for person re-ID and attribute recognition from 1.4 million synthetic images(Elsevier, 2021-05-26) Aslan, C.; Ercan, B.; Ates, T.; Celikcan, U.; Erdem, A.; Erdem, E.; Üner, Onur CanLearning robust representations is critical for the success of person re-identification and attribute recognition systems. However, to achieve this, we must use a large dataset of diverse person images as well as annotations of identity labels and/or a set of different attributes. Apart from the obvious concerns about privacy issues, the manual annotation process is both time consuming and too costly. In this paper, we instead propose to use synthetic person images for addressing these difficulties. Specifically, we first introduce Synthetic18K, a large-scale dataset of over 1 million computer generated person images of 18K unique identities with relevant attributes. Moreover, we demonstrate that pretraining of simple deep architectures on Synthetic18K for person re-identification and attribute recognition and then fine-tuning on real data leads to significant improvements in prediction performances, giving results better than or comparable to state-of-the-art models.