Hasil untuk "Computer engineering. Computer hardware"

M. Abdelguerfi, Kam-Fai Wong

Zhong Sun, G. Pedretti, E. Ambrosi et al.

Significance Linear algebra is involved in virtually all scientific and engineering disciplines, e.g., physics, statistics, machine learning, and signal processing. Solving matrix equations such as a linear system or an eigenvector equation is accomplished by matrix factorizations or iterative matrix multiplications in conventional computers, which is computationally expensive. In-memory computing with analog resistive memories has shown high efficiencies of time and energy, through realizing matrix-vector multiplication in one step with Ohm’s law and Kirchhoff’s law. However, solving matrix equations in a single operation remains an open challenge. Here, we show that a feedback circuit with cross-point resistive memories can solve algebraic problems such as systems of linear equations, matrix eigenvectors, and differential equations in just one step. Conventional digital computers can execute advanced operations by a sequence of elementary Boolean functions of 2 or more bits. As a result, complicated tasks such as solving a linear system or solving a differential equation require a large number of computing steps and an extensive use of memory units to store individual bits. To accelerate the execution of such advanced tasks, in-memory computing with resistive memories provides a promising avenue, thanks to analog data storage and physical computation in the memory. Here, we show that a cross-point array of resistive memory devices can directly solve a system of linear equations, or find the matrix eigenvectors. These operations are completed in just one single step, thanks to the physical computing with Ohm’s and Kirchhoff’s laws, and thanks to the negative feedback connection in the cross-point circuit. Algebraic problems are demonstrated in hardware and applied to classical computing tasks, such as ranking webpages and solving the Schrödinger equation in one step.

R. L. Manogna, Vijay Dharmaji, S. Sarang

Abstract This study presents a comprehensive analysis of agricultural price volatility forecasting using a hybrid long short-term memory (LSTM)-Generalized Autoregressive Conditional Heteroskedasticity (GARCH) model. Agricultural price volatility poses critical challenges for food security, economic stability, and the livelihoods of millions, particularly in developing countries like India. Accurately forecasting these price fluctuations is vital for effective policymaking and strategic decision-making in agricultural markets. This study investigates the potential of deep learning models, specifically LSTM, and their integration with GARCH for forecasting agricultural commodity price volatility. Using extensive historical price data for 23 commodities across 165 markets in India from February 2010 to June 2024, the proposed hybrid model demonstrates significantly enhanced accuracy and robustness compared to standalone econometric or deep learning models. The results suggest that this hybrid approach effectively addresses price instability, offering improved predictive capabilities. These findings provide valuable implications for policymakers and stakeholders, emphasizing the adoption of advanced machine learning techniques for better market risk management and policy interventions tailored to agricultural price dynamics.

Liia Buhhanevits, Hesam Ramezani, Charles de Kergariou et al.

Hygromorphic materials and composites adjust their shape and curvature in response to changes in relative humidity, similar to the pinecone scales existing in nature. This work introduces a new class of poligromorphic materials‐composites that adapt across various environments, including water, humidity, and hydrocarbons. Composed of carbon fibers (CF), polyvinyl alcohol (PVA), and maleic anhydride ethylene‐propylene (MA‐EPR) matrices, these materials are sensitive to fluctuations in fluids like water, isooctane, and toluene. Their bioinspired internal architecture mimics the asymmetric stacking of pinecone scales. MA‐EPR/CF composites show limited actuation in water but strong responsiveness in isooctane and toluene. PVA/CF materials are more sensitive to water while retaining functionality in hydrocarbons. Importantly, their actuation is reversible and stable through multiple cycles of environmental aging, and the carbon fibers provide significant load‐bearing capabilities, with stiffening effects when passing from dry to wet and immersed states.

Namkyung Yoon, Hwangnam Kim

With the development of artificial intelligence technology, the need for a large amount of high-quality learning data is increasing to be used in various fields. This paper proposes RelayGAN, a new generative model that integrates knowledge inherent in multiple energy data based on the Generative Adversarial Network(GAN) sequentially, similar to relay running. To evaluate the effectiveness of RelayGAN, we conducted extensive experiments using quantitative methods. We employ three statistical metrics, including the Pearson correlation coefficient, the Mann–Whitney U test, and the Kolmogorov–Smirnov test, to validate the quality of the generated data. This shows that RelayGAN improves the performance of conventional multitasking learning-based GAN under the same conditions. Through this, we demonstrate that RelayGAN consistently outperforms state-of-the-art generative models in terms of data quality and pattern preservation. Furthermore, we verify that RelayGAN leverages sequential knowledge transfer to reduce redundant learning processes in accordance with the principles of sustainable AI development, increasing computational efficiency and contributing to eco-friendly AI. Beyond energy data, RelayGAN is a promising approach for multi-source data generation in various edge intelligence applications, ultimately contributing to data-driven innovation.

Huiwen Jia, Ying Liu, Chunming Tang et al.

In this work, we explore the recent developments related to lattice-based signature and preimage sampling, and specify a compact identity-based signature (IBS) on an ideal lattice for practical use. Specifically, we first propose an ellipsoid version of the G + G signature scheme (Asiacrypt 2023) that achieves slightly better signature size and higher security. Then, by adapting a specific preimage sampling algorithm to the modified G + G signature, we obtain an efficient IBS scheme. In addition, we prove its security in the quantum random oracle model (QROM), following the paradigm introduced by Zhangdry (Crypto 2012). Finally, a complete specification of the IBS, featuring three distinct parameter sets, is accompanied by a proof-of-concept implementation. We believe that the combination of the preimage sampling with the Fiat–Shamir transformation holds potential for application in the other advanced digital signature schemes.

Andrea Mio, Romeo Danielis, Giovanni Carrosio et al.

The Italian steel industry requires revamping through strong actions both in the short and medium term. It is essential that Italy meets its steel demand through domestic production, not only to reduce its dependence on imports but also for social and economic reasons. In this context, three decarbonization scenarios for the steel sector in Italy have been developed, including a Conservative pathway, a Potential scenario, and a Desirable one. The Conservative scenario envisions a short-term perspective in which corrective actions mainly involve the addition of CO2 capture to the existing technologies. Potential scenario envisions a medium-term perspective that introduces substantial modifications to production processes (blue hydrogen-based DRI) to achieve complete decarbonization of the sector in the long term. Lastly, Desirable scenario envisions a long-term perspective in which primary steel will be produced using DRI technology based on the use of green hydrogen. Each scenario has been analysed from different viewpoint, considering the CO2 overall emissions, the Levelized Cost of Production (LCOP) of steel and the employment repercussions. The outcomes highlight a good reduction of CO2 for every scenario, with a substantial improvement for Potential and Desirable ones, with 68% less CO2 emissions. From the economic viewpoint, the best results have been achieved by blue hydrogen-based DRI, followed by Conservative scenario and Desirable one. The employment rates are best for green hydrogen-based DRI, due to the relocation of workers into the renewable energy sector.

Adnan Yusuf, Laksha Bhardwaj, Kamlesh Pandey et al.

Barnabás Kiss, Áron Ballagi, Miklós Kuczmann

The objective of this study is the control technology of quadcopters. The aim of this article is to propose further simulation assessment opportunities and other control implementations for investigating the transfer function of a quadrotor BLDC (Brushless Direct Current Electric Motor) motor obtained from experimental results in a previously published paper by separate authors. In this article, an LQ (linear-quadratic) controller is implemented based on the transmission function, during which the response of the controller to a unit step signal is examined. It is proved that LQ control can significantly enhance the autonomy of UAVs (Unmanned Aerial Vehicles) compared to PID (Proportional-Integral-Derivative Controller) control, as a faster and more accurate step response is achieved during system analysis. Additionally, how the LQ controller and the PID controller respond to a randomly generated white noise is examined. The results are compared with those implemented with a PID controller presented in a separate article.

Haiqi Dong, Amanda S Barnard, Amanda J Parker

Acquisition of scientific data can be expensive and time-consuming. Active learning is a solution to reduce costs and time by guiding the selection of scientific experiments. Autonomous and automatic identification of the most essential samples to annotate by active learning can also help to mitigate human bias. Previous research has demonstrated that unlabelled samples causing the largest gradient norms of neural network models can promote active learning in classification. However, gradient norm estimation in regression is non-trivial because the continuous one-dimensional output of regression significantly differs from classification. In this study, we propose a new active learning method that uses meta-learning to estimate the gradient norm of the unlabelled sample in regression. Specifically, we use a separate model to be a selector that learns knowledge from the previous active learning results and is used to predict the gradient norms of unlabelled samples. In each active learning iteration, we estimate and select unlabelled samples with the largest gradient norms to annotate. Our method is evaluated on six regression data sets in various domains, which include costly scientific data.

Stefan Heinen, Danish Khan, Guido Falk von Rudorff et al.

For many machine learning applications in science, data acquisition, not training, is the bottleneck even when avoiding experiments and relying on computation and simulation. Correspondingly, and in order to reduce cost and carbon footprint, training data efficiency is key. We introduce minimal multilevel machine learning (M3L) which optimizes training data set sizes using a loss function at multiple levels of reference data in order to minimize a combination of prediction error with overall training data acquisition costs (as measured by computational wall-times). Numerical evidence has been obtained for calculated atomization energies and electron affinities of thousands of organic molecules at various levels of theory including HF, MP2, DLPNO-CCSD(T), DFHFCABS, PNOMP2F12, and PNOCCSD(T)F12, and treating them with basis sets TZ, cc-pVTZ, and AVTZ-F12. Our M3L benchmarks for reaching chemical accuracy in distinct chemical compound sub-spaces indicate substantial computational cost reductions by factors of ∼1.01, 1.1, 3.8, 13.8, and 25.8 when compared to heuristic sub-optimal multilevel machine learning (M2L) for the data sets QM7b, QM9 $^\mathrm{LCCSD(T)}$ , Electrolyte Genome Project, QM9 $^\mathrm{CCSD(T)}_\mathrm{AE}$ , and QM9 $^\mathrm{CCSD(T)}_\mathrm{EA}$ , respectively. Furthermore, we use M2L to investigate the performance for 76 density functionals when used within multilevel learning and building on the following levels drawn from the hierarchy of Jacobs Ladder: LDA, GGA, mGGA, and hybrid functionals. Within M2L and the molecules considered, mGGAs do not provide any noticeable advantage over GGAs. Among the functionals considered and in combination with LDA, the three on average top performing GGA and Hybrid levels for atomization energies on QM9 using M3L correspond respectively to PW91, KT2, B97D, and τ -HCTH, B3LYP $\ast$ (VWN5), and TPSSH.

WANG Cheng, LIU Yuansheng, LIU Shengjie

Pedestrian detection is vital to applications in unmanned environment perception.Most existing pedestrian-detection algorithms focus only on ordinary pedestrian targets and do not consider the low accuracy caused by the insufficient pedestrian feature information of small targets;furthermore, they do not offer favorable real-time performance when applied to embedded devices.Hence, a small-target pedestrian-detection algorithm, YOLOv4-DBF, is proposed herein.The conventional convolution is replaced with deeply separable convolution in the YOLOv4 algorithm, which reduces the number of parameters and the computation time of the model, as well as improves the detection speed and real-time performance of the algorithm.Additionally, the concurrent spatial and channel Squeeze & Excitation(scSE) attention module is introduced into the feature fusion component of the YOLOv4 backbone network to enhance the important channels and spatial features of the input pedestrian feature map as well as to enable the network to learn more meaningful feature information.The feature fusion component of the Feature Pyramid Network(FPN) in the YOLOv4 neck is improved to enhance the multiscale feature learning of the pedestrian target in the image, which improves the detection accuracy but increases the amount of computation.After training and verification based on the VOC07+12+COCO dataset, the results show that compared with the original YOLOv4 algorithm, YOLOv4-DBF increases the Average Precision(AP) by 4.16 percentage points and the speed by 27%.Finally, YOLOv4-DBF is accelerate deployed on the TX2 equipment of an unmanned vehicle for real-time testing, where the maximum speed reaches 23FPS.The algorithm proposed herein can effectively improve the accuracy and real-time performance of small-target pedestrian detection.

Haeyong Kang, Chang D. Yoo

An unbiased scene graph generation (SGG) algorithm referred to as Skew Class-Balanced Re-Weighting (SCR) is proposed for considering the unbiased predicate prediction caused by the long-tailed distribution. The prior works focus mainly on alleviating the deteriorating performances of the minority predicate predictions, showing drastic dropping recall scores, i.e., losing the majority predicate performances. It has not yet correctly analyzed the trade-off between majority and minority predicate performances in the limited SGG datasets. In this paper, to alleviate the issue, the <i>Skew Class-Balanced Re-Weighting</i> (SCR) loss function is considered for the unbiased SGG models. Leveraged by the skewness of biased predicate predictions, the SCR estimates the target predicate weight coefficient and then re-weights more to the biased predicates for better trading-off between the majority predicates and the minority ones. Extensive experiments conducted on the standard Visual Genome dataset and Open Image V4 and V6 show the performances and generality of the SCR with the traditional SGG models.

Terlumun Adagba, Aliyu Abubakar, Abubakar Sabo Baba

The use of supplementary cementitious materials in the cement and construction industry is growing rapidly owing to the numerous benefits the application of these materials offers. In this study, the effects of metakaolin (MK) and sugarcane bagasse ash (SCBA) used to partially replace cement on concrete are investigated. The experimental plan was designed using a constant 5% MK and 0-20% SCBA contents by weight. The mix design of 1:2:3 and water-binder (w/b) ratio of 0.5 was employed. Samples prepared were tested at the ages of 7, 14, 28, and 60 days respectively. Concrete workability, water absorption, and densities all showed a decrease with an increase in the percentage of SCBA. The compressive strengths at lower percentages of SCBA (5% and 10%) recorded higher values compared to that of 5% MK and 0% SCBA. An increase in the percentage of SCBA above 10% however led to a decrease in compressive strength. The maximum compressive strength of 22.17N/mm2 was obtained at 60 days in concrete containing 5% MK and 10% SCBA. Both the T-statistics and F-statistics values calculated were statistically significant and exceeded their critical values. This suggests that there is a good relationship between the compressive strength of SCBA and the curing period and that the variation in the curing period and SCBA also causes a variation in the concrete compressive strength. From the results obtained, it is concluded that 5% MK and 10% SCBA can be applied to replace cement for structural concrete production.

Huijun Cao, Wenbo Fan, Lixiang Wang

Jin Sun, Nengbo He, Jiawen Zhang et al.

Raisa Abedin Disha, Sajjad Waheed

Abstract To protect the network, resources, and sensitive data, the intrusion detection system (IDS) has become a fundamental component of organizations that prevents cybercriminal activities. Several approaches have been introduced and implemented to thwart malicious activities so far. Due to the effectiveness of machine learning (ML) methods, the proposed approach applied several ML models for the intrusion detection system. In order to evaluate the performance of models, UNSW-NB 15 and Network TON_IoT datasets were used for offline analysis. Both datasets are comparatively newer than the NSL-KDD dataset to represent modern-day attacks. However, the performance analysis was carried out by training and testing the Decision Tree (DT), Gradient Boosting Tree (GBT), Multilayer Perceptron (MLP), AdaBoost, Long-Short Term Memory (LSTM), and Gated Recurrent Unit (GRU) for the binary classification task. As the performance of IDS deteriorates with a high dimensional feature vector, an optimum set of features was selected through a Gini Impurity-based Weighted Random Forest (GIWRF) model as the embedded feature selection technique. This technique employed Gini impurity as the splitting criterion of trees and adjusted the weights for two different classes of the imbalanced data to make the learning algorithm understand the class distribution. Based upon the importance score, 20 features were selected from UNSW-NB 15 and 10 features from the Network TON_IoT dataset. The experimental result revealed that DT performed well with the feature selection technique than other trained models of this experiment. Moreover, the proposed GIWRF-DT outperformed other existing methods surveyed in the literature in terms of the F1 score.

X. Sheng, Catherine E. Graves, Suhas Kumar et al.

Using memristors, such as oxide and phase change resistive switches, as tunable resistors to construct analog computing hardware accelerators is gaining keen attention. Such accelerators have demonstrated the potential to significantly outperform digital computers in highly relevant applications such as machine learning and image processing. However, improvements in device‐level performance of memristors, including reducing power consumption and high current–induced metal migration in interconnects, need continued developments. Nanoscaling and complementary metal‐oxide semiconductor (CMOS) integration are also of significant importance in commercialization of such accelerators. Here tantalum oxide memristors scaled down to 25 nm sizes and integrated on CMOS transistor circuits are presented. The memristor conductance is programmable with a 6 order‐of‐magnitude operating range, especially with 3‐bits below 10 µS for low current operation. The stability of such levels and the size scaling of the operating parameters are further studied. These results will aid device engineering of memristors and bolster development of neuromorphic hardware accelerators.

Wenjing Hu, Xueyi Zhang, Huihui Geng et al.

Aiming at the problem of uncontrollable magnetic field of permanent magnet generators, a new hybrid excitation generator (HEG) with parallel magnetic circuit is proposed. The HEG consists of combined permanent magnet rotor (PMR) and brushless electric excitation rotor (EER). The PMR has surface-mounted and embedded magnets. The PMR provides the main air gap field, and the brushless EER is used to adjust the air gap field. The operating principle and electromagnetic design scheme of the proposed generator are given in detail. Besides, the matching with two different types of rotors and the flux regulation characteristics is analyzed by using the finite element method. Finally, the output performance of the proposed generator including no-load and load characteristics and output voltage are tested. The results show that the two different types of rotors can be matched efficiently and operated reliably. The internal magnetic flux is easy to adjust in both directions, and the proposed HEG can output stable voltage in the range of wide speed and load.

Daniel González Vega, Maylen Rodríguez Simón, Yaima Malagón Chala

Las habilidades sociales son las capacidades o destrezas sociales específicas requeridas para ejecutar competentemente una tarea interpersonal (Monjas y González, 2000).  Situaciones como pedir la palabra para responder o defender criterios frente al grupo se tornan verdaderamente engorrosas para algunos estudiantes.  El Objetivo de este trabajo fue valorar el comportamiento de las Habilidades Sociales de los estudiantes en cursos virtuales desde la experiencia y percepción de los profesores/tutores del Centro Nacional de Educación a Distancia (CENED). Se desarrolló un estudio descriptivo de corte trasversal, cuya muestra de estudio fue de 18 profesores/tutores. Para la recolección de los datos se usó una encuesta. Como resultado se obtuvo que 6 (33.3%) profesores/tutores refieren que frecuentemente los estudiantes tienen dificultad para hablar en público y 8 (44.4%) profesores/tutores afirman que frecuentemente los estudiantes tienen dificultad para afrontar las críticas. El 55.5% de los profesores/tutores afirman que piensan frecuentemente en la carencia o tenencia de habilidades sociales de los estudiantes en sus cursos virtuales. El 83.3%, justificado por 15 profesores/tutores, hacen referencia al componente comportamental en el estudio de las habilidades sociales. Las herramientas digitales de comunicación que más se utilizan son los foros y los chats de acuerdo con la percepción de 15 (83.3%) de los profesores/tutores en el caso de los foros y 12 (63.7%) en el caso de los chats. Se concluyó que los profesores/tutores perciben la existencia de dificultades en los estudiantes de cursos virtuales respecto a la tenencia de habilidades sociales.