Hasil untuk "Architecture"

Yahui Luo, Guangsheng Gao, Wenwu Hu et al.

To address the challenges of rapid and accurate grape variety identification in natural orchard environments, along with the demand for efficient deployment on mobile devices, we propose in this paper YOLOv11n-GrapeLite, a lightweight model built upon an enhanced YOLOv11n architecture. First, an Efficient Channel Attention (ECA) mechanism is incorporated into the Neck layer. This mechanism adaptively recalibrates feature channel weights to emphasize those relevant to grape variety recognition, suppress background interference, and enhance target feature perception in complex scenes. Second, an adaptive downsampling (ADown) strategy is employed to replace the traditional convolutional downsampling module, reducing computational complexity while preserving critical features. Finally, the original C3k2 module is redesigned as a multi-scale convolution block (MSCB). This block integrates depthwise separable convolutions with multi-scale convolutions, which achieves significant parameter compression and enhances multi-scale feature extraction. Experimental results demonstrate that the proposed model achieves a mean average precision (mAP) of 91.5%, representing a 0.2% improvement over the original YOLOv11n, along with a 0.6% increase in recall. These results indicate outstanding robustness in complex field scenarios. The model’s parameter count was reduced to 1.87 million, computational complexity to 5.0 GFLOPS, and model size to 4.1 MB. These figures represent reductions of 27.8%, 23.1%, and 25.5%, respectively, compared to the original YOLOv11n, demonstrating significant lightweight optimization. Compared to mainstream models such as YOLOv6, YOLOv8n, YOLOv9s, YOLOV12, YOLOv13 and YOLOv26, the proposed model achieves superior performance in parameter count, computational load, and model size, while maintaining competitive detection accuracy. The YOLOv11n-GrapeLite model efficiently adapts to mobile terminal deployment, providing a feasible and efficient technical solution for real-time, precise identification of grape varieties in complex field scenarios.

Satoshi Otsuki, Miki Taya, Kenichi Nakashima et al.

This paper presents a novel hierarchical control architecture designed for navigating multiple tugboats to perform complex maneuvers, including approaching, enclosing, and capturing a large target vessel, while taking into account external disturbances and operational constraints. The proposed hierarchical control architecture includes a high-level nominal controller for trajectory generation, tracking, and coordination, and a low-level model-following controller for multitask execution. The low-level controller integrates constraint-driven control to refine nominal control inputs generated by the high-level controller, and local control to mitigate disturbances. To achieve multitask execution across all phases, the constraint-driven controller activates only essential constraints, enforcing seamless maneuver transitions without relying on ad-hoc controller switching. Specifically, we design constraints to facilitate effective enclosing behaviour while guiding the tugboat fleet toward the target in formation. We also adopt a so-called Prescribed-Time Safety Filter to enforce mild contact with the surface of the target vessel within a specified finite-time interval. Simulation studies validate the proposed control architecture across all maneuvers throughout the operation and demonstrate its capability to achieve complex multi-tugboat coordination.

Daniel Lindenschmitt, Marcos Rates Crippa, Hans D. Schotten

The emergence of nomadic mobile communication networks for sixth-generation (6G) introduces a paradigm shift in how network infrastructure is conceptualized, deployed, and operated. Unlike traditional fixed systems, Nomadic Networks (NNs) consist of mobile and self-organizing nodes that provide radio infrastructure capabilities in motion. This paper explores the architectural implications of such systems, with a particular focus on the design and evolution of network interfaces. We analyze the requirements for inter-node communication, service discovery, and control delegation in dynamic environments. Furthermore, we examine the regulatory and licensing challenges that arise when infrastructure elements traverse jurisdictional boundaries. Based on current 6G visions and relevant research, we identify limitations in existing architectures and propose a set of interface principles tailored to nomadicity. By synthesizing findings from mobile, non-terrestrial, and organic network domains, this work contributes to the architectural foundation for future nomadic 6G communication systems and outlines directions for interface standardization in decentralized, mobile infrastructures.

Pratyush Dhingra, Janardhan Rao Doppa, Partha Pratim Pande

Transformer architectures have become the standard neural network model for various machine learning applications including natural language processing and computer vision. However, the compute and memory requirements introduced by transformer models make them challenging to adopt for edge applications. Furthermore, fine-tuning pre-trained transformers (e.g., foundation models) is a common task to enhance the model's predictive performance on specific tasks/applications. Existing transformer accelerators are oblivious to complexities introduced by fine-tuning. In this paper, we propose the design of a three-dimensional (3D) heterogeneous architecture referred to as Atleus that incorporates heterogeneous computing resources specifically optimized to accelerate transformer models for the dual purposes of fine-tuning and inference. Specifically, Atleus utilizes non-volatile memory and systolic array for accelerating transformer computational kernels using an integrated 3D platform. Moreover, we design a suitable NoC to achieve high performance and energy efficiency. Finally, Atleus adopts an effective quantization scheme to support model compression. Experimental results demonstrate that Atleus outperforms existing state-of-the-art by up to 56x and 64.5x in terms of performance and energy efficiency respectively

Janith Petangoda, Chatura Samarakoon, James Meech et al.

Computing systems interacting with real-world processes must safely and reliably process uncertain data. The Monte Carlo method is a popular approach for computing with such uncertain values. This article introduces a framework for describing the Monte Carlo method and highlights two advances in the domain of physics-based non-uniform random variate generators (PPRVGs) to overcome common limitations of traditional Monte Carlo sampling. This article also highlights recent advances in architectural techniques that eliminate the need to use the Monte Carlo method by leveraging distributional microarchitectural state to natively compute on probability distributions. Unlike Monte Carlo methods, uncertainty-tracking processor architectures can be said to be convergence-oblivious.

Agboola Oluwagbemiga Paul, Nnezi Uduma-Olugu

The study's primary focus is to gather profound insights into user perceptions and preferences related to living in smart environments prioritizing thermal comfort and energy efficiency. This objective is pursued through meticulous surveys designed to capture the nuanced experiences of residents, unravelling the intricate relationship between smart technologies and their impact on daily life within these intelligent living spaces. The findings from these quantitative surveys become a valuable repository of information that sheds light on the intricate dynamics of user satisfaction and experience, providing a holistic understanding of the role played by smart technologies in shaping thermal comfort in the built environment. Additionally, the research sets out to assess the specific influence of smart technologies on thermal comfort within the urban settings of Turkey. By narrowing the focus to this geographical context, the study aims to draw region-specific insights that can be instrumental in tailoring smart living solutions to the unique needs and nuances of the Turkish urban landscape. This contextual analysis allows for a nuanced understanding of how smart technologies operate in diverse urban environments, providing a foundation for targeted interventions and improvements. A core objective of the study is to distil actionable recommendations for architects, designers, and urban planners. These recommendations are crafted with the intention of guiding the creation of user-centric spaces within the built environment that not only meet but exceed expectations in optimizing thermal comfort. By translating survey findings and regional assessments into practical suggestions, this research aims to empower professionals in the field to integrate smart technologies seamlessly into their designs, ultimately contributing to the development of intelligent, sustainable, and people-focused spaces. In summary, the study positions itself as a comprehensive exploration of the symbiotic relationship between smart technologies and thermal comfort. Through a meticulous examination of user perceptions, regional influences, and actionable recommendations, the research seeks to chart a course for a future where the built environment actively supports and enhances thermal comfort, thereby improving the overall quality of life for its residents.

Sebastián Seguel-Vargas, Carlos Rubio-Bellido, Lucía Pereira-Ruchansky et al.

Reducing energy consumption in the construction sector is urgently needed. In Chile, where income distribution is unequal and the cost of energy is high, energy demand is seriously affected, especially in vulnerable households. Hence, it is essential to establish public policies with more realistic energy-saving goals to address this situation. However, reliably predicting the energy performance of buildings remains a challenge. For this reason, this study aims to identify and evaluate the impact of the variables associated with energy performance in vulnerable households in Central-Southern Chile and propose values that would reduce the gap. A sensitivity analysis was conducted to achieve this, adjusting the energy performance parameters in a base model with data analyzed using local standards. In addition, field information was collected in 93 households to obtain the actual energy consumption. The main results show that the variables that most impacted performance were infiltration, COP, heating setpoints, and schedules, which generated a 60% difference between the theoretical and actual consumption.

Daniel Sanchez‐Morillo, Antonio León‐Jiménez, María Guerrero‐Chanivet et al.

Abstract Engineered stone silicosis (ESS), primarily caused by inhaling respirable crystalline silica, poses a significant occupational health risk globally. ESS has no effective treatment and presents a rapid progression from simple silicosis (SS) to progressive massive fibrosis (PMF), with respiratory failure and death. Despite the use of diagnostic methods like chest x‐rays and high‐resolution computed tomography, early detection of silicosis remains challenging. Since routine blood tests have shown promise in detecting inflammatory markers associated with the disease, this study aims to assess whether routine blood biomarkers, coupled with machine learning techniques, can effectively differentiate between healthy individuals, subjects with SS, and PMF. To this end, 107 men diagnosed with silicosis, ex‐workers in the engineered stone (ES) sector, and 22 healthy male volunteers as controls not exposed to ES dust were recruited. Twenty‐one primary biochemical markers derived from peripheral blood extraction were obtained retrospectively from clinical hospital records. Relief‐F features selection technique was applied, and the resulting subset of 11 biomarkers was used to build five machine learning models, demonstrating high performance with sensitivities and specificities in the best case greater than 82% and 89%, respectively. The percentage of lymphocytes, the angiotensin‐converting enzyme, and lactate dehydrogenase indexes were revealed, among others, as blood biomarkers with significant cumulative importance for the machine learning models. Our study reveals that these biomarkers could detect a chronic inflammatory status and potentially serve as a supportive tool for the diagnosis, monitoring, and early detection of the progression of silicosis.

张钊(ZHANG Zhao), 毛义华(MAO Yihua), 王凯(WANG Kai) et al.

With the impact of the epidemic and the deepening of population aging in China, the distribution and quality of medical resources have become important factors affecting housing prices, gradually generating the capitalization effects of medical resources. In this study, the differences in the resident's preference for the infectious hospital and general 3A hospitals in Harbin were explored in depth through a questionnaire survey and a comparative analysis of their capitalization effects. Furthermore, the social heterogeneity of the capitalization effects of medical resources and the homogeneity of the two kinds of medical resources were analyzed based on quantile regression models and interaction effects tests. The results show that (1) the infectious hospital depresses the prices of nearby housings, and general 3A hospitals increase the prices of nearby housings. The capitalization effect of both medical resources gradually decreases with increasing distance. (2) Medium-priced housings are more sensitive to the proximity of the infectious hospital, and the capitalization effect of general 3A hospitals gradually increases as the price of housings increases. (3) There is an interaction between the capitalization effects of the two kinds of medical resources, and the proximity of general 3A hospitals enhances the NIMBY (not in my backyard) effect of infectious hospitals.(受新型冠状病毒感染冲击以及我国人口老龄化程度加深的影响，医疗资源的分布与质量成为影响住宅价格的重要因素，从而产生了医疗资源资本化效应。通过问卷调查，分析了城市居民对哈尔滨市传染病医院与全科三甲医院两种医疗资源的偏好差异。基于分位数回归模型与交互效应检验，深入探讨了医疗资源资本化效应的空间异质性、社会异质性以及两种医疗资源的同质性。结果表明：（1）传染病医院抑制了附近的住宅价格，全科三甲医院提升了附近的住宅价格，且随着距离的增加，两种医疗资源的资本化效应均逐渐减弱；（2）中等价位住宅对与传染病医院的距离更敏感，随着住宅价格的提高，全科三甲医院的资本化效应逐渐增强；（3）两种医疗资源具有交互资本化效应，住宅与全科三甲医院临近增强了传染病医院的邻避效应。)

Dhandeep Challagundla, Ignatius Bezzam, Riadul Islam

While general-purpose computing follows Von Neumann's architecture, the data movement between memory and processor elements dictates the processor's performance. The evolving compute-in-memory (CiM) paradigm tackles this issue by facilitating simultaneous processing and storage within static random-access memory (SRAM) elements. Numerous design decisions taken at different levels of hierarchy affect the figure of merits (FoMs) of SRAM, such as power, performance, area, and yield. The absence of a rapid assessment mechanism for the impact of changes at different hierarchy levels on global FoMs poses a challenge to accurately evaluating innovative SRAM designs. This paper presents an automation tool designed to optimize the energy and latency of SRAM designs incorporating diverse implementation strategies for executing logic operations within the SRAM. The tool structure allows easy comparison across different array topologies and various design strategies to result in energy-efficient implementations. Our study involves a comprehensive comparison of over 6900+ distinct design implementation strategies for EPFL combinational benchmark circuits on the energy-recycling resonant compute-in-memory (rCiM) architecture designed using TSMC 28 nm technology. When provided with a combinational circuit, the tool aims to generate an energy-efficient implementation strategy tailored to the specified input memory and latency constraints. The tool reduces 80.9% of energy consumption on average across all benchmarks while using the six-topology implementation compared to baseline implementation of single-macro topology by considering the parallel processing capability of rCiM cache size ranging from 4KB to 192KB.

Qingmin Jia, Yujiao Hu, Xiaomao Zhou et al.

With the development of new Internet services such as computation-intensive and delay-sensitive tasks, the traditional "Best Effort" network transmission mode has been greatly challenged. The network system is urgently required to provide end-to-end transmission determinacy and computing determinacy for new applications to ensure the safe and efficient operation of services. Based on the research of the convergence of computing and networking, a new network paradigm named deterministic computing power networking (Det-CPN) is proposed. In this article, we firstly introduce the research advance of computing power networking. And then the motivations and scenarios of Det-CPN are analyzed. Following that, we present the system architecture, technological capabilities, workflow as well as key technologies for Det-CPN. Finally, the challenges and future trends of Det-CPN are analyzed and discussed.

Zhongkai Yu, Shengwen Liang, Tianyun Ma et al.

Deploying advanced large language models on edge devices, such as smartphones and robotics, is a growing trend that enhances user data privacy and network connectivity resilience while preserving intelligent capabilities. However, such a task exhibits single-batch computing with incredibly low arithmetic intensity, which poses the significant challenges of huge memory footprint and bandwidth demands on limited edge resources. To address these issues, we introduce Cambricon-LLM, a chiplet-based hybrid architecture with NPU and a dedicated NAND flash chip to enable efficient on-device inference of 70B LLMs. Such a hybrid architecture utilizes both the high computing capability of NPU and the data capacity of the NAND flash chip, with the proposed hardware-tiling strategy that minimizes the data movement overhead between NPU and NAND flash chip. Specifically, the NAND flash chip, enhanced by our innovative in-flash computing and on-die ECC techniques, excels at performing precise lightweight on-die processing. Simultaneously, the NPU collaborates with the flash chip for matrix operations and handles special function computations beyond the flash's on-die processing capabilities. Overall, Cambricon-LLM enables the on-device inference of 70B LLMs at a speed of 3.44 token/s, and 7B LLMs at a speed of 36.34 token/s, which is over 22X to 45X faster than existing flash-offloading technologies, showing the potentiality of deploying powerful LLMs in edge devices.

Kirsten Carter McKee

Over the last decade, analyses of Scotland’s historic global diasporas have incorporated more pronounced conversations on how Scotland’s current political, social, and economic contexts are rooted in the legacies of the British Empire. While this has produced narratives highlighting Scotland’s key role in imperial expansion, the resonance of this in establishing and perpetuating systems of white oppression are less widely addressed in Scotland’s consciousness of its own identity. Through consideration of how architecture’s cultural analogies reflect and represent Imperial ideologies, this paper will explore the resonance of architectural urban discourse funded by the outputs of the British Empire. It will discuss how an architecturally focused reading of our built environment can clearly recognise the systemic legacies of colonialism and imperialism within our urban realm, and further enhance inclusive narratives of Scotland’s heritage. This will highlight how a more nuanced approach to reading the historic built environment is necessary to challenge established current authorised heritage discourse of white male histories. It will demonstrate the function of the built environment in telling stories of Scotland’s prominent role in Empire and how this supports a human-rights based approach to heritage analysis.

Pouya Houshmand, Jiacong Sun, Marian Verhelst

In-memory-computing is emerging as an efficient hardware paradigm for deep neural network accelerators at the edge, enabling to break the memory wall and exploit massive computational parallelism. Two design models have surged: analog in-memory-computing (AIMC) and digital in-memory-computing (DIMC), offering a different design space in terms of accuracy, efficiency and dataflow flexibility. This paper targets the fair comparison and benchmarking of both approaches to guide future designs, through a.) an overview of published architectures; b.) an analytical cost model for energy and throughput; c.) scheduling of workloads on a variety of modeled IMC architectures for end-to-end network efficiency analysis, offering valuable workload-hardware co-design insights.

Ji Zhou, Zhenping Xing, Haide Wang et al.

To cope with the explosive bandwidth demand, significant progress has been made in the ITU-T standardization sector to define a higher-speed passive optical network (PON) with a 50Gb/s line rate. Recently, 50G PON becomes mature gradually, which means it is time to discuss beyond 50G PON. For ensuring an acceptable optical power budget, beyond 50G PON will potentially use coherent technologies, which can simultaneously promote the applications of flexible multiple access such as time/frequency-domain multiple access (TFDMA). In this paper, we will introduce the architectures, algorithms, and demonstrations for TFDMA-based coherent PON. The system architectures based on an ultra-simple coherent transceiver and specific signal spectra are designed to greatly reduce the cost of ONUs. Meanwhile, fast and low-complexity digital signal processing (DSP) algorithms are proposed for dealing with upstream and downstream signals. Based on the architectures and algorithms, we experimentally demonstrate the first real-time TFDMA-based coherent PON, which can support at most 256 end users, and peak line rates of 100Gb/s and 200Gb/s in the upstream and downstream scenarios, respectively. In conclusion, the proposed technologies for the coherent PON make it more possible to be applied in the future beyond 50G PON.

Oscar Gonzalez de Dios, Ramon Casellas, Filippo Cugini et al.

Both public and private innovation projects are targeting the design, prototyping and demonstration of a novel end-to-end integrated packet-optical transport architecture based on Multi-Band (MB) optical transmission and switching networks. Essentially, MB is expected to be the next technological evolution to deal with the traffic demand and service requirements of 5G mobile networks, and beyond, in the most cost-effective manner. Thanks to MB transmission, classical telco architectures segmented into hierarchical levels and domains can move forward toward an optical network continuum, where edge access nodes are all-optically interconnected with top-hierarchical nodes, interfacing Content Delivery Networks (CDN) and Internet Exchange Points (IXP). This article overviews the technological challenges and innovation requirements to enable such an architectural shift of telco networks both from a data and control and management planes.

Disha Sushant Wankhede, R. Selvarani

A correct diagnosis of brain tumours is crucial to making an accurate treatment plan for patients with the disease and allowing them to live a long and healthy life. Among a few clinical imaging modalities, attractive reverberation imaging gives extra different data about the tissues. The use of MRI-Magnetic Resonance Imaging tests is a significant method for identifying disorders throughout the human body. Deep learning provides a solution for efficiently detecting Brain Tumour. The work has used MRI images for predicting the glioblastoma of brain tumours. Initially, data is retrieved from hospitals in form of an image database to continue with the brain tumour prediction. Pre-processing of dataset images is a mandatory step to enhance the accuracy and smooth line supplementary stages. The intensity value of each MRI (Magnetic Resonance Imaging) is subtracted by the mean intensity value and standard deviation of the brain region. Further, reduce the medical image noise by employing a bilateral filter. Further, the preprocessed medical images are used for extracting the radiomics features from images as well as tumour segmentation. Thus the work adopts the tumor is automatically segmented into four compartments using mutually exclusive rules using Modified Fuzzy C Means Clustering (MFCM). The clustering-based approach is very beneficial in MR tumour segmentation; it categorizes the pixels using certain radiomics features. The most important problem in the radiomics-based machine learning model is the dimension of data. Moreover, using a GWO (Grey Wolf Optimizer) with rough set theory, we propose a novel dimensionality reduction algorithm. This method is employed to find the significant features from the extracted images and differentiate HG (high-grade) and LG (Low-grade) from GBM while varying feature correlation limits were applied to remove redundant features. Finally, the article proposed the dynamic architecture of Multilevel Layer modelling in Faster R-CNN (MLL-CNN) approach based on feature weight factor and relative description model to build the selected features. This reduces the overall computation and performs long-tailed classification. This results in the development of CNN training performance more accurate. Results show that the general endurance expectation of GBM cerebrum growth with more prominent exactness of about 95% with the decreased blunder rate to be 2.3%. In the calculation of similarity between segmented tissues and ground truth, different tools produce correspondingly different predictions.

Chenyang He, Yindi Zhao, Jihong Dong et al.

Mastering urban change information is of great importance and significance in practical areas such as urban development planning, land management, and vegetation cover. At present, high-resolution remote sensing images and deep learning techniques have been widely used in the detection of urban information changes. However, most of the existing change detection networks are Siamese networks based on encoder–decoder architectures, which tend to ignore the pixel-to-pixel relationships and affect the change detection results. To solve this problem, we introduced a generative adversarial network (GAN). The change detection network based on the encoder–decoder architecture was used as the generator of the GAN, and the Jensen-Shannon(JS) scatter in the GAN model was replaced by the Wasserstein distance. An urban scene change detection dataset named XI’AN-CDD was produced to verify the effectiveness of the algorithm. Compared with the baseline model of the change detection network, our generator outperformed it significantly and had higher feature integrity. When the GAN was added, the detected feature integrity was better, and the F1-score increased by 4.4%.

Amirreza Asnafi, Mohsen Haji Zeinolabedini, Faezeh Ahmadipour

Access to the required information in all available scientific disciplines is one of the most important factors in the survival of that field. In the architecture field, the type of information format differs from other disciplines. The purpose of this study was to identify the behavior of images in the architecture of Shahid Beheshti University. The present study is an applied target and uses a descriptive survey method. The statistical population of the study consists of two groups of students and professors in the architecture major of Shahid Beheshti University. To determine the sample size, the Cochran formula was used and the sample size in this formula was 296 people. The results showed that the architects mainly used images for identifying creative ideas and taking advantage of the details of architectural structures. The type of image content they used was mostly photos, maps, and charts, which could be found in engines and image databases by limiting the size of the image and following related links as long as the image was taken. One of the major obstacles in finding images for architects was the lack of familiarity with the way they were searched. Creativity, proximity to the subject, credibility, and quality of the images were the criteria for selecting content. Considering the library's share in retrieving research-based images, it is suggested that library and library librarians conduct awareness-raising activities at the university's research groups such as brochures, conferences, library visits, and workshops.

Stanislav Fedorenko, Lesya Vasylenko, Yuliia Bereznytska et al.

The development of environmental management in Ukraine is determined by the urgent need to overcome environmental problems and ensure the environmental safety of society, especially under the conditions of martial law. Today, the domestic economy is three times more resource-intensive than the world economy, the technological base and infrastructure complex of public production are rapidly wearing out, which leads to a decrease in the level of technological and environmental safety. Environmental management is related to the national economy and forms information about the need to use natural resources when promoting effective development. A comprehensive project-targeted approach to the development of new forms of ownership and market economy reflects the interrelationship of all parts of the nature management project. The development of the scientific foundations of nature management is facilitated by the formulation of a general plan for the placement of productive forces. The ecological situation in Ukraine has long been called a crisis. In recent decades, new scientific directions have appeared, the result of which have been new ideas about human, society and nature and their coexistence. One of these directions is environmental management, which today is the ideology of production activity management, as it provides an effective toolkit for solving current problems and preventing the emergence of new production environmental and economic issues.