Search

Abstract: This article uses lightweight J2EE tools based on B/S architecture to design an auto-generated paper management system. This system includes several modules, namely: user management, subject management, classification management, question entry, question management, paper generation, and paper management. The design process performs the analysis and composes the examination paper using an efficient algorithm with a high possibility of success. With this algorithm, the user needs to identify the subject, the question type and the difficulty level. From this input, the examination paper will be generated automatically. On the basis of a study of the JSP and Microsoft Word interface, we find that the system combines with Word very well. Therefore the editing of questions and answers is performed using Word and the final paper may be stored as “.doc” files. The system has characteristics such as easy operation, a good interface, good usability, high security, high stability, and reliability. [Copyright &y& Elsevier]

The integration of Adaptive Business Intelligence systems in healthcare has garnered significant attention due to their potential to manage the ever-growing volume of healthcare data and enhance the quality of care provided to society. ABI systems also play a crucial role in supporting hospital administrators in making strategic decisions. To facilitate the transparency and interoperability of these solutions, the scientific community has embarked on various studies to develop technologic architectures capable of meeting the complex requirements of healthcare settings. One of the key challenges in adopting this technology is the creation and integration of prediction and optimization models in an automated and semi-autonomous manner. This article presents a novel and robust microservices architecture designed to streamline the deployment of intelligent models and seamlessly integrate them within the ABI system. This paper begins by introducing the problem of deploying and integrating intelligent models into ABI systems, providing essential context on ABI systems within the healthcare domain. Subsequently, it details the proposed architecture, outlining its technical approaches and highlighting the advantages it brings to the healthcare ecosystem. Finally, the paper concludes by summarizing the contributions and future directions for research in this critical area, emphasizing the potential impact of this architecture on improving healthcare intelligence systems. [ABSTRACT FROM AUTHOR]

Human exploration of Mars and beyond will demand unprecedented levels of self-sufficiency due to the exceedingly far distances from Earth and lengthy mission durations. As such, astronaut crews must be able to function in a progressively autonomous manner, as they will be unable to rely on timely logistical resupply and/or operational guidance from ground-based mission support teams. One means of increasing autonomy is by incorporating intelligent (i.e., smart) technologies into the mission architecture and habitat design. As plans for sending humans to deep space continue to develop and mature, identifying current and emerging smart technologies and which have the potential to enable these missions or maintain habitability for the crew can be used to explore design options prioritized for autonomous operations. This paper surveys select commercial-off-the-shelf (COTS) technologies currently available and marketed terrestrially for use in integrated smart homes and aligns them for consideration by their functional relevance toward potential Smart Habitat (SmartHab) applications in deep-space. • Self-sufficiency becomes increasingly necessary as distance from Earth prohibits timely communication and resupply capability. • Habitability affects a crew's physical, physiological, psychological, and social health, as well as human performance. • Self-sufficiency and habitability for deep space missions may benefit from utilizing 'smart systems'to enable autonomy. • Over 90 smart technologies were reviewed resulting in 97 space-relevant applications for deep space missions. [ABSTRACT FROM AUTHOR]

This paper proposes an integrated architecture for navigation and attitude control of low-cost suborbital launch vehicles, propelled by a solid motor. Single-nozzle, two Degrees-of-Freedom (DoF) Thrust Vector Control (TVC) actuation is adopted. For architecture design purposes, a non-linear, unstable, 6 DoF model for the generic thrust-vector-controlled launcher dynamics and kinematics is deduced, and a linear state-space representation is proposed. The navigation system provides full-state estimates resorting to novel complementary kinematic filters, whose design allows to establish an explicit relation with steady-state Kalman filtering. A globally stable estimation solution is obtained, apart from the singularities arising from the use of Euler angles. The attitude control law is derived from the Linear Quadratic Regulator (LQR) using the state-space models for each linearization point of the reference trajectory, with an integral action (LQI) added to improve robustness and to provide null steady state attitude tracking error. A correction method is proposed to allow for pitch and yaw control in the presence of spinning motion, precluding the need of a supplementary roll control system. The control system is implemented through gain scheduling, resorting to an altitude-based linear parametric varying method. The architecture is implemented in a realistic simulation environment, composed by the 6 DoF non-linear model, the Navigation and Control solutions, and the environmental disturbances, to assess its performance through Monte Carlo simulations. The navigation system is able to provide accurate estimates of the state vector, while the control system satisfies attitude tracking performance and robustness to both external disturbances and model parametric uncertainties. • An integrated architecture for navigation and attitude control of suborbital launch vehicles is proposed. • An original time-varying state-space representation is deduced. • Novel complementary kinematic filters are designed to provide accurate full-state estimation. • LQI-based control allows for accurate pitch and yaw reference tracking via thrust vectoring. • The architecture performance is assessed through extensive Monte Carlo simulations. [ABSTRACT FROM AUTHOR]

Change detection in high-resolution Earth observation is a fundamental Earth vision task to understand the subtle temporal dynamics of Earth's surface, significantly promoted by generic vision technologies in recent years. Vision Transformer is a powerful component to learning spatiotemporal representation but with enormous computation complexity, especially for high-resolution images. Besides, there is still lacking principles in designing macro architectures integrating these advanced vision components for various change detection tasks. In this paper, we present a deep probabilistic change model (DPCM) to provide a unified, interpretable, modular probabilistic change process modeling to address multiple change detection tasks, including binary change detection, one-to-many semantic change detection, and many-to-many semantic change detection. DPCM describes any complex change process as a probabilistic graphical model to provide theoretical evidence for macro architecture design and generic change detection task modeling. We refer to this probabilistic graphical model as the probabilistic change model (PCM), where DPCM is the PCM parameterized by deep neural networks. For parameterization, the PCM is factorized into many easy-to-solve distributions based on task-specific assumptions, and then we can use deep neural modules to parameterize these distributions to solve the change detection problem uniformly. In this way, DPCM has both theoretical macro architecture from PCM and strong representation capability of deep networks. We also present the sparse change Transformer for better parameterization. Inspired by domain knowledge, i.e., the sparsity of change and the local correlation of change, the sparse change Transformer computes self-attention within change regions to model spatiotemporal correlations, which has a quadratic computational complexity of the change region size but independent of image size, significantly reducing computation overhead for high-resolution image change detection. We refer to this instance of DPCM with sparse change Transformer as ChangeSparse to demonstrate their effectiveness. The experiments confirm ChangeSparse's superiority in speed and accuracy for multiple real-world application scenarios, such as disaster response and urban development monitoring. The code is available at https://github.com/Z-Zheng/pytorch-change-models. More resources can be found in http://rsidea.whu.edu.cn/resource_sharing.htm. [ABSTRACT FROM AUTHOR]

Smart Cities are complex systems where digital innovations are applied to design enhanced city services and enable public value creation across various domains (e.g., mobility, energy and infrastructure, environment, and so). Such complexity challenges the alignment between city strategies with city services and underlying information systems, leading to providing services that may fail to achieve city goals and meet the needs of citizens. This paper defines a set of design principles to ensure strategic alignment in Smart City Enterprise Architectures (SCEA). The design principles were derived from the literature and validated by smart city domain experts, following a design-oriented research approach. The main contribution of this paper is the formulated design principles that deepen the knowledge of Enterprise Architecture design in the smart cities field. These design principles can be used by researchers and practitioners to guide the design of innovative digital services aligned with urban strategies. [ABSTRACT FROM AUTHOR]

A multitude of outbreak prediction models are employed on a global scale to facilitate informed decision-making and the implementation of appropriate control measures in light of the COVID-19 pandemic. The authorities are now focusing on a straightforward epidemiological and statistical methodology for forecasting the global COVID-19 epidemic. These models are widely recognized within the media landscape. Long-term prediction has been found to have poor precision in typical models due to the presence of considerable uncertainty and insufficient vital data. Despite several efforts that have been undertaken to tackle this issue, there is a necessity to enhance the fundamental generalization and robustness of current models. There is a need for a comparative analysis of recurrent neural networks (RNN) and long short-term memory models (LSTM) in order to forecast the breakout of COVID-19, in comparison to the currently existing models. The architectural design incorporates the concept of the attention mechanism, specifically, the use of attention inside hidden data. This methodology enables the identification of patterns and the prioritization of time-sensitive information, leading to the creation of a network that is easily comprehensible. This research presents a proposal for utilizing RNN with LSTM to anticipate outbreaks of COVID-19. Additionally, we have incorporated a learning vector time embedding into our methodology, as the inclusion of time in a vector representation may be seamlessly integrated into many architectures. The model provides a forecast of the progression of the COVID-19 pandemic in Europe's most severely affected nations, including Italy, France, Spain, as well as a country in North America, specifically Canada. This study proposes that variations in behavior across different nations indicate the potential efficacy of employing machine learning as a modelling tool for understanding and predicting the spread of epidemics. This paper provides an initial benchmarking analysis to demonstrate the potential of recurrent neural networks (RNNs) for future academic investigations. The study presents a unique method that combines RNNs with LSTM to effectively forecast flare-ups. The experimental results show that our proposed LSTM model performs better than existing models with lesser RMSE and MAPE values. [ABSTRACT FROM AUTHOR]

IT-systems supporting business operations are a cornerstone of industry 4.0 and smart manufacturing. IT-systems must not be considered as an end in itself. The enterprise IT architecture must be aligned with the business goals and strategy. Current approaches for designing enterprise IT architectures provide little guidance for greenfield environments in which there are little given boundaries, such as operation procedures. In smart manufacturing this is challenging due to the complexity of the shop floor involving OT and IT. This paper addresses this gap by combining the TOGAF architecture development method with the manufacturing operations management standard IEC 62264. The paper explores this combination by the example of designing an enterprise IT architecture for the Fraunhofer battery manufacturing research line involving the TOGAF phases architecture vision, business architecture, information systems architecture and technology architecture for manufacturing. [ABSTRACT FROM AUTHOR]

This paper traces the early ideas of lunar architecture developed by the US and Soviet space programs. Starting with a historical overview of those achievements, highlighting specific engineering, architectural, and design elements; illustrated with images and drawings. Some of them will be described in more detail, showing the design and decision-making process from the first sketch to its implementation. One example of the great engineering and management capabilities is the development and inclusion of the deployable lunar roving vehicle (LRV). Other examples include the project DLB Zvezda by Barmin's Design Bureau of General Engineering that was under development between 1962 and 1974, and a US project Horizon – a study to set up a Lunar Military outpost conducted in 1959. The paper includes personal statements from astronauts and cosmonauts, as well as people involved along design and implementation processes. Based on a historical overview of achievements, this paper highlights particular architectural and design unique approaches that defined design strategies for creating space systems, spacecraft and habitats for decades to follow. • Lunar architecture developed by the US and Soviet space programs. • Specific engineering, architectural and design elements. • Lessons learned from earlier lunar base designs. • Space habitat designs. [ABSTRACT FROM AUTHOR]

Museums environment is complicated and defined by indoor air quality, thermal and lighting comfort. Artifacts deterioration is factored by extreme conditions of thermal exposure or excessive lighting. With high control required for exposed old monuments and visitor comfort, museums energy systems reach extreme levels. In this paper, we aim to find the gap in museums high energy loads and reach solutions through architectural design. The paper studies comparison results for different techniques in fifty recent case studies to identify the specific factors that matter most to museum buildings. These factors are implemented simultaneously on one base model in three climatic states by multi-objective optimization. The best option will showcase each climate optimum conditions. The paper introduces optimum architectural procedures optimizing sDA and ASE to minimum 70% and maximum 10% respectively while decreasing thermal load. The results help architects and policy makers achieve daylighting and energy optimization in museums through architecture. [ABSTRACT FROM AUTHOR]

• ADA learns preferences from market data and predicts price of newly generated building design. • Integrate ADA with 3D building simulation to study price gain and value at risk. • Changes to Visual Capital identify the exposure and financial sensitivity to land-use change. • Blueprint for integrating economic evaluation within site-selection, design optimization. Uncertainty of spillover effects – including property devaluation - from proposed land-use change elicits opposition to local development. This hinders cities' ability to implement land-use policy aimed at housing affordability and environmental sustainability. The uncertainty arises from the limitation of current methods to quantify both the physical exposure and financial sensitivity to changes in urban environmental characteristics and amenities. Hence, this paper introduces an integrated approach to quantify the direct and indirect effects of a proposed urban development, using parametric modeling, building performance simulation and automated design appraisal (ADA). ADA, a novel aspect of this study, predicts the market price of a generated building design concept within a local urban context. To demonstrate its practical utility, this paper examines property value at risk due to nearby development associated with changes to visual landscape quality. The results shed light on the importance and spatial variability of the price-amenity gradient, identifying properties that are financially at risk to the visual impacts of local development. Beyond building-specific risk modeling for the development planning stage, this work serves as a blueprint for architectural design optimization procedures, in which economic performance is evaluated based on learned preferences derived from financial market data. [ABSTRACT FROM AUTHOR]

It is widely recognized that the sustainability of the built environment is one of the key strategies to tackle global crises such as climate change, resource depletion, and human health. An environmentally driven approach to architectural design and urban planning through simulations has been around for many decades. However, architects and planners haven't fully taken advantage of its potential yet due to the complexity of the tasks and the fragmentation of tools and methods in different fields of expertise. This paper aims to demonstrate how the nascent field of environmental architectural design exploration is a viable answer to complex tasks and challenges through the integration of sustainable design, environmental simulations, building performance analysis, and computational procedures leveraging parametric design environments. After defining the context that favored its development, the ten questions answered present the environmental metrics and design parameters, propose a taxonomy of simulation tools and methods, present optimization processes, identify the required data, and introduce a classification of co-simulation and design exploration workflows, emphasizing their potential for the design of sustainable buildings and districts. Then focus is given to the scales and domains of application and to the potential of machine learning methods. The paper concludes by presenting applications in practice and discussing challenges and future opportunities. Strategies and operational procedures emerge from the ten questions. They emphasize the use of holistic approaches and the selection of balanced design solutions allowed by environmental architectural design exploration for the development of a sustainable built environment. [Display omitted] • The paper introduces environmental architectural design exploration. • Metrics, design parameters, tools and methods are presented and classified. • Optimization and machine learning processes are presented. • Integration of simulation and design exploration tools and methods is discussed. • Examples from practice, challenges and opportunities are presented and discussed. [ABSTRACT FROM AUTHOR]

Guided image denoising recovers clean target images by fusing guidance images and noisy target images. Several deep neural networks have been designed for this task, but they are black-box methods lacking interpretability. To overcome the issue, this paper builds a more interpretable network. To start with, an observation model is proposed to account for modality gap between target and guidance images. Then, this paper formulates a deep prior regularized optimization problem, and solves it by alternating direction method of multipliers (ADMM) algorithm. The update rules are generalized to design the network architecture. Extensive experiments conducted on FAIP and RNS datasets manifest that the novel network outperforms several state-of-the-art and benchmark methods regarding both evaluation metrics and visual inspection. • This is the first paper to raise the concept of modality gap. • This paper profoundly studies the statistical distribution of modality gap. • The proposed networks outperform the SOTA methods. [ABSTRACT FROM AUTHOR]

In recent years, many image denoising methods have been proposed based on convolutional neural networks (CNNs). While these methods have shown continuous performance improvement by introducing various mechanisms and structures, their computational cost tends to become increasingly expensive, owing to the resulting complex network architectures. This paper aims at winning the trade-off between computational efficiency and denoising performance for CNN-based image denoisers. Towards this end, we draw inspirations from traditional variational models with wavelet analysis operators for CNN architecture design. A model-inspired CNN is proposed with four key modules: iterative encoding-decoding units inspired by the iterative denoising process, directional convolutions inspired by the separable wavelet filters, inception modules inspired by the multi-scale analysis of wavelets, and stage-wise connections inspired by the adding noise back operation. In experiments, our CNN shows high computational efficiency in both training and test, with competitive results to state-of-the-art approaches. • A neural network inspired by the key ingredients of wavelet analysis model. • A sequential encoding-decoding network for image denoising. • A deep image denoiser well balancing effectiveness and computational efficiency. [ABSTRACT FROM AUTHOR]

This paper launches a thorough discussion on the locality of local neural operator (LNO), which is the core that enables LNO great flexibility on varied computational domains in solving transient partial differential equations (PDEs). We investigate the locality of LNO by looking into its receptive field and receptive range, carrying a main concern about how the locality acts in LNO training and applications. In a large group of LNO training experiments for learning fluid dynamics, it is found that an initial receptive range compatible with the learning task is crucial for LNO to perform well. On the one hand, an over-small receptive range is fatal and usually leads LNO to numerical oscillation; on the other hand, an over-large receptive range hinders LNO from achieving the best accuracy. We deem rules found in this paper general when applying LNO to learn and solve transient PDEs in diverse fields. Practical examples of applying the pre-trained LNOs in flow prediction are presented to confirm the findings further. Overall, with the architecture properly designed with a compatible receptive range, the pre-trained LNO shows commendable accuracy and efficiency in solving practical cases. [ABSTRACT FROM AUTHOR]

The widespread prevalence of cardiovascular diseases (CVDs) mandates meticulous and continuous monitoring for effective management and treatment. Wearable technologies have garnered substantial attention due to their seamless integration with bodily movements and biological systems. Researchers are actively exploring wearable technology from multidimensional angles, encompassing materials, design, and bioelectronics, to enhance CVD detection with greater sophistication and comfort. Enduring challenges, notably those surrounding material selection, persist, encompassing biocompatibility, conductivity, sensitivity, accuracy, and flexibility. Addressing these challenges is pivotal for adequate progress in wearable devices across many applications. Here, our review highlights the advancements in developing novel materials tailored for wearable technologies to detect cardiovascular diseases. The paper explicitly accentuates potential materials, architectural designs, operative mechanisms, and recent breakthroughs in flexible wearable sensors for CVD detection. The discussion explores diverse sensing mechanisms to monitor vital cardiac indicators, including piezoelectric, piezoresistive, capacitive, and triboelectric modalities. Furthermore, the paper provides a consolidated overview of contemporary efforts by different research teams in pulse wave sensors, heart sound sensors, ultrasound sensors, wearable ECG electrodes, and electro-biochemical sensors. We envision that the comprehensive analysis and juxtaposition of these distinct sensing mechanisms provide a more nuanced comprehension of their potential applications, constraints, and performance attributes within the wearable CVD health monitoring device framework. [Display omitted] • Underscores the worldwide prevalence of cardiovascular diseases, emphasizing the need for continuous monitoring for rational management and treatment. • Elaborate on how researchers are actively exploring wearable technology from bench materials to commercialization for CVD monitroing and diagnosis. • Discuss the persistent challenges of wearable technology, particularly concerning material selection, sensing mechanisms and CVDs Applications. • Emphasize the advancements in novel materials, architectural designs, operative mechanisms, CVD Parameters and wearable sensors for CVD detection. • Discuss the ongoing challenges of translating wearable technologies into commercially viable devices and their current FDA status. [ABSTRACT FROM AUTHOR]

• A new design method to obtain more flexible and diversified forms of a chalice-shaped column controlled by multivariable parameters including varied sectional thickness which can be directly applied to single column structure and multi-column structures in architectural design;. • SubD modeling technology after mesh editing is adopted to solve the problem of surface smooth transition between column and slab in early form-finding stage;. • Horizontal loading together with other natural forces are taken into consideration;. • Multi-objective optimization with an integrated final solution selection strategy to maintain balance between structural efficiency and form elegance which is specially needed for architectural aesthetics. This study proposes a design method to generate multivariable control chalice-shaped columns based on multi-objective optimization leveraging SubD modelling technology to smooth the geometric transition between slab and column. Taking horizontal force into consideration, the paper introduces total material usage as both an optimization objective and an important basis for final solution selection to gain a more elegant and dynamic form needed in architectural design. The paper applies the method to examples of both single-column and multi-column structures. The outcome is chalice-shaped columns balancing structural efficiency and aesthetic requirements. [ABSTRACT FROM AUTHOR]

This paper presents the embedded systems along with the first in-orbit results of FloripaSat-I, a complete CubeSat mission developed at the Federal University of Santa Catarina, Brazil. The satellite's service module has three main subsystems, each with an embedded processor and its peripherals. This paper presents an overview of the nanosatellite architecture designed for the mission, including the embedded systems, functionalities, and components. Launched in December 2019 atop a Chinese Long March 4B rocket, FloripaSat-I has been since then transmitting mission data, which is a metric used to validate its open-source service platform through flight heritage. The results collected in orbit are presented and discussed, as well as the student's perception about the challenges in the project. With the analysis of the results, it is possible to identify technical points to be improved by external groups with interest to reuse the open-source service platform for the next missions, and the impact that this teaching–learning methodology has in the students. • We present the FloripaSat-I embedded systems. • Flight results regarding the first three months are presented. • We validate the FloripaSat-I service platform with flight inheritance. [ABSTRACT FROM AUTHOR]

In this paper we show how simple text-driven variations of given statements in mathematics can lead to interesting new problems and push forward a whole theory around simple initial questions. We exemplify this in two cases. Case 1 deals with problem-posing activities suitable for pupils and case 2 is a rational reconstruction of the organisation of mathematical knowledge within problems of graph colorings. Mathematicians learn to systematically look for subsequent problems around a given problem. We argue that this toy-model captures a nontrivial part of professional mathematical research within the pure fields and conjecture that it even grasps high level developments in mathematics. By doing this, we implicitly encourage a very simplistic view on criteria, so to speak a "cowpath" approach to progress in mathematics. The term "cowpath" is borrowed from architecture and software design, where it is commonly used. While we can contemplate which pathways are ideal, we may also just plant grass and see where people choose to walk. Those pathways are also self-enforcing, since we are less hesitant to walk on those rather than criss-cross the landscape. [ABSTRACT FROM AUTHOR]

This paper concerns modeling, simulations and control design of turbo-electric distributed propulsion (TeDP) systems needed to power future hybrid aircraft systems. The approach taken is the one of control co-design by which the sizing and hardware selection of components and the TeDP architecture design are pursued so that potential effects of control and automation are accounted for from the very beginning. Unique to this approach is a multi-layered modular modeling and control approach in which technology-specific modules comprising the complex dynamical system are characterized using unified interaction variables at their interfaces with the rest of the system. The dynamical performance of the interconnected system is assessed using these technology-agnostic interface variable specifications and, as such, can be applied to any candidate architecture of interest. Importantly, even the inputs to the TeDP system coming from pilot commands are modeled using such interface variables. This new multi-layered modeling captures the dynamics of energy and power as interactions. It also has a rather straightforward physical interpretation. The paper builds on our earlier results introduced for terrestrial power systems, including small micro-grids. We show how system feasibility and stability can be checked in real-time operations by modules exchanging the information about their interaction variables and adjusting in a near-autonomous manner so that, as system conditions vary, the interconnected system still functions. No such systematic control co-design exists to the best of our knowledge, but it is needed as both new technologies and more complex, often conflicting performance objectives emerge. We illustrate the approach on a representative TeDP architecture and compare it to today's state-of-the-art. We close with a discussion on the generalization of the method for any given candidate architecture. Having such an approach dramatically reduces the R&D&D of novel candidate architectures. [ABSTRACT FROM AUTHOR]

In this paper we give an overview of both hardware and software architectures of real time systems used in devices for various purposes — from lab bench contraptions to cars. The goal of the paper is to reveal separate classes of such architectures as well as to define preconditions for choosing a particular architecture. [ABSTRACT FROM AUTHOR]

Deep neural networks have revolutionized many fields, including image processing, inverse problems, text mining and more recently, give very promising results in systems and control. Neural networks with hidden layers have a strong potential as an approximation framework of predictive control laws as they usually yield better approximation quality and smaller memory requirements than existing explicit (multi-parametric) approaches. In this paper, we first show that neural networks with HardTanh activation functions can exactly represent predictive control laws of linear time-invariant systems. We derive theoretical bounds on the minimum number of hidden layers and neurons that a HardTanh neural network should have to exactly represent a given predictive control law. The choice of HardTanh deep neural networks is particularly suited for linear predictive control laws as they usually require less hidden layers and neurons than deep neural networks with ReLU units for representing exactly continuous piecewise affine (or equivalently min–max) maps. In the second part of the paper we bring the physics of the model and standard optimization techniques into the architecture design, in order to eliminate the disadvantages of the black-box HardTanh learning. More specifically, we design trainable unfolded HardTanh deep architectures for learning linear predictive control laws based on two standard iterative optimization algorithms, i.e., projected gradient descent and accelerated projected gradient descent. We also study the performance of the proposed HardTanh type deep neural networks on a linear model predictive control application. [ABSTRACT FROM AUTHOR]

Although contemporary children's learning environments highlight promoting physical and social development–related play behaviours and safety, there are no valid means to analyse children's dynamic, complex behaviours. To address this limitation, the paper explores the impacts of agent-based simulation on architects' trade-offs in designing children's play-oriented learning environments. To simulate children's subtle behavioural responsiveness to the given environments, this paper adopts reinforcement learning (RL) as a method to develop autonomous play behaviours. A comparative experiment was conducted with 14 professional architects to investigate the capacities of the RL-powered agents. The systemic qualitative analysis indicates that the RL agent supported the coordination of complex physical constraints and new insights into child-oriented dimensions when evaluating the learning environment design. • RL agent is developed for designing children's play-oriented learning environments. • Capacities of RL agents-powered simulation is explored with professional architects. • Simulation facilitates calibration and refinement of design goals and strategies. • Simulation promotes optimisation of physical layouts via removing unused elements. • Simulation motivates experts and novices to design with child-centric perspective. [ABSTRACT FROM AUTHOR]

Sentence classification is an important task in natural language processing. The task makes use of deep networks to enclose a mass of features with different granularities in a sentence. However, the classification usually suffer from severe performance degradation when stacking a large number of networks. The main reason is that, in a deep architecture, the silent feature representations are easily weakened and mixed with noisy information, which is not effective in learning contextual features and constructing semantic dependencies in a sentence. In this paper, a deep penetration network (DPN) is designed to improve deep architectures' ability to preserve the favourable semantic features. The DPN enables salient features to penetrate through a deeper architecture and to construct long semantic dependencies between them. This approach is evaluated on seven public datasets. Our experiments show that the DPN exhibits a stable performance with deeper architectures. It improves the performance on three types of sentence classification tasks, outperforming the existing state-of-the-art models. • A DPN network is proposed for learning semantic features of a sentence. • The DPN enables salient features to penetrate through deeper architectures. • It also enables to construct semantic dependencies between abstract features. • A DPN-based deep architecture is designed to support sentence classification. • Our model achieves the SOTA performance on seven public datasets. [ABSTRACT FROM AUTHOR]

Construction is an indispensable part of people's life. The development of the construction industry is naturally overwhelming. In addition, the improvement of people's living standards has led to the rapid shortage of building resources and the increasingly serious energy problem. In fact, traditional buildings constructed by traditional building technologies are almost all high energy consumption, which has led to a sharp reduction of non renewable resources. People have also gradually found this important problem. Therefore, the concept of application of ecological energy saving technology in building reconstruction has been proposed. On this basis, this paper analyzes the energy conservation design of the external environment of the building, the energy conservation design of doors and windows, the energy conservation design of the building wall and the energy conservation design of the roof through mathematical analysis. The final results show that the energy-saving architectural design concept can really avoid the waste of non renewable resources and improve people's living experience. [ABSTRACT FROM AUTHOR]

Due to the variation in spatial resolution and the diversity of object scales, the interpretation of optical remote sensing images is extremely challenging. Deep learning has become the mainstream solution to interpret such complex scenes. However, the explosion of deep learning model architectures has resulted in the need for hundreds of millions of remote sensing images for which labels are very costly or often unavailable publicly. This paper provides an in-depth analysis of the main reasons for this data thirst, i.e., (i) limited representational power for model learning, and (ii) underutilization of unlabeled remote sensing data. To overcome the above difficulties, we present a scalable and adaptive self-supervised Transformer (AST) for optical remote sensing image interpretation. By performing masked image modeling in pre-training, the proposed AST releases the rich supervision signals in massive unlabeled remote sensing data and learns useful multi-scale semantics. Specifically, a cross-scale Transformer architecture is designed to collaboratively learn global dependencies and local details by introducing a pyramid structure, to facilitate multi-granular feature interactions and generate scale-invariant representations. Furthermore, a masking token strategy relying on correlation mapping is proposed to achieve adaptive masking of partial patches without affecting key structures, which enhances the understanding of visually important regions. Extensive experiments on various optical remote sensing interpretation tasks show that AST has good generalization capability and competitiveness. [ABSTRACT FROM AUTHOR]

The current paper analyzes the feasibility of a modular web application's migration procedure from on-premises to the cloud. Our focus is on identifying cost savings and options for hosting the application in the cloud. The research specifically examines the impact of architectural decisions records (ADR) on a modular monolith use case, utilizing.NET Core for the backend and Angular for the front end, with Clean Architecture as the design pattern. We investigate different cloud models (IaaS, PaaS, Serverless) considering project management's triple constraints (time, cost, performance). The study demonstrates that a modular monolith can be migrated with varying effort and costs depending on the chosen cloud model and technology stack. We explore optimizations such as resource utilization, licensing fees, and cost reduction through infrastructure reservations. Our findings show that the migration cost can range from a 20% increase with IaaS to approximately 70% cost reduction with the Serverless strategy compared to the on-premises environment, using equivalent resources. We also explore methods to lower expenses for each model, including resource modifications, Linux operating systems, and longer resource reservations. Considering limitations, we propose a two-stage migration strategy: initially lifting and shifting the application to IaaS with cost optimization, and subsequently migrating to PaaS for scalability and simplified resource management in the long term. [ABSTRACT FROM AUTHOR]

In this paper, we propose the strategy-following distributed attentional actor architecture after conditional attention (sfDA6-X) for multi-agent deep reinforcement learning (MADRL). The architecture is designed to provide controllability of coordinated behaviors in MADRL through the use of a saliency vector that captures conditions from the environment and rough, high-level instructions given by external experts such as the system designers or users. We introduce destination channels, which enable agents to be instructed to change their behaviors only by specifying the regions where individual agents should work. To validate the effectiveness of sfDA6-X, we conducted experiments in the object collection game and analyzed how agents change their coordinated and cooperative behaviors based on the given instructions. Our findings suggest that our approach provides a preliminary but promising solution for controlling coordinated behaviors in MADRL via arbitrary instructions represented in destination channels. [ABSTRACT FROM AUTHOR]

The utilization of physical models is a teaching strategy that has been compared to teaching based on sketching in design studios. This paper aims to determine the effectiveness of using physical models as a key instrument in first-year architecture design studios and as an alternative in cases of graphic sketching and 3D visualization deficiency. The paper adopts descriptive and analytical approaches in a specific case study, in which first-year architecture students were divided in the design studio among three different teaching strategies, namely: model-making, sketching and a blended strategy of both. Observations were validated on the basis of answers given by students to a brief questionnaire, investigating the extent to which they relied on models as supportive tools while designing and evaluating their projects. The results revealed that physical models play a significant role in the development of students' performance throughout the entire process of project design. [ABSTRACT FROM AUTHOR]

Vehicle re-identification (re-ID) has numerous applications in real life, such as video surveillance, information retrieval, and public security. However, it suffers from the misalignment of vehicles, which is a critical problem caused by inaccurate detection and different views. This paper proposes a novel network named Part Alignment Network (PAN) which can properly handle the misalignment of vehicles. It is therefore adapted to the vehicle re-ID task, avoiding the use of any additional pre-processing steps such as the annotation of vehicle key points and part segmentation by hand. In PAN, cross-correlation is adopted to the alignment of vehicle parts. Then, an effective network architecture is designed to extract the discriminative aligned features. By combining complementary aligned features and original features, more robust feature representations are learned. To show the effectiveness of PAN, this paper conducts experiments on three vehicle re-ID databases (VD1, VD2, and VehicleID), on which it improves the current state-of-the-art performance. [ABSTRACT FROM AUTHOR]

Traffic prediction is an important component in Intelligent Transportation Systems(ITSs) for enabling advanced transportation management and services to address worsening traffic congestion problems. The methodology for traffic prediction has evolved significantly over the past decades from simple statistical models to recent complex integration of different deep learning models. In this paper, we focus on evaluating recent hybrid deep learning models in the task of traffic prediction. To this end, we first conducted a review and taxonomize the reviewed models based on their feature extraction methods. We analyze their constituent modules and architectural designs. We select ten models representative of different architectural choices from our taxonomy and conducted a performance comparison study. For this, we reconstruct the selected models and performed a series of comparative experiments under identical conditions with three well-known real-world datasets collected from large-scale road networks. We discuss the findings and insights based on our results, highlighting the differences in the achieved prediction accuracy by models with different design decisions. • A specific review and taxonomy on the recent models for traffic prediction. • Architectural analysis of hybrid deep learning models based on our taxonomy. • A comparison study of representative models. • Experiments conducted on real-world datasets under identical conditions. [ABSTRACT FROM AUTHOR]

Cyber-physical oriented architectures are very recent approaches that aim at coupling physical system architectures with computer programs to facilitate the control and management of complex interactions. In the field of urban mobility, the context is marked by strong demographic growth and rapid urbanization of cities, making travel difficult. While mobility models (physical and informational) applied in industrialized countries help to control this problem while promoting continuous improvement, developing countries (DCs) are faced with a dilemma: directly transpose existing proven models (with the risk of poor adequacy to immediate local needs), or develop new models that may lack maturity. The objective of this paper is to propose a cyber-physical mobility architecture that combines the maturity of proven models with the strong suitability of a new approach to the developing country context. The aim is to set up a decision support system in the form of recommendations, based on the coupling between the definition of a physical transport system and the definition of the appropriate information system, according to a system engineering approach. [ABSTRACT FROM AUTHOR]

Advances in sensor technology, IOTs, communication technologies, computing, data management and analytics is changing the way modern global manufacturing enterprises are managed. With appropriate design and architecture of these technologies it is possible to use digital data streams to manage these enterprises in real time. This paper focuses on issues related to the design of these systems. Specifically, we focus on developing approaches for the design of autonomous services that receive data from local devices (e.g. sensors) in real time, do some processing in real time, communicate with other services, and respond to stimuli received from other services in real time. This allows localization of appropriate decision making. These services pass on relevant information to other related services in a web like architecture. [ABSTRACT FROM AUTHOR]

In this paper, we present FaceTuneGAN, a new 3D face model representation decomposing and encoding separately facial identity and facial expression. We propose a first adaptation of image-to-image translation networks, that have successfully been used in the 2D domain, to 3D face geometry. Leveraging recently released large face scan databases (FaceScape and CoMA), a neural network has been trained to decouple factors of variations with a better knowledge of the face, enabling facial expressions transfer and neutralization of expressive faces. Specifically, we design an adversarial architecture adapting the base architecture of FUNIT and using SpiralNet++ for our convolutional and sampling operations. Applied on these two datasets, FaceTuneGAN has a better identity decomposition and face neutralization than state-of-the-art techniques. It also outperforms classical deformation transfer approach by predicting blendshapes closer to ground-truth data and with less of undesired artifacts due to too different facial morphologies between source and target. [Display omitted] • 3D face model representation. • Morphology aware expression transfer. • Style based 3D facial generative model. • Semi-supervised learning of 3D identity and expression separation. [ABSTRACT FROM AUTHOR]

• Neural architecture search addressed with Genetic Programming and Backpropagation. • Genetic Programming efficiently provides blueprints for neural network architectures. • Backpropagation significantly improves the performance of candidate blueprints. • Proper fusion of hand-crafted saliency methods can outperform deep learning methods. • Proper fusion of deep learning methods outperforms the state of the art. Saliency detection methods proposed in the literature exploit different rationales, visual clues, and assumptions, but there is no single best saliency detection algorithm that is able to achieve good results on all the different benchmark datasets. In this paper we show that fusing different saliency detection algorithms together by exploiting neural network architectures makes it possible to obtain better results. Designing the best architecture for a given task is still an open problem since the existing techniques have some limits with respect to the problem formulation, to the search space, and require very high computational resources. To overcome these problems, in this paper we propose a three-step fusion approach. In the first step, genetic programming techniques are exploited to combine the outputs of existing saliency algorithms using a set of provided operations. Having a discrete search space allows us a fast generation of the candidate solutions. In the second step, the obtained solutions are converted into backbone Convolutional Neural Networks (CNNs) where operations are all implemented with differentiable functions, allowing an efficient optimization of the corresponding parameters (in a continuous space) by backpropagation. In the last step, to enrich the expressiveness of the initial architectures, the networks are further extended with additional operations on intermediate levels of the processing that are once again efficiently optimized through backpropagation. Extensive experimental evaluations show that the proposed saliency fusion approach outperforms the state-of-the-art on the MSRAB dataset and it is able to generalize to unseen data of different benchmark datasets. [ABSTRACT FROM AUTHOR]

This paper aims to investigate what is unique about videogame spaces and their construction, and how they might be utilised by architects. This design-led study establishes videogames as an architectural medium through practice-based research projects, examining game worlds as architectural spaces by developing design methods to analyze their environments. It then discusses developing games as architectural representations combining virtual space with computational rules. The paper's final study suggests how game structures might be applied in design approaches for ostensibly real sites. Through four case study projects, I outline a new form of architectural design practice operating between physical and virtual, reflecting the cultural influence of game worlds and exploring videogame environments as new ways to 'realise' architecture. • In videogame worlds player interactivity is integral to their architectural design. • Architecture designed within game spaces can impart meaning through interaction. • Videogame worlds can be sited within the history of architectural representation. • Analyzing videogames through architectural design methods expands existing games studies scholarship. • New forms of architectural practice will operate across the virtual and physical. [ABSTRACT FROM AUTHOR]

Thallium Bromide (TlBr) single crystal detector is widely recognized as an excellent semiconductor nuclear radiation detector. And its detection performance has already reached commercial level of the majority semiconductor detector. However, the long-term stability of TlBr detector always exhibited a serious polarization effect and the commercial application of TlBr detector is heavily limited. In this paper, the polarization phenomenon, polarization mechanism and the influence of polarization effect on detection performance are systematic investigated. As the metal-semiconductor interface plays an important role in detector performance, two typical metal-semiconductor-metal structure, Au/TlBr/Au and Ti/TlBr/Ti detectors are designed and fabricated in this paper. The XPS in-situ analysis results reveal that different oxidation/reduction reactions would happen at the metal-semiconductor interface of Au/TlBr/Au and Ti/TlBr/Ti detectors. The serious polarization effect in Au/TlBr/Au detectors resulted from the aggregated [AuBr 4 ]- ion at the Au/TlBr interface. Meanwhile, the relative weaken polarization effect in Ti/TlBr/Ti detectors is ascribed to the combined effects of oxidic titanium ions (Ti3+, Ti4+) and metallic Ti–Tl layer at the Ti/TlBr interface. Thus, the detector Ti/TlBr/Ti has a relative better energy resolution and long-term stability compared to detector Au/TlBr/Au. This paper systematic investigate the polarization mechanism and the influence of polarization effects on TlBr detector performance, and this will provide a potential solution to suppress the polarization problem by electrode architecture designing in the near future. • The polarization phenomenon, polarization mechanism and the influence on detection performance are systematic investigated. • Two typical MSM structure, Au/TlBr/Au and Ti/TlBr/Ti detectors are designed and fabricated in this work. • A XPS in-situ measure technology is used to study the reaction mechanism at the interface of Au/TlBr and Ti/TlBr. • The serious polarization effect in Au/TlBr/Au detectors resulted from the aggregated [AuBr 4 ]- ion at the Au/TlBr interface. • The polarization effect of Ti/TlBr/Ti detectors is ascribed to the combined effects of ions (Ti3+, Ti4+) and Ti-Tl layer. [ABSTRACT FROM AUTHOR]

• Travel demand generator for anywhere in Germany. • Open-source and requires only OpenStreetMap data as input. • Output is wholly disaggregated spatially and temporally. • Validated with household travel survey data. • Architecture designed for transferability to other regions. This paper introduces OMOD (OpenStreetMap Mobility Demand Generator), a new open-source activity-based mobility demand generation tool. OMOD uses a data-driven approach, calibrated with household travel survey data, to generate a population of agents with detailed daily activity schedules that state what activities each agent plans to conduct, where, and for how long. The temporal aspect of the output is wholly disaggregated, while the spatial aspect is given on the level of individual buildings. In contrast to other existing models, OMOD is freely available, open-source, works out-of-the-box, can be applied to anywhere in Germany with the ambition to widen the scope to other countries, and only requires freely available OpenStreetMap (OSM) data from the user. With OMOD, it is easy for non-experts to create realistic mobility demand, which can be used in transportation studies, energy system modeling, communications system research, et cetera. This paper describes OMOD's architecture and validates the model for three cities ranging from 200,000 to 2.5 million inhabitants. [ABSTRACT FROM AUTHOR]

· Exploration of optimal Trojan resistant design using area-latency tradeoff. · A novel methodology for functional Trojan-resistant hardware IP design. · Presents low-cost optimized Trojan-resistant design using PSO-DSE. · Attains Trojan resistance with minimal design cost overhead. Hardware Trojans that have the capability to change the computed functional output in intellectual property (IP) cores, integrated into computing systems can be a vital reliability concern in the context of correct system operation. Therefore, determining an optimal Trojan-resistant hardware design architecture that considers multi-objective orthogonal parameters such as area and delay is crucial. This paper presents a novel exploration of optimal hardware IP core design methodology with Trojan defense capability (i.e., detection and isolation) during high level synthesis (HLS) that provides isolation of functional Trojan in a system design to ensure reliable and correct functional behavior. The proposed methodology is robust and provides the capability to yield the correct output value using HLS-based triple modular redundancy (TMR) logic and a distinct multivendor allocation policy. Therefore, the proposed HLS methodology can generate an optimal hardware IP core/system-on-chip (SoC) design with functional Trojan defense capability. The paper presents an overall flow of the proposed methodology along with a demonstrative case study on designing optimal Trojan resistant finite impulse response filter (FIR) hardware SoC design. Results of the proposed approach are evaluated in terms of design cost, convergence time, security and optimality analysis, and comparison with prior works. The proposed approach is able to generate fully functional Trojan-resistant optimal SoC designs with minimum overhead, as evident from optimality analysis and design cost. [ABSTRACT FROM AUTHOR]

Extreme weather, chain failures, and other events have increased the probability of wide-area blackouts, which highlights the importance of rapidly and efficiently restoring the affected loads. This paper proposes a multi-agent reinforcement learning method for distribution system restoration. Firstly, considering that the topology of the distribution system may change during network reconfiguration, a dynamic agent network (DAN) architecture is designed to address the challenge of input dimensions changing in neural network. Two encoders are created to capture observations of the environment and other agents respectively, and an attention mechanism is used to aggregate an arbitrary-sized neighboring agent feature set. Then, considering the operation constraints of the DSR, an action mask mechanism is implemented to filter out invalid actions, ensuring the security of the strategy. Finally, an IEEE 123-node test system is used for validation, and the experimental results showed that the proposed algorithm can effectively assist agents in accomplishing collaborative DSR tasks. • A multi-agent reinforcement learning is used for distribution system restoration. • An action mask is used to filter out unsafe actions and improve learning performance. • A dynamic agent network adapts to varying input dimensions from topology changes. [ABSTRACT FROM AUTHOR]

Analysis of 3D textures, also known as relief patterns is a challenging task that requires separating repetitive surface patterns from the underlying global geometry. Existing works classify entire surfaces based on one or a few patterns by extracting ad-hoc statistical properties. Unfortunately, these methods are not suitable for objects with multiple geometric textures and perform poorly on more complex shapes. In this paper, we propose a neural network for binary segmentation to infer per-point labels based on the presence of surface relief patterns. We evaluated the proposed architecture on a high resolution point cloud dataset, surpassing the state-of-the-art, while maintaining memory and computation efficiency. [Display omitted] • A deep learning model for geometric texture segmentation on 3D surfaces. • Architecture design based on distinctive features of relief patterns. • Application of geodesic Voronoi diagrams to reduce memory usage during training. [ABSTRACT FROM AUTHOR]

Security is paramount in all mission-critical domains, including the aerospace industry. Cyber-attacks are increasing both in number and sophistication. Zero-trust is an emerging initiative that has proven very effective for enterprise systems in the Information Technology domain; however, research is lacking on applicable zero-trust mechanisms and their assurance for cyber–physical systems (CPS). We have already identified various zero-trust mechanisms in our previous work. In this paper, we present our zero-trust architecture design patterns and provide a methodology for the assurance of these mechanisms. Towards this objective, we have identified an initial set of assurance patterns covering individual zero-trust components in a system design. Our design and assurance patterns are made available to system engineers in pattern libraries. Engineers can model system architectures and utilize one or more of these patterns to provide design assurance based on individual zero-trust security requirements to improve the overall system cyber-security. To demonstrate our approach, we apply our assurance patterns to an unmanned aerial vehicle surveillance application. We discuss how our framework leverages the use of these patterns to develop zero-trust-enabled systems with different security requirements. Furthermore, our assurance patterns enable engineers to identify any design flaws and correct them during the initial system design phase, thus saving development time, effort, and cost. As a result, the overall approach can be utilized to design system models with specific zero-trust security requirements to improve the security posture of a CPS. [ABSTRACT FROM AUTHOR]

Machine learning assisted optical diagnostics could reduce conventional extensive experimental data collecting and post-processing costs and time in academic and industrial combustion measurements. In this paper, a novel Trident-Net (T-Net) architecture is designed and assisted for retrieval of high-fidelity soot temperature, volume fraction (SVF), and diameter fields simultaneously from soot radiation measurements in laminar sooting flames. Uniquely, the T-net is subtly fabricated in one branch of the encoder and three branches of decoders, which enables three adjustable cost functions and the corresponding decoder branch for soot three respective parameter fields. Contrasted with previous Back-propagation (BP) and U-net models, the T-Net is more efficient and achieves a higher entire score in terms of individual decoder manipulation. In addition, owing to the generalization performance improvement of multi-task learning, the T-Net model demonstrates decent prediction performance under few-shot learning. Thus, this proposed T-Net model showcases the great opportunities for real-time, in situ, monitoring of practical combustion pollutant emission, by embedded in a response strategy of the detection devices. • The T-Net model significantly changes the model methodolgy via the basic architecture. • Three decoder branches of the model are subtly designed for soot three parameter fields. • The decent performance of T-Net under extremely limited train data turns it into a priority. [ABSTRACT FROM AUTHOR]

Architects have increasingly turned to Machine learning techniques to streamline various aspects of the architectural design process. Although ML excels at discovering intricate patterns, predicting architectural design parameters through ML remains relatively unexplored. The complexity of the design process entails aspects that affect the form generation process. This paper investigated the feasibility of a framework that employed ML algorithms to predict numeric values for creating 3D models. A single villa was designed parametrically to generate hundreds of samples ensuring a human-centered design process. Four datasets were created from the samples for predicting form and windows parameters. Various regression and classification algorithms were applied, with ensemble learning methods demonstrating impressive performance across all datasets. Regression tasks achieved high R2 of up to 0.97, 0.79, and 0.99 while the best classification algorithm achieved 98 % accuracy. These findings underscored the potential efficacy of the framework in predicting architectural parameters, contingent upon well-designed datasets. [Display omitted] • Machine learning approaches can serve as human-centered in architectural design. • Creating a numeric data set of architectural parameters requires being thoughtfully designed with correlated variables. • Machine learning models can find the patterns and predict the building's parameters with well-developed data sets. • Ensemble learning models can give high accuracy and R2 scores in problems related to predicting architectural parameters. • Utilizing machine learning in the form generation phase can automate the design process giving precise parameters values. [ABSTRACT FROM AUTHOR]

Over the last half century, the main application of Brain–Computer Interfaces (BCIs) has been controlling wheelchairs and neuroprostheses, or generating text or commands for people with restricted mobility. There has been very limited attention in the field to applications for computer-aided design (CAD), especially in the built environment, despite the potential of BCIs to provide a new form of environmental interaction. This paper introduces Neuron, a BCI tool that enables designers with little experience in neuroscience or computer programming to: (1) gain access to neurological data, along with established metrics relevant to design; (2) create BCI interaction prototypes, both with digital on-screen objects and physical devices; and (3) evaluate designs based on neurological information and record measurements for further analysis. After detailing the BCI tool's development, the article showcases its capabilities through two case studies within the architectural design and built environment, followed by an evaluation of the tool's performance and a discussion of its implications, limitations, and areas for future improvement. • A BCI software tool was developed for EEG data acquisition, signal processing, feature extraction, and visualization. • NEURON enables brain-controlled generative design process • NEURON enables BCI prototyping for intelligence control system with visual programming • NEURON provides an accessible platform for integrating BCI into design, promoting interdisciplinary research and practical applications [ABSTRACT FROM AUTHOR]

Long-term and effective detection of epileptic seizures is a crucial aspect of epilepsy monitoring and treatment. Addressing the resource overhead issue of wearable epilepsy detection devices, this paper proposes a lightweight hardware implementation scheme for epilepsy detection based on a reusable architecture empirical mode decomposition (EMD) and K-Nearest Neighbors (KNN). Firstly, EMD is used to extract epileptic features from electroencephalogram (EEG), optimized through a reusable architecture design and sawtooth transform to reduce hardware resource usage. Subsequently, a KNN classifier with similarity judgment mechanism is designed to improve the recognition efficiency. Implemented on TSMC 65 nm process, the circuit area is 1.91 mm2, operates at 1 V and 20 MHz, with a power consumption of 4.034 mW. Evaluation on the Bonn EEG dataset yielded a classification accuracy of 96 %, sensitivity of 98 %, and a single detection delay of 1.51 ms. The hardware design offers a simple structure, high accuracy, and low resource consumption, making it suitable for wearable epilepsy detection devices. [ABSTRACT FROM AUTHOR]

Security-constrained unit commitment (SCUC) is a crucial procedure in power system planning and operation. As renewable resources are integrated, it is suggested to perform sub-hourly SCUC with a 15-minute interval. This change increases the computational burden due to more binary commitment variables. Despite the use of advanced MIP solvers, poor performance continues to be a challenge. Therefore, this paper proposes a feasible-enabled integer-variable warm-start strategy to provide feasible estimated starting values for MIP solvers before optimization. To achieve this objective, a data-driven model based on a deep neural network architecture is designed. This data-driven model takes into consideration the structural characteristics of input data, allowing it to predict the corresponding value of binary commitment variables effectively. Subsequently, an auxiliary optimization model is constructed by combining predicted values with the physical constraints of SCUC, ensuring estimated starting values are within the feasible region and mitigating the adverse effects of incorrect predicted values. Case studies conducted on two large-scale testing systems illustrate the effectiveness of the proposed method. • A strategy for estimating the starting value of binary variables in SCUC is proposed. • A neural network model is proposed to predict binary commitment variables. • An auxiliary model is proposed to enhance the feasibility of predicted values. • The accuracy and efficiency are demonstrated via numerical simulations. [ABSTRACT FROM AUTHOR]

Learning building information modeling (BIM) systems has always been a challenge for BIM adoption. Although groundbreaking performances of large language models (LLMs) have inspired many researchers to consider an LLM as a potential BIM control method using natural language, a specific method of utilizing LLMs for automated BIM model detailing has not yet been proposed. This paper proposes an LLM-BIM chaining framework to enable architectural design detailing using natural language, instead of using menu-based user interfaces, named "Natural-language-based Architectural Detailing through Interaction with AI (NADIA)". The NADIA framework is based on three main approaches: 1) separating the specification of the wall layers from the creation of the wall layers; 2) appropriate instruction prompting to guide the LLM to minimize irrational responses and produce engineering rational details; and 3) LLM-BIM chaining to seamlessly link a BIM authoring tool and an LLM. The effectiveness of NADIA was validated based on two main aspects: its accuracy in generating details that adhere to specified design requirements from users—as a design assistant—and its compliance with general engineering requirements—as a design consultant. The validation was achieved through tasks that involved generating 240 and 1,920 exterior wall details, respectively. NADIA achieved an average accuracy of 83.33% in generating logically coherent details in line with the required design conditions. For thermal performance requirements, it demonstrated a mean accuracy of 98.54% in complying with the American Society of Heating, Refrigerating, and Air-Conditioning Engineers (ASHRAE) 20.1–2019 standard. Despite being in its early stages, NADIA's potential for developing and refining architectural details through natural language-based interactions between architects and machines is promising. [ABSTRACT FROM AUTHOR]

Algorithmic Design (AD) tools enable the creation of geometrically complex architectural shapes that might be challenging to manufacture. This paper presents an overview of recent design-to-fabrication processes based on AD. It reflects on how AD can help overcome fabrication limitations and enhance the connection between architectural geometry, material, and manufacturing, approximating design exploration and construction. After analysing the literature on AD-based design-to-fabrication processes, the research identifies methodological trends, showing that AD is often used to increase (1) design flexibility, (2) tool interoperability, and (3) control over design manufacturing. It also reveals that custom and standard algorithms are equally used at early-stage design tasks, but that custom algorithms are markedly more used at later stages, where the amount of data and level of detail are higher. The research concludes with a critical reflection on the still-existing design-to-fabrication barriers and proposes future research directions. • Literature review on recent design-to-fabrication processes relying on Algorithmic Design (AD). • Analysis of the impact of AD in closing the gap between design exploration and fabrication. • Identification of methodological trends in AD to increase design flexibility, tool interoperability, and manufacturing control. • Critical reflection on the existing design-to-fabrication barriers and proposal of future research directions. [ABSTRACT FROM AUTHOR]

[Display omitted] • We explore a novel multistage approach that can generate high quality images. • we design a cross-stage fusion module for more effective information fusion. • We upsample the features with bilinear interpolation and deformable convolution. • Qualitative and quantitative evaluations demonstrate the superiority of our method. In this paper, we propose a multistage fusion network to solve the infrared and visible image fusion (IVIF) problem. Unlike other deep learning methods, our network architecture is designed to the complex balance between spatial details and contextualized information. In order to optimally balance these competing goals, our main proposal is a multistage architecture that progressively learns IVIF functions for the source images (infrared and visible images), thereby breaking down the overall recovery process into more manageable steps. Specifically, our model first learns the contextural features using the encoder-decoder architecture with downsampling operations and later combines them with a full-resolution branch that retains local details. Between stages, we introduce a cross-stage fusion module (CSFM) to propagate multiscale contextual features from an earlier stage to a later stage. In addition, we introduce an upsampling module that can conquer both checkerboard artifacts and blurring effect by a bilinear interpolation operation followed by a deformable convolution. The resulting tightly interlinked multistage fusion network, named MSFNet, demonstrates the superiority of our method over state-of-the-art performance on publicly available datasets. [ABSTRACT FROM AUTHOR]