Nagornov, Nikolay Nikolaevich

Abdulkadirov R., Lyakhov P., Butusov D., Nagornov N., Kalita D.

In this paper, we propose a physics-informed neural network controller for quadcopter dynamics modeling. Physics-aware machine learning methods, such as physics-informed neural networks, consider the UAV dynamics model, solving the system of ordinary differential equations entirely, unlike proportional–integral–derivative controllers. The more accurate control action on the quadcopter reduces flight time and power consumption. We applied our fractional optimization algorithms to decreasing the solution error of quadcopter dynamics. Including advanced optimizers in the reinforcement learning model, we achieved the trajectory of UAV flight more accurately than state-of-the-art proportional–integral–derivative controllers. The advanced optimizers allowed the proposed controller to increase the quality of the building trajectory of the UAV compared to the state-of-the-art approach by 10 percentage points. Our model had less error value in spatial coordinates and Euler angles by 25–35% and 30–44%, respectively.

Abdulkadirov R., Lyakhov P., Butusov D., Nagornov N., Reznikov D., Bobrov A., Kalita D.

The current development of machine learning has advanced many fields in applied sciences and industry, including remote sensing. In this area, deep neural networks are used to solve routine object detection problems, satisfying the required rules and conditions. However, the growing number and difficulty of such problems cause the developers to construct machine learning models with higher computational complexities, such as an increased number of hidden layers, epochs, learning rate, and rate decay. In this paper, we propose the Yolov8 architecture with decomposed layers via canonical polyadic and Tucker methods for accelerating the solving of the object detection problem in satellite images. Our positive–negative momentum approaches enabled a reduction in the loss in precision and recall assessments for the proposed neural network. The convolutional layer factorization reduces the shapes and accelerates the computations at kernel nodes in the proposed deep learning models. The advanced optimization algorithms achieve the global minimum of loss functions, which makes the precision and recall metrics superior to the ones for their known counterparts. We examined the proposed Yolov8 with decomposed layers, comparing it with the conventional Yolov8 on the DIOR and VisDrone 2020 datasets containing the UAV images. We verified the performance of the proposed and known neural networks on different optimizers. It is shown that the proposed neural network accelerates the solving object detection problem by 44–52%. The proposed Yolov8 with Tucker and canonical polyadic decompositions has greater precision and recall metrics than the usual Yolov8 with known analogs by 0.84–0.94 and 0.228–1.107 percentage points, respectively.

Abdulkadirov R.I., Lyakhov P.A., Baboshina V.A., Nagornov N.N.

Abdulsalyamova A., Abdulkadirov R., Lyakhov P., Nagornov N.

The main problem of artificial intelligence is increasing productivity and quality of problem solutions. Due to the growing architecture of modern neural networks, one needs to engage advanced mathematical methods. Deep learning models use more hardware resources, which increases the computational complexity. Therefore, it is necessary to apply modifications of machine learning models at a fundamental level using alternative matrix multiplication methods. This article proposes a comparative analysis of the computational complexity of matrix multiplication implemented by the standard Strassen and Strassen-Winograd methods. We consider data time complexity for int32, int64, float32, and float64 data types. In addition, the number of recursions for each matrix size is determined. According to the experimental results, we can conclude that the Strassen-Winograd matrix multiplication method has minimal time costs compared to the Strassen method and standard approaches by 3%–6% and 30%–40%, respectively. It is possible to incorporate such an approach into convolutional, spike, and auto-encoding layers.

Nagornov N.N., Lyakhov P.A., Bergerman M.V., Kalita D.I.

Kalita D.I., Lyakhov P.A., Nagornov N.N.

Lyakhov P.A., Nagornov N.N., Semyenova N.F., Abdulsalyamova A.S.

Abdulkadirov R.I., Lyakhov P.A., Nagornov N.N.

Kiladze M.R., Lyakhova U.A., Lyakhov P.A., Nagornov N.N., Vahabi M.

Lyakhov P., Semyonova N., Nagornov N., Bergerman M., Abdulsalyamova A.

Wavelets are actively used to solve a wide range of image processing problems in various fields of science and technology. Modern image processing systems cannot keep up with the rapid growth in digital visual information. Various approaches are used to reduce the computational complexity and increase computational speeds. The Winograd method (WM) is one of the most promising. However, this method is used to obtain sequential values. Its use for wavelet image processing requires expanding the calculation methodology to cases of downsampling. This paper proposes a new approach to reduce the computational complexity of wavelet image processing based on the WM with decimation. Calculations have been carried out and formulas have been derived that implement digital filtering using the WM with downsampling. The derived formulas can be used for 1D filtering with an arbitrary downsampling stride. Hardware modeling of wavelet image filtering on an FPGA showed that the WM reduces the computational time by up to 66%, with increases in the hardware costs and power consumption of 95% and 344%, respectively, compared to the direct method. A promising direction for further research is the implementation of the developed approach on ASIC and the use of modular computing for more efficient parallelization of calculations and an even greater increase in the device speed.

Lyakhov P.A., Bergerman M.V., Abdulkadirov R.I., Abdulsalyamova A.S., Nagornov N.N., Voznesensky A.S., Minenkov D.V., Kaplun D.I.

Sign detection is a non-modular operation in the residue number system (RNS). It requires the calculation of the number positional characteristic represented in the RNS. This work proposes a new sign detection method based on the Chinese Remainder Theorem (CRT) with fractional values implemented using the Wallace tree and the modified Kogge-Stone adder. Hardware modelling on FPGA for the proposed method shows that it provides 1.3 - 36.3 times less hardware costs than the other state-of-the-art (SOTA) methods, and for ASIC modelling the proposed method provides 1.14 - 35.74 times less hardware costs than the other SOTA methods. The presented sign detection method can be helpful in RNS-based devices in implementing comparison and division operations, providing an extension of the RNS application in areas such as cryptography, machine learning, and digital signal processing.

Abdulsalyamova A.S., Kalita D.I., Lyakhov P.A., Nagornov N.N., Bergerman M.V.

Lyakhova U.A., Bondarenko D.N., Boyarskaya E.E., Nagornov N.N.

Skin cancer is the most common cancer in humans today and is usually caused by exposure to ultraviolet radiation. There are many diagnostic methods for visual analysis of pigmented neoplasms. However, most of these methods are subjective and largely dependent on the experience of the clinician. To minimize the influence of the human factor, it is proposed to introduce artificial intelligence technologies that have made it possible to reach new heights in terms of the accuracy of classifying medical data, including in the field of dermatology. Artificial intelligence technologies can equal and even surpass the capabilities of an dermatologists in terms of the accuracy of visual diagnostics. The article proposes a web application based on a multimodal neural network system for recognizing pigmented skin lesions as an additional auxiliary tool for oncologist. The system combines and analyzes heterogeneous dermatological data, which are images of pigmented neoplasms and such statistical information about the patient as age, gender, and localization of pigmented skin lesions. The recognition accuracy of the proposed web application was 85.65%. The use of the proposed web application as an auxiliary diagnostic method will expand the possibilities of early detection of skin cancer and minimize the impact of the human factor.

Nagornov N.N., Semyonova N.F., Abdulsalyamova A.S.

Wavelets are actively used for solving of image processing problems in various fields of science and technology. Modern imaging systems have not kept pace with the rapid growth in the amount of digital visual information that needs to be processed, stored, and transmitted. Many approaches are being developed and used to speed up computations in the implementation of various image processing methods. This paper proposes the Winograd method (WM) to speed up the wavelet image processing methods on modern microelectronic devices. The scheme for wavelet image filtering using WM has been developed. WM application reduced the computational complexity of wavelet filtering asymptotically to 72.9% compared to the direct implementation. An evaluation based on the unit-gate model showed that WM reduces the device delay to 66.9%, 73.6%, and 68.8% for 4-, 6-, and 8-tap wavelets, respectively. Revealed that the larger the processed image fragments size, the less time is spent on wavelet filtering, but the larger the transformation matrices size, the more difficult their compilation and WM design on modern microelectronic devices. The obtained results can be used to improve the performance of wavelet image processing devices for image compression and denoising. WM hardware implementation on a field-programmable gate arrays and an application-specific integrated circuits to accelerate wavelet image processing is a promising direction for further research.

Lyakhov P.A., Nagornov N.N., Semyonova N.F., Abdulsalyamova A.S.

Modern computer technology devices do not keep pace with the high growth rate of quantitative and qualitative characteristics of digital images. The computational complexity of the wavelet transform must be reduced for the hardware-friendly implementation of wavelet image processing methods on microelectronic devices. This paper proposes a new approach to reduce the computational complexity of wavelet image processing based on the Winograd method. Group pixel processing using Winograd method reduces the asymptotic computational complexity by up to 72.9% compared to the traditional pixel-by-pixel processing approach, according to the results obtained. A theoretical evaluation of the resource costs of a wavelet image processing device based on the unit-gate model showed that Winograd method reduces device delay to 73.62% and device area to 34.03% compared to the direct implementation. The greatest reduction in resource costs is observed mainly when obtaining fragments of the processed image with 5 pixels. At the same time, the greatest rate of resource reduction is observed when obtaining fragments of the processed image with 3 pixels. Further increase in the fragments size leads to a significantly smaller reduction in resource costs while increasing the complexity of circuits design. Separation of filters into several components is more hardware-friendly when using high-order wavelets. Verification of all obtained results on field-programmable gate arrays and application-specific integrated circuits is a promising direction for further research.

Kukrál M., Pham D.T., Kohout J., Kohek Š., Havlík M., Grygarová D.

Laptev N.V., Laptev V.V., Pleshkov M.O., Martusevich Y.A., Starikova E.G., Tolmachev I.V.

Ramesh E., Ganesan A., Lakshmi K.C., Natarajan P.M.

ObjectiveThe present study aims to employ and compare the artificial intelligence (AI) convolutional neural networks (CNN) Xception and MobileNet-v2 for the diagnosis of Oral leukoplakia (OL) and to differentiate its clinical types from other white lesions of the oral cavity.Materials and methodsClinical photographs of oral leukoplakia and non-oral leukoplakia lesions were gathered from the SRM Dental College archives. An aggregate of 659 clinical photos, based on convenience sampling were included from the archive in the dataset. Around 202 pictures were of oral leukoplakia while 457 were other white lesions. Lesions considered in the differential diagnosis of oral leukoplakia like frictional keratosis, oral candidiasis, oral lichen planus, lichenoid reactions, mucosal burns, pouch keratosis, and oral carcinoma were included under the other white lesions subset. A total of 261 images constituting the test sample, were arbitrarily selected from the collected dataset, whilst the remaining images served as training and validation datasets. The training dataset were engaged in data augmentation to enhance the quantity and variation. Performance metrics of accuracy, precision, recall, and f1_score were incorporated for the CNN model.ResultsCNN models both Xception and MobileNetV2 were able to diagnose OL and other white lesions using photographs. In terms of F1-score and overall accuracy, the MobilenetV2 model performed noticeably better than the other model.ConclusionWe demonstrate that CNN models are capable of 89%–92% accuracy and can be best used to discern OL and its clinical types from other white lesions of the oral cavity.

Cassiano L.D., da Silva J.P., Martins A.A., Barbosa M.T., Rodrigues K.T., Barbosa Á.R., da Silva Gomes G.E., Maia P.R., de Oliveira P.T., de Sousa Lopes M.L., da Silva I.M., de Aquino Martins A.R.

Kim K., Kim K., Shin Y.

This study presents a novel localization framework that leverages the unique properties of chirp signals combined with a time division multiple access (TDMA)-based tactical data link to achieve high-precision positioning. Chirp signals, with their wide bandwidth and high temporal resolution, enable an oversampling-like effect, significantly enhancing distance estimation accuracy without the need for additional sampling rates. The proposed framework integrates chirp-based ranging and localization algorithms, incorporating raised cosine interpolation and circular shift techniques to improve temporal resolution and ensure precise peak detection. By utilizing the time differential of arrival (TDoA) and Fang’s algorithm, the system demonstrates robust performance, effectively mitigating challenges posed by multipath interference and jamming. The TDMA system provides synchronized time slots, allowing the seamless integration of communication and localization functionalities while ensuring stable and efficient operation. Experimental evaluations under various environmental conditions, including dense multipath and high-jamming scenarios, confirm the framework’s superiority over conventional localization methods in terms of accuracy, reliability, and resilience. These results highlight the framework’s potential applications in diverse fields, such as Internet of Things (IoT) networks, smart city infrastructure, and tactical communication systems, where high precision and robust localization are critical.

Dai Y., Zhang W., Li Q., Wu Z., Liu Y.

Liu Y., Jiang S., Wang Y.

BackgroundStandardised management of chronic sinusitis (CRS) is a challenging but vital area of research. Not only is accurate diagnosis and individualised treatment plans required, but post-treatment chronic disease management is also indispensable. With the development of artificial intelligence (AI), more “AI + medical” application models are emerging. Many AI-assisted systems have been applied to the diagnosis and treatment of CRS, providing valuable solutions for clinical practice.ObjectiveThis study summarises the research progress of various AI-assisted systems applied to the clinical diagnosis and treatment of CRS, focusing on their role in imaging and pathological diagnosis and prognostic prediction and treatment.MethodsWe used PubMed, Web of Science, and other Internet search engines with “artificial intelligence”、“machine learning” and “chronic sinusitis” as the keywords to conduct a literature search for studies from the last 7 years. We included literature eligible for AI application to CRS diagnosis and treatment in our study, excluded literature outside this scope, and categorized it according to its clinical application to CRS diagnosis, treatment, and prognosis prediction. We provide an overview and summary of current advances in AI to optimize the diagnosis and treatment of CRS, as well as difficulties and challenges in promoting standardization of clinical diagnosis and treatment in this area.ResultsThrough applications in CRS imaging and pathology diagnosis, personalised medicine and prognosis prediction, AI can significantly reduce turnaround times, lower diagnostic costs and accurately predict disease outcomes. However, a number of challenges remain. These include a lack of AI product standards, standardised data, difficulties in collaboration between different healthcare providers, and the non-interpretability of AI systems. There may also be data privacy issues involved. Therefore, more research and improvements are needed to realise the full potential of AI in the diagnosis and treatment of CRS.ConclusionOur findings inform the clinical diagnosis and treatment of CRS and the development of AI-assisted clinical diagnosis and treatment systems. We provide recommendations for AI to drive standardisation of CRS diagnosis and treatment.

Suleimenov I., Bakirov A.

The method of non-standard algebraic extensions based on the use of additional formal solutions of the reduced equations is extended to the case corresponding to three-dimensional space. This method differs from the classical one in that it leads to the formation of algebraic rings rather than fields. The proposed approach allows one to construct a discrete coordinate system in which the role of three basis vectors is played by idempotent elements of the ring obtained by a non-standard algebraic extension. This approach allows, among other things, the identification of the symmetry properties of objects defined through discrete Cartesian coordinates, which is important, for example, when using advanced methods of digital image processing. An explicit form of solutions of the equations is established that allow one to construct idempotent elements for Galois fields GFp such that p−1 is divisible by three. The possibilities of practical use of the proposed approach are considered; in particular, it is shown that the use of discrete Cartesian coordinates mapped onto algebraic rings is of interest from the point of view of improving UAV swarm control algorithms.

Hussein M.K., Ucan O.N.

Abdulkadirov R., Lyakhov P., Butusov D., Nagornov N., Kalita D.

In this paper, we propose a physics-informed neural network controller for quadcopter dynamics modeling. Physics-aware machine learning methods, such as physics-informed neural networks, consider the UAV dynamics model, solving the system of ordinary differential equations entirely, unlike proportional–integral–derivative controllers. The more accurate control action on the quadcopter reduces flight time and power consumption. We applied our fractional optimization algorithms to decreasing the solution error of quadcopter dynamics. Including advanced optimizers in the reinforcement learning model, we achieved the trajectory of UAV flight more accurately than state-of-the-art proportional–integral–derivative controllers. The advanced optimizers allowed the proposed controller to increase the quality of the building trajectory of the UAV compared to the state-of-the-art approach by 10 percentage points. Our model had less error value in spatial coordinates and Euler angles by 25–35% and 30–44%, respectively.

Abdulkadirov R., Lyakhov P., Butusov D., Nagornov N., Reznikov D., Bobrov A., Kalita D.

The current development of machine learning has advanced many fields in applied sciences and industry, including remote sensing. In this area, deep neural networks are used to solve routine object detection problems, satisfying the required rules and conditions. However, the growing number and difficulty of such problems cause the developers to construct machine learning models with higher computational complexities, such as an increased number of hidden layers, epochs, learning rate, and rate decay. In this paper, we propose the Yolov8 architecture with decomposed layers via canonical polyadic and Tucker methods for accelerating the solving of the object detection problem in satellite images. Our positive–negative momentum approaches enabled a reduction in the loss in precision and recall assessments for the proposed neural network. The convolutional layer factorization reduces the shapes and accelerates the computations at kernel nodes in the proposed deep learning models. The advanced optimization algorithms achieve the global minimum of loss functions, which makes the precision and recall metrics superior to the ones for their known counterparts. We examined the proposed Yolov8 with decomposed layers, comparing it with the conventional Yolov8 on the DIOR and VisDrone 2020 datasets containing the UAV images. We verified the performance of the proposed and known neural networks on different optimizers. It is shown that the proposed neural network accelerates the solving object detection problem by 44–52%. The proposed Yolov8 with Tucker and canonical polyadic decompositions has greater precision and recall metrics than the usual Yolov8 with known analogs by 0.84–0.94 and 0.228–1.107 percentage points, respectively.

Abraham O.L., Ngadi M.A.

Awodeyi A.I., Ibok O.A., Omokaro I., Ekwemuka J.U., Ighofiomoni M.O.

Amin A.A., Mubarak A., Waseem S.

Burhan K.

Somanagoudar A.G., Mérida W.

Ramirez I., Pino J., Pardo D., Sanz M., del Rio L., Ortiz A., Morozovska K., Aizpurua J.I.

Cheng H., Zhang M., Shi J.Q.

Alhawsawi A.N., Khan S.D., Rehman F.U.

Crowd counting in aerial images presents unique challenges due to varying altitudes, angles, and cluttered backgrounds. Additionally, the small size of targets, often occupying only a few pixels in high-resolution images, further complicates the problem. Current crowd counting models struggle in these complex scenarios, leading to inaccurate counts, which are crucial for crowd management. Moreover, these regression-based models only provide the total count without indicating the location or distribution of people within the environment, limiting their practical utility. While YOLOv8 has achieved significant success in detecting small targets within aerial imagery, it faces challenges when directly applied to crowd counting tasks in such contexts. To overcome these challenges, we propose an improved framework based on YOLOv8, incorporating a context enrichment module (CEM) to capture multiscale contextual information. This enhancement improves the model’s ability to detect and localize tiny targets in complex aerial images. We assess the effectiveness of the proposed framework on the challenging VisDrone-CC2021 dataset, and our experimental results demonstrate the effectiveness of this approach.

Abdulkadirov R.I., Lyakhov P.A., Baboshina V.A., Nagornov N.N.

Zhang S., Li Z., Liu W., Zhao J., Qin T.

Sharifuzzaman S.A., Tanveer J., Chen Y., Chan J.H., Kim H.S., Kallu K.D., Ahmed S.

Remote sensing technology has been modernized by artificial intelligence, which has made it possible for deep learning algorithms to extract useful information from images. However, overfitting and lack of uncertainty quantification, high-resolution images, information loss in traditional feature extraction, and background information retrieval for detected objects limit the use of deep learning models in various remote sensing applications. This paper proposes a Bayes by backpropagation (BBB)-based system for scene-driven identification and information retrieval in order to overcome the above-mentioned problems. We present the Bayes R-CNN, a two-stage object detection technique to reduce overfitting while also quantifying uncertainty for each object recognized within a given image. To extract features more successfully, we replace the traditional feature extraction model with our novel Multi-Resolution Extraction Network (MRENet) model. We propose the multi-level feature fusion module (MLFFM) in the inner lateral connection and a Bayesian Distributed Lightweight Attention Module (BDLAM) to reduce information loss in the feature pyramid network (FPN). In addition, our system incorporates a Bayesian image super-resolution model which enhances the quality of the image to improve the prediction accuracy of the Bayes R-CNN. Notably, MRENet is used to classify the background of the detected objects to provide detailed interpretation of the object. Our proposed system is comprehensively trained and assessed utilizing the state-of-the-art DIOR and HRSC2016 datasets. The results demonstrate our system’s ability to detect and retrieve information from remote sensing scene images.

Sharifuzzaman S.A., Chen Y., Xie Y., Kim H.S.

Remote sensing scene understanding is crucial for extracting valuable information from high-resolution images, including object detection and classification. Traditional object detection methods face challenges in handling the diverse scales, orientations, and complex backgrounds present in remote sensing data. In this paper, we propose a novel remote sensing scene understanding system called multiscale-attention R-CNN (MSA R-CNN), which incorporates a super multiscale feature extraction network (SMENet) for enhanced feature extraction from multiscale images, an adaptive dynamic inner lateral (ADIL) connection module to tackle information loss in feature pyramid networks (FPN), and a distributed lightweight attention module (DLAM) to refine feature information processing. Furthermore, a new dataset combining the DIOR and DOTA datasets is introduced to extract the background information of detected objects and evaluate the proposed system’s performance. MSA R-CNN achieved an mAP of 74.37% on the DIOR dataset when the gamma value was set to 0.2 and 81.97% on the DOTA dataset when the gamma value was set to 0.1 with the same learning rate, outperforming state-of-the-art models on both datasets. The proposed system demonstrates significant improvements in both object detection and background information extraction, providing a comprehensive solution for remote sensing scene understanding.

Jia X., Feng X., Yong H., Meng D.

Weight decay (WD) is a fundamental and practical regularization technique in improving generalization of current deep learning models. However, it is observed that the WD does not work effectively for an adaptive optimization algorithm (such as Adam), as it works for SGD. Specifically, the solution found by Adam with the WD often generalizes unsatisfactorily. Though efforts have been made to mitigate this issue, the reason for such deficiency is still vague. In this article, we first show that when using the Adam optimizer, the weight norm increases very fast along with the training procedure, which is in contrast to SGD where the weight norm increases relatively slower and tends to converge. The fast increase of weight norm is adverse to WD; in consequence, the Adam optimizer will lose efficacy in finding solution that generalizes well. To resolve this problem, we propose to tailor Adam by introducing a regularization term on the adaptive learning rate, such that it is friendly to WD. Meanwhile, we introduce first moment on the WD to further enhance the regularization effect. We show that the proposed method is able to find solution with small norm and generalizes better than SGD. We test the proposed method on general image classification and fine-grained image classification tasks with different networks. Experimental results on all these cases substantiate the effectiveness of the proposed method in help improving the generalization. Specifically, the proposed method improves the test accuracy of Adam by a large margin and even improves the performance of SGD by $0.84\%$ on CIFAR 10 and $1.03 \%$ on CIFAR 100 with ResNet-50. The code of this article is public available at xxx.

Chang X., Liu X., Hou L., Qi J.

Sharma P., Saurav S., Singh S.

Lack of proper maintenance of power line infrastructures is one of the main reasons behind power shortages and major blackouts. Current inspection methods are human-dependent, which is time-consuming and expensive. Recent progress in Unmanned Aerial Vehicles (UAVs) and digital cameras enforces the use of UAVs for power line inspection, reducing the cost and time to a great extent. Deep learning methods have recently proved their efficacy in the automatic analysis of power line data; however, they suffer from numerous challenges. Unlike generic object detection, power line inspection does not have large datasets. The data collection of power line objects is challenging compared to data collection for generic objects. As deep learning methods are data-hungry, difficulty in collecting training data raises class imbalance problems . Also, the real-time inspection of power line components demands compute-efficient deep learning methods, which is also challenging because of the high computational requirements of the generic deep learning-based object detectors. Despite being researched for decades, no object detectors can eliminate the effect of diverse challenges on the performance of deep learning methods. With these considerations, this study thoroughly reviews the existing works in the literature and the methods and approaches adopted in power line inspection to overcome these challenges. We also provide the type of faults addressed in the literature with details on the methods employed for their analysis. Finally, we conclude the review by providing insights into future research directions in power line inspection. • In this survey, various power line components, their faults and consequences, and sensors used to capture these faults are described. • The challenges of vision-based methods for automatic power line inspection are identified. • The existing open-source power line datasets are discussed and the reported results on these datasets are summarized. • The available techniques to solve the challenges of power line infrastructure are examined and valuable insights are provided for further studies.

Wang H., Jiang H., Sun J., Zhang S., Chen C., Hua X., Luo X.

Guan Y., Zou S., Peng H., Ni W., Sun Y., Gao H.

Abdulkadirov R., Lyakhov P., Bergerman M., Reznikov D.

The modern machine learning theory finds application in many areas of human activity. One of the most dispersed tasks is pattern recognition on satellite images. It is difficult for a person to recognize a large number of images in a short time. It made the researchers develop the automation process, such as neural network engagement. The loss function minimization and ensemble learning raise the pattern recognition accuracy. We propose the robust difference gradient positive-negative momentum optimization algorithm that achieves the global minimum of the loss function with higher accuracy and fewer iterations than known analogs. Such an optimization algorithm contains the generalized average moving estimation approach and more effective learning rate control by additional parameters. The proposed optimizer has the regret-bound rate estimation, belonging to OT, and converges to the global minimum. However, the main problems in optimization theory are vanishing and blowing gradient values, where the standard gradient-based algorithms fail to achieve the required objective function value. The vanishing and blowing gradient problems meet in Rastrigin and Rosebrock test functions, where the proposed optimization algorithm attains the global extreme in the shortest number of iterations and has a more stable convergence process than state-of-the-art methods. Afterward, there are trained deep convolutional neural networks with different optimizers on satellite images from the University of California merced dataset containing 21 object classes, where the proposed algorithm gives the highest accuracy. There is a suggested ensemble-learning model consisting of 4 networks with different optimizers. The prediction results receive weight coefficients distributed according to the majority voting and ensemble neural network retrains with the higher pattern recognition accuracy. The suggested ensemble-learning model with the developed optimizer raised the accuracy by 1 %–4 % percentage points.

Sowmya R., Premkumar M., Jangir P.

The Newton-Raphson-Based Optimizer (NRBO), a new metaheuristic algorithm, is suggested and developed in this paper. The NRBO is inspired by Newton-Raphson's approach, and it explores the entire search process using two rules: the Newton-Raphson Search Rule (NRSR) and the Trap Avoidance Operator (TAO) and a few groups of matrices to explore the best results further. The NRSR uses a Newton-Raphson method to improve the exploration ability of NRBO and increase the convergence rate to reach improved search space positions. The TAO helps the NRBO to avoid the local optima trap. The performance of NRBO was assessed using 64 benchmark numerical functions, including 23 standard benchmarks, 29 CEC2017 constrained benchmarks, and 12 CEC2022 benchmarks. In addition, the NRBO was employed to optimize 12 CEC2020 real-world constrained engineering optimization problems. The proposed NRBO was compared to seven state-of-the-art optimization algorithms, and the findings showed that the NRBO produced promising results due to its features, such as high exploration and exploitation balance, high convergence rate, and effective avoidance of local optima capabilities. In addition, the NRBO also validated on challenging wireless communication problem called the internet of vehicle problem, and the NRBO was able to find the optimal path for data transmission. Also, the performance of NRBO in training the deep reinforcement learning agents is also studied by considering the mountain car problem. The obtained results also proved the NRBO's excellent performance in handling challenging real-world engineering problems.