Özcan Çataltaş

Mr. Özcan Çataltaş

orcid

Research Focus & Metrics

Academic Career Length

8Years (2019-Present)

(Calculated once Scilit publication data is synced: current year − year of first publication + 1)

Representative Publications

View All Publications
9 March 2023 in PeerJ Computer Science
Verified
Journal Article
Open access
PDF file
ÖÇÖzcan Çataltaş
Özcan Çataltaş
KTKemal Tutuncu
Kemal Tutuncu
Background: Analysis of the nutritional values and chemical composition of grain products plays an essential role in determining the quality of the products. Near-infrared spectroscopy has attracted the attention of researchers in recent years due to its advantages in the analysis process. However, preprocessing and regression models in near-infrared spectroscopy are usually determined by trial and error. Combining newly popular deep learning algorithms with near-infrared spectroscopy has brought a new perspective to this area. Methods: This article presents a new method that combines a one-dimensional convolutional autoencoder with near-infrared spectroscopy to analyze the protein, moisture, oil, and starch content of corn kernels. First, a one-dimensional convolutional autoencoder model was created for three different spectra in the corn dataset. Thirty-two latent variables were obtained for each spectrum, which is a low-dimensional spectrum representation. Multiple linear regression models were built for each target using the latent variables of obtained autoencoder models. Results: R2, RMSE, and RMSPE were used to show the performance of the proposed model. The created one-dimensional convolutional autoencoder model achieved a high reconstruction rate with a mean RMSPE value of 1.90% and 2.27% for calibration and prediction sets, respectively. This way, a spectrum with 700 features was converted to only 32 features. The created MLR models which use these features as input were compared to partial least squares regression and principal component regression combined with various preprocessing methods. Experimental results indicate that the proposed method has superior performance, especially in MP5 and MP6 datasets.

1 January 2021 in Computer Science and Information Systems
Verified
Journal Article
Open access
PDF file
KTKemal Tütüncü
Kemal Tütüncü
ÖÇÖzcan Çataltaş
Özcan Çataltaş
This study proposes a new hybrid n-LSB (Least Significant Bit) substitution-based image steganography method in the spatial plane. The previously proposed n-LSB substitution method by authors of this paper is combined with the Rivest-Shamir-Adleman (RSA), RC5, and Data Encryption Standard (DES) encryption algorithms to improve the security of the steganography, which is one of the requirements of steganography, and the Lempel-Ziv-Welch (LZW), Arithmetic and Deflate lossless compression algorithms to increase the secret message capacity. Also, embedding was done randomly using a logistic map-based chaos generator to increase the security more. The classical n-LSB substitution method and the proposed hybrid approaches based on the previously proposed n-LSB were implemented using different secret messages and cover images. When the results were examined, it has been seen that the proposed hybrid n-LSB approach showed improvement in all three criteria of steganography. The proposed hybrid approach that consists of previously proposed n-LSB, RSA, Deflate, and the logistic map had the best results regarding capacity, security, and imperceptibility.

26 May 2025 in New Trends in Computer Sciences
Verified
Journal Article
Open access
PDF file
ÖÇÖzcan Çataltaş
Özcan Çataltaş
In this study, a deep learning-based denoising autoencoder approach is proposed to increase the robustness of near-infrared spectroscopy data to random noise and improve quantitative modeling accuracy. Artificial Gaussian noise at four different levels (10, 15, 20, and 25 dB) was added to the near-infrared spectra obtained from milk samples to mimic the real measurement conditions. The noisy spectra were denoised by processing with an autoencoder architecture consisting of fully connected layers. The noise removal performance is quantitatively evaluated with both theoretical and measured signal-to-noise ratio values. The results show that the AE model significantly improves the spectral signal quality at all signal-to-noise ratio levels. In particular, at the lowest signal-to-noise ratio level (10 dB), the signal-to-noise ratio value nearly tripled to 29.6 dB with the autoencoder. At all other levels, an average increase of 18-20 dB was observed in the signal-to-noise ratio of the denoised spectra. In the second stage of the study, Partial Least Squares Regression models were built using both the noisy and cleaned spectra and evaluated on the test set with root mean square error and coefficient of determination. The Partial Least Squares Regression models built with the denoised spectra achieved lower root mean square error and higher coefficient of determination values at all signal-to-noise ratio levels. Especially at the 10 dB signal-to-noise ratio level, the coefficient of determination value of the model increased from 0.44 to 0.71, while the root means square error decreased from 0.60 to 0.43. The results show that the deep learning-based AE architecture can effectively reduce random noise in near-infrared spectral data and significantly improve both spectral signal quality and quantitative modeling performance. This approach provides an effective solution to improve model reliability and accuracy in near-infrared spectroscopy analysis.

2 December 2025 in PeerJ Computer Science
Verified
Journal Article
Open access
ÖÇÖzcan Çataltaş
Özcan Çataltaş
In the Turkish higher education system, make-up exams are a key mechanism enabling students to recover failed courses, yet the decision to take the make-up exam depends on multiple student- and course-level factors. This study develops an artificial neural network (ANN) model to predict whether students will take the make-up exam, using routinely available academic and course-related features. To obtain a comprehensive sample reflecting both student achievement and course characteristics, the dataset was drawn from courses spanning different academic levels and difficulty tiers. During data preparation, normalization and one-hot encoding were applied to facilitate model learning. Training, validation, and test subsets were formed using a Stratified Sample Projection in X and y (Stratified SPxy) sampling strategy to maintain representativeness across splits, and hyperparameters were optimized via random search to identify the best-performing configuration. The final model achieved 87.68% test accuracy and an 87.62% test F1-score, indicating good generalization. Notably, to the best of our knowledge, this is the first study to address make-up exam participation using only easily accessible institutional data, which makes the proposed approach practical and adaptable for real-world university settings. Despite relying on simple and readily available parameters, the ANN delivers strong predictive performance, suggesting that efficient models can be developed for non-complex datasets. Such predictions can enable early identification of at-risk students and trigger targeted academic support, inform course- and program-level adjustments, and improve operational planning, thereby contributing to more effective educational management and improved student progression and completion rates.

25 December 2022 in Selcuk Journal of Agriculture and Food Sciences
Verified
Journal Article
Open access
PDF file
ÖÇÖzcan Çataltaş
Özcan Çataltaş
KTKemal Tutuncu
Kemal Tutuncu
Sugarcane is a plant whose quality parameters are required to be determined both for being one of the substances used in sugar production and for being used as animal feed. Near-infrared spectroscopy is a technique that has already been used for predicting the parameters of various plants and has gained popularity in recent years. This study proposes a near-infrared spectroscopy-based model for the rapid and effortless analysis of acid detergent fiber fraction and vitro organic matter digestibility parameters of the sugarcane plant. Partial least squares regression was combined with common preprocessing methods for modeling. This model yielded an R2CV value of 0.935 and 0.953 for the acid detergent fiber fraction and vitro organic matter digestibility parameters, respectively. Then, the spectra from three handheld spectrometers were combined using a proposed combination method to generate new spectra with higher spectral resolution. New models were built using these generated spectra and compared to the previous result. Obtained results showed that combining spectra from different spectrometers can improve model performance.

1 November 2019 in IOP Conference Series: Materials Science and Engineering
Verified
Conference Paper
Open access
PDF file
ÖÇÖzcan Çataltaş
Özcan Çataltaş
KTKemal Tutuncu
Kemal Tutuncu
Today, secure communication is one of the biggest goals that people want to achieve. There are many different methods to ensure security in data communication. Steganography is one of the methods that can be used in secure data communication. In this study, n-LSB embedding method based on reducing the amount of distortion generated by the classical Least Significant Bit substitution method that we introduced in another study has been applied in both RGB color space and HSI, XYZ and YCbCr color spaces. In the embedding process, 3 message bits were embedded in each color channel. 7.94 kb data is embedded on 3 different cover images. The obtained results were compared with the PSNR criterion, which is the criterion of image quality evaluation, and it is seen that proposed n-LSB approach gives better results in HSI color space for 3 cover images. The highest PSNR value obtained was 83.907 dB.