Transactions on Machine Intelligence2821-16934320210901Concealment of Speech Signals within Speech by Combining Cryptography and Steganography10911720921510.47176/TMI.2021.109ENJ.ShiraziAssistant Professor, Department of Electrical Engineering, Gonabad Branch, Islamic Azad University, Razavi Khorasan, Gonabad, IranJournal Article20210604The concealment of speech signals, particularly within other speech signals, encompasses the integration of cryptographic techniques and steganographic methods. This literature review synthesizes current research findings and identifies existing gaps in knowledge, while also suggesting potential future research directions. In this paper, to enhance data security in speech transmission, a combination of steganography and cryptography is utilized. In the encryption phase, the Discrete Cosine Transform (DCT) is applied to frames of the speech signal. This phase involves scrambling time-domain samples, transform coefficients, and time-domain samples obtained from the inverse transform of the signal using a chaotic function that generates random numbers. In the steganography phase, the encrypted data replaces the low-value coefficients of the Discrete Wavelet Transform (DWT) of the host speech signal. To evaluate the proposed hybrid method, both qualitative and quantitative metrics were used. The results indicate a high level of various metric measures for the method.https://www.tmachineintelligence.ir/article_209215_d5a4052636c4fa68d93e3bac077abc18.pdfTransactions on Machine Intelligence2821-16934320210901Noise Reduction in CT Scan Images Using Wavelet Transform and Fuzzy Logic11812721865110.47176/TMI.2021.118END.KargaranDepartment of Electrical Engineering, Payam Higher Education Institute, Golpayegan, IranP.MoallemAssistant Professor, Department of Electrical Engineering, Faculty of Engineering, University of Isfahan, Isfahan, IranM.HashemiLecturer, Department of Electrical Engineering, Payam Higher Education Institute, Golpayegan, IranJournal Article20210612Images are often contaminated with noise due to sensor errors or during transmission. Noise in an image reduces the effectiveness of subsequent image processing operations, such as edge detection, segmentation, and object recognition. The use of an appropriate filter to reduce or eliminate noise in medical images shortens imaging time and consequently minimizes the exposure of patients to X-ray radiation. In the medical field, this improvement can enhance a physician’s analysis of captured images, thereby reducing medical errors. In conventional noise removal methods using wavelet coefficient thresholding, the threshold value is uniformly applied to all image coefficients, leading to abrupt changes in values close to the threshold and causing image distortions. To address this issue, fuzzy logic is employed to achieve more effective noise reduction, overcoming the shortcomings of traditional thresholding methods and ensuring that the resulting image is not only quantitatively enhanced but also subjectively satisfactory. Due to its ability to model uncertain and approximate values, fuzzy logic can be effectively utilized for image denoising. By accepting approximation in data processing, fuzzy logic often provides more suitable behavior in many cases. This study aims to introduce an innovative approach for noise removal from CT scan images with the highest precision and sensitivity, considering evaluation criteria such as Peak Signal-to-Noise Ratio (PSNR) and Signal-to-Noise Ratio (SNR), among others. The proposed method integrates wavelet transform and fuzzy logic to establish a novel approach to noise reduction, overcoming existing challenges in this domain through fuzzy logic rules.https://www.tmachineintelligence.ir/article_218651_0d0d0475282db704d1c87e6d7e3d2a6b.pdfTransactions on Machine Intelligence2821-16934320210901Identity Recognition from Ear Images Using Local Binary Pattern and Local Phase Quantization12813621865310.47176/TMI.2021.128ENA.FathiAssistant Professor, Department of Computer Science and Information Technology, Razi University, Kermanshah, IranSh.SajjadiM.Sc. Student in Information Technology Engineering, Department of Computer Science and Information Technology, Razi University, Kermanshah, IranJournal Article20210503One of the challenges posed by technological advancements in modern society is the issue of identity verification and authentication. Among various biometric identification methods, ear biometrics is a relatively new approach. Identity recognition has garnered significant attention in the realm of biometrics, particularly with the advancement of image processing techniques. Among various biometric traits, ear recognition is emerging as a reliable method due to the unique structure of human ears. The human ear possesses unique characteristics that make it suitable for identity recognition. In this study, we employ Local Binary Pattern (LBP) and Local Phase Quantization (LPQ) operators for pattern recognition in ear images. To reduce the feature vector size and enhance classifier accuracy, we apply Principal Component Analysis (PCA) to the features extracted using LBP. Finally, we utilize the k-Nearest Neighbors (k-NN) algorithm with the Canberra similarity measure for classification. To evaluate the efficiency of our proposed method, we conducted experiments on the USTB-1 database, which contains 180 images from 60 individuals, achieving an accuracy of 98.33%.https://www.tmachineintelligence.ir/article_218653_a3dd6041c26f6ccabad08f6dc5fbd622.pdfTransactions on Machine Intelligence2821-16934320210901Feature Selection Method Based on the Rough Set Theory and the Intelligent Water Drops Algorithm13715221865610.47176/TMI.2021.137ENE.DalvandDepartment of Electrical and Computer Engineering, Shahid Beheshti University, Tehran, IranM.BafandkarDepartment of Computer Engineering, Shiraz Branch, Islamic Azad University, Shiraz, IranJournal Article20210605Since datasets are often collected for purposes other than data mining, they typically contain numerous redundant and irrelevant features. The presence of such features can pose challenges for learning systems, including increased computational costs and reduced accuracy. Consequently, feature selection as a preprocessing step can enhance system performance in applications that utilize datasets. This paper presents a method that employs rough set theory as a criterion for evaluating the quality of a feature subset and the intelligent water drops algorithm as a search method. The primary objective of this study is to reduce the computational complexity associated with applying rough set theory in feature selection while improving the quality of the final solution set through the use of the intelligent water drops search algorithm. The proposed method has been tested on multiple datasets from the University of California, Irvine (UCI) repository, and its results have been compared with existing similar methods. The findings demonstrate that eliminating redundant computations can significantly reduce the time complexity of rough set-based feature selection and that the intelligent water drops algorithm serves as an efficient search strategy for feature selection.https://www.tmachineintelligence.ir/article_218656_47eaaec8e1fc41136698efbf7d065d69.pdfTransactions on Machine Intelligence2821-16934320210901Optimized Reactive Power Distribution for Enhancing Power System Efficiency Using a Generalized Teaching-Learning-Based Optimization Algorithm15316821866110.47176/TMI.2021.153ENM.Elahi NejadMSc Student, Power Engineering Department, Faculty of Engineering, Boroujerd Branch, Islamic Azad University, IranA.SamimiAssistant Professor, Department of Electrical Engineering, Arak University of Technology, Arak, IranM.NikzadAssistant Professor, Electrical Engineering Department, Boroujerd Branch, Islamic Azad University, IranJournal Article20210604Electrical energy generation in power systems aims to minimize the total production cost of active units within the power network, making it one of the most critical aspects of modern power systems. Optimal reactive power distribution is a key approach to ensuring the reliable and economical operation of power systems. The primary objective of reactive power distribution in power networks is to determine the control variables that minimize the objective function while adhering to system constraints. In this paper, a hybrid optimization algorithm is introduced, combining the Multi-Objective Teaching-Learning-Based Optimization (MTLBO) method with a fuzzy decision-making algorithm to solve the Reactive Power Dispatch (RPD) problem. The proposed method addresses multiple objectives, including reducing active power losses, improving voltage profiles, and enhancing network security. To assess the effectiveness of the proposed approach, simulations were performed on IEEE 57-bus and 118-bus test systems. The simulation results confirm the efficiency and superiority of the proposed method compared to conventional optimization techniques.https://www.tmachineintelligence.ir/article_218661_efe0cec96bdcb5d3a95caa3c9e3062e7.pdfTransactions on Machine Intelligence2821-16934320210901Enhancement of the Ring-Shaped Photonic Crystal Raman Amplifier Using Optofluidic Materials16918121871210.47176/TMI.2021.169ENA.SeyedfarajiAssociate Professor, Department of Electrical Engineering, Faculty of Engineering, Alzahra University, Tehran, IranJournal Article20210407This study explores the potential of a ring-shaped photonic crystal structure for Raman amplification, focusing on the integration of optofluidic materials to enhance performance. By incorporating optofluidic cavities on both sides of the signal transmission path, the effective refractive index of the medium is modified, leading to a reduction in the group velocity of both the pump and signal waves. This reduction enhances the interaction time between the optical waves, thereby increasing the Raman gain. To further improve performance, a dual-ring structure is introduced and analyzed, comparing its efficiency with that of a single-ring configuration. The impact of structural modifications on the achievable bit rate is also investigated. The numerical analysis is conducted using the Finite-Difference Time-Domain (FDTD) method, solving Maxwell's equations while accounting for various nonlinear effects, including two-photon absorption (TPA), free-carrier absorption (FCA), the Kerr effect, and self-phase modulation (SPM) in a hybrid photonic crystal waveguide. The proposed design, with a compact length of only 100 µm, demonstrates a significant Raman gain of 19.01 dB. Additionally, the system achieves an impressive bit rate of 0.6493 × 10¹² pulses per second, making it a promising candidate for high-speed, high-gain optical signal amplification in next-generation photonic communication networks.https://www.tmachineintelligence.ir/article_218712_d178d6856252d78e6b3a06f6ea930a64.pdf