Prathyush P. Poduval, Hamza Errahmouni Barkam, Xiangjian Liu
et al.
Hyperdimensional Computing (HDC) is a neurally inspired computing paradigm that leverages lightweight, high-dimensional operations to emulate key brain functions. Recent advances in HDC have primarily targeted two domains: learning, where the goal is to extract and generalize patterns for tasks such as classification, and cognitive computation, which requires accurate information retrieval for human-like reasoning. Although state-of-the-art HDC methods achieve strong performance in both areas, they lack a principled understanding of the fundamentally different requirements imposed by learning vs. cognition. In particular, existing works provide limited guidance on designing encoding methods that generate optimal hyperdimensional representations for these distinct tasks. In this study, we proposed the first universal hyperdimensional encoding method that dynamically adapts to the needs of both learning and cognitive computation. Our approach is based on neural-symbolic techniques that assign random complex hypervectors to atomic bases (e.g., alphabet definitions) and then apply algebraic operations in the high-dimensional hyperspace to control the correlation structure among encoded data points. Through theoretical analysis, we show that learning tasks benefit from correlated representations to maximize memorization and generalization capacity, whereas cognitive tasks require orthogonal, highly separable representations to enable accurate decoding and reasoning. We further derived a separation metric that quantifies this trade-off and validated it empirically across image classification and decoding tasks. Our results demonstrate that tuning the encoder to increase correlation improves classification accuracy from 65% to 95%, while maximizing separation enhances decoding accuracy from 85% to 100%. These findings provide the first systematic framework for designing hyperdimensional encoders that unify learning and cognition under a single, theoretically grounded representation model.
Cloud-native microservices improve development velocity and elasticity, but they also create complex and dynamic service dependencies. Resource contention, queue buildup, and downstream slowdowns can propagate through call chains, amplifying end-to-end tail latency (e.g., p95/p99) and increasing Service Level Objective (SLO) violation risks. While many studies focus on post-hoc anomaly detection and root-cause analysis, industrial operations increasingly demand proactive capabilities, like predicting performance risks before a request finishes, issuing early warnings from partial trace prefixes, and producing actionable signals for mitigation. This mini-review synthesizes recent progress on trace-driven proactive SLO management. We summarize problem formulations and evaluation protocols for SLO violation and tail-quantile prediction, prefix early warning under precision constraints, and actionable intermediate outputs such as bottleneck candidate ranking and what-if estimation. We then survey modeling approaches spanning feature-based baselines, sequence models, graph neural networks, sequence-graph fusion, and multimodal/causal extensions, highlighting practical issues such as class imbalance, sampling-induced missing spans, and topology drift. Finally, we survey commonly used public benchmarks and traces, and discuss open challenges toward deployable, trustworthy proactive SLO management.
Abstract Most people who have considered the matter have concluded that machines cannot be moral agents. Responsibility for acting on the outputs of machines must always rest with a human being. A key problem for the ethical use of AI, then, is to ensure that it does not block the attribution of responsibility to humans or lead to individuals being unfairly held responsible for things over which they had no control. This is the “responsibility gap”. In this paper, we argue that the claim that machines cannot be held responsible for their actions has unacknowledged implications for the conditions under which the outputs of AI can serve as reasons for belief. Following Robert Brandom, we argue that, because the assertion of a claim is an action, moral agency is a necessary condition for the giving and evaluating of reasons in discourse. Thus, the same considerations that suggest that machines cannot be held responsible for their actions suggest that they cannot be held to account for the epistemic value — or lack of value — of their outputs. If there is a responsibility gap, there is also a “testimony gap.” An under-recognised problem with the use of AI, then, is to ensure that it does not block the attribution of testimony to human beings or lead to individuals being held responsible for claims that they have not asserted. More generally, the “assertions” of machines are only capable of serving as justifications for belief or action where one or more people accept responsibility for them.
BackgroundLarge language models (LLMs) have demonstrated impressive performance on medical licensing and diagnosis-related exams. However, comparative evaluations to optimize LLM performance and ability in the domain of comprehensive medication management (CMM) are lacking. The purpose of this evaluation was to test various LLMs performance optimization strategies and performance on critical care pharmacotherapy questions used in the assessment of Doctor of Pharmacy students.MethodsIn a comparative analysis using 219 multiple-choice pharmacotherapy questions, five LLMs (GPT-3.5, GPT-4, Claude 2, Llama2-7b and 2-13b) were evaluated. Each LLM was queried five times to evaluate the primary outcome of accuracy (i.e., correctness). Secondary outcomes included variance, the impact of prompt engineering techniques (e.g., chain-of-thought, CoT) and training of a customized GPT on performance, and comparison to third year doctor of pharmacy students on knowledge recall vs. knowledge application questions. Accuracy and variance were compared with student’s t-test to compare performance under different model settings.ResultsChatGPT-4 exhibited the highest accuracy (71.6%), while Llama2-13b had the lowest variance (0.070). All LLMs performed more accurately on knowledge recall vs. knowledge application questions (e.g., ChatGPT-4: 87% vs. 67%). When applied to ChatGPT-4, few-shot CoT across five runs improved accuracy (77.4% vs. 71.5%) with no effect on variance. Self-consistency and the custom-trained GPT demonstrated similar accuracy to ChatGPT-4 with few-shot CoT. Overall pharmacy student accuracy was 81%, compared to an optimal overall LLM accuracy of 73%. Comparing question types, six of the LLMs demonstrated equivalent or higher accuracy than pharmacy students on knowledge recall questions (e.g., self-consistency vs. students: 93% vs. 84%), but pharmacy students achieved higher accuracy than all LLMs on knowledge application questions (e.g., self-consistency vs. students: 68% vs. 80%).ConclusionChatGPT-4 was the most accurate LLM on critical care pharmacy questions and few-shot CoT improved accuracy the most. Average student accuracy was similar to LLMs overall, and higher on knowledge application questions. These findings support the need for future assessment of customized training for the type of output needed. Reliance on LLMs is only supported with recall-based questions.
Water resources from the Indus Basin sustain over 270 million people. However, water security in this region is threatened by climate change. This is especially the case for the upper Indus Basin, where most frozen water reserves are expected to decrease significantly by the end of the century, leaving rainfall as the main driver of river flow. However, future precipitation estimates from global climate models differ greatly for this region. To address this uncertainty, this paper explores the feasibility of using probabilistic machine learning to map large-scale circulation fields, better represented by global climate models, to local precipitation over the upper Indus Basin. More specifically, Gaussian processes are trained to predict monthly ERA5 precipitation data over a 15-year horizon. This paper also explores different Gaussian process model designs, including a non-stationary covariance function to learn complex spatial relationships in the data. Going forward, this approach could be used to make more accurate predictions from global climate model outputs and better assess the probability of future precipitation extremes.
Cultural tourism is important for preserving cultural history and giving visitors immersive experiences, but tailoring it to each visitor's needs is still a major problem. It offers a distinct method of improving cultural tourism by combining Virtual Reality (VR), Genetic Algorithm (GA), and individual customization. Premature convergence and inadequate population variety are addressed by the Dynamic variety-Enhanced Genetic Algorithm (DDE-GA), a variation of the conventional GA. DDE-GA improves the investigation of possible solutions by dynamically modifying selection pressure according to population diversity, it makes it particularly useful for tackling optimization problems that are complicated, multi-modal, and highly dimensional. Creating an immersive environment that enables visitors to experience cultural heritage in a manner that is entirely tailored to their preferences, interests, and schedules is the objective of virtual reality technology. By adjusting to these individual parameters, the algorithm cleverly optimizes tourist itineraries. The DDE-GA-powered VR system works better than current methods, according to experimental data, with improvements in reaction time (1.1 s), accuracy (98 %), precision (97 %), and modeling error (0.10). When compared to convolutional algorithms, the suggested approach specifically enhances accuracy and drastically lowers error. This invention assists not only in satisfying tourists with individualized experiences but also in popularizing and preserving cultural traditions via the use of modern technology. The research concludes that integrating DDE-GA with VR technology substantially enhances personalized cultural tourism by optimizing routes based on user-specific preferences. This approach yields notable improvements in accuracy, precision, and response time while minimizing modeling errors. Furthermore, it contributes to both enriching tourist experiences and advancing cultural heritage conservation through innovative technological applications.
Information technology, Electronic computers. Computer science
ABSTRACT High entropy alloys (HEAs) have gained attention as effective reinforcements for enhancing the properties of metal matrix composites (MMCs), thanks to their distinct properties in contrast to traditional reinforcement particles. In view of that, this study develops HEA‐reinforced aluminum matrix composites (AMCs) consolidated through the pulse electric current sintering (PECS) technique and examines how the HEA reinforcement influences the microstructural, tribological, and nanomechanical properties of these consolidated composites. Appropriate thermodynamic and phase identification equations were used to determine a suitable combination of elements for the development of the HEA reinforcement, and an optimized sintering process was used to achieve effective bonding within the matrix. The resulting composites exhibited enhanced densification, with Laves phase, BCC, and FCC HEA phases present. Furthermore, incorporating HEA reinforcement greatly improved the mechanical properties such as wear resistance, microhardness, and nanoindentation characteristics of the composites such that the composite with 10% HEA displayed about a 191% increase in microhardness, with a significantly lower average coefficient of friction (ACOF) and higher wear resistance as compared to the unreinforced aluminum matrix.
The paper proposes an innovative model able to predict the output signals of resistance and capacitance (RC) low-pass filters for machine-controlled systems. Specifically, the work is focused on the analysis of the parametric responses in the time- and frequency-domain of the filter output signals, by considering a white generic noise superimposed onto an input sinusoidal signal. The goal is to predict the filter output using a black-box model to support the denoising process by means of a double-stage RC filter. Artificial neural networks (ANNs) and random forest (RF) algorithms are compared to predict the output of noisy signals. The work is concluded by defining guidelines to correct the voltage output by knowing the predictions and by adding further RC elements correcting the distorted signals. The model is suitable for the implementation of Industry 5.0 Digital Twin (DT) networks applied to manufacturing processes.
With the advantages of high efficiency and convenience, shared bikes are rapidly gaining popularity in China, and become an important connection mode for rail transit stations, but it also brings many problems which affect urban transportation and environment. Based on multi-source big data such as Xiamen shared bikes and urban space, this article analyzes the spatial and temporal characteristics of shared bikes travel by Python and GIS, and explores their riding characteristics and the effects concerning land use, urban environment and other aspects in the station area. Shared bikes have the characteristics of short spatial and temporal distance utilization, which provides an important support for the connecting traffic of urban subway stations. The connection and utilization of shared bikes around the station are affected by various urban factors, which also reflect the stages and differences of the station development. By analyzing the characteristics of shared bikes trips, rail transit stations can be classified into four types, and the develop strategies should be taken according to the difference of station types, to improve the urban slow traffic system, promote the development level of rail transit stations.(共享单车凭借其高效便利的优势在全国各大城市快速普及，已成为轨道交通站点重要的接驳方式，但其供需失衡、乱停乱放等问题也给城市交通与环境带来挑战。利用厦门市共享单车及城市空间等多源大数据，通过Python、GIS解析共享单车出行的时空特征，探究轨道站域骑行特征及其来自土地利用、城市建成环境等方面的影响效应，结果表明：共享单车具有短时空距离利用的特征，为城市地铁站点的接驳交通提供了重要支撑；站点周边共享单车的接驳利用受城市多种因素影响，同时也反映了站点开发的阶段性和差异性；根据共享单车出行特征，轨道交通站点可归纳为四类，应根据站点的不同类型因站施策，完善城市慢行系统，促进共享单车骑行，加强轨道交通站域开发。)
Jose Dixon, Oluwatunmise Akinniyi, Abeer Abdelhamid
et al.
The accurate classification of brain tumors is an important step for early intervention. Artificial intelligence (AI)-based diagnostic systems have been utilized in recent years to help automate the process and provide more objective and faster diagnosis. This work introduces an enhanced AI-based architecture for improved brain tumor classification. We introduce a hybrid architecture that integrates vision transformer (ViT) and deep neural networks to create an ensemble classifier, resulting in a more robust brain tumor classification framework. The analysis pipeline begins with preprocessing and data normalization, followed by extracting three types of MRI-derived information-rich features. The latter included higher-order texture and structural feature sets to harness the spatial interactions between image intensities, which were derived using Haralick features and local binary patterns. Additionally, local deeper features of the brain images are extracted using an optimized convolutional neural networks (CNN) architecture. Finally, ViT-derived features are also integrated due to their ability to handle dependencies across larger distances while being less sensitive to data augmentation. The extracted features are then weighted, fused, and fed to a machine learning classifier for the final classification of brain MRIs. The proposed weighted ensemble architecture has been evaluated on publicly available and locally collected brain MRIs of four classes using various metrics. The results showed that leveraging the benefits of individual components of the proposed architecture leads to improved performance using ablation studies.
In this paper, the analytical solution to the problem of heat transfer and fluid flow was obtained by using the quadruple Laplace transform method. Temperature distribution and fluid flow distribution were shown, temperature and fluid flow increase when the value of z increases, as well as the effect of the radiation parameter shown, it was concluded that the temperature increase with the increase in the value of the radiation coefficient . Matlab was used to plot the results.
Huimin Jiang, Xianhui Wu, Farzad Sabetzadeh
et al.
Abstract In online sales platforms, product design attributes influence consumer preferences, and consumer preferences also have a significant impact on future product design optimization and iteration. Online review data are the most intuitive feedback from consumers on products. Using the value of online review information to explore consumer preferences is the key to optimize the products, improve consumer satisfaction and meet consumer requirements. Therefore, the study of consumer preferences based on online reviews is of great importance. However, in previous research on consumer preferences based on online reviews, few studies have modeled consumer preferences. The models often suffer from the nonlinear structure and the fuzzy coefficients, making it challenging to build explicit models. Therefore, this study adopts a fuzzy regression approach with a nonlinear structure to model consumer preferences based on online reviews to provide reference and insight for subsequent studies. First, smartwatches were selected as the research object, and the sentiment scores of product reviews under different topics were obtained by text mining on the product online data. Second, a polynomial structure between product attributes and consumer preferences was generated to investigate the association between them further. Afterward, based on the existing polynomial structure, the fuzzy coefficients of each item in the structure were determined by the fuzzy regression approach. Finally, the mean relative error and mean systematic confidence of the fuzzy regression with nonlinear structure method were numerically calculated and compared with fuzzy least squares regression, fuzzy regression, adaptive neuro fuzzy inference system (ANFIS) and K-means-based ANFIS, and it was found that the proposed method was relatively more effective in modeling consumer preferences.
Electronic computers. Computer science, Information technology
Granular computing imitates human being's thinking.It shows great promise as a new way for data mining and know-ledge discovery in the context of big data.To solve the problem of knowledge acquisition in inconsistent generalized decision multi-scale ordered information systems,by employing evidence theory,the optimal scale combination and rule extraction in inconsistent generalized decision multi-scale ordered information systems are studied.Dominance relations are first introduced into decision multi-scale information systems,and some basic concepts in decision multi-scale ordered information systems are introduced.With reference to the notion of scale combinations in inconsistent generalized decision multi-scale ordered information systems,representations of information granules as well as lower and upper approximations of sets under different scale combinations are presented and their relationships are examined.With different scales of decisions,several types of optimal scale combinations in inconsistent generalized decision multi-scale ordered information systems are further defined and their relationships are clarified.Finally,a method of discernibility matrix attribute reduction and rule acquisition based on generalized dominance decision functions are explored.
Tradition wireless sensor networks (WSNs) transmit data by single or multiple hops. However, some sensor nodes (SNs) close to a static base station forward data more frequently than others, which results in the problem of energy holes and makes networks fragile. One promising solution is to use a mobile node as a mobile sink (MS), which is especially useful in energy-constrained networks. In these applications, the tour planning of MS is a key to guarantee the network performance. In this paper, a novel strategy is proposed to reduce the latency of mobile data gathering in a WSN while the routing strategies and tour planning of MS are jointly optimized. First, the issue of network coverage is discussed before the appropriate number of clusters being calculated. A dynamic clustering scheme is then developed where a virtual cluster center is defined as the MS sojourn for data collection. Afterwards, a tour planning of MS based on prediction is proposed subject to minimizing the traveling distance to collect data. The proposed method is simulated in a MATLAB platform to show the overall performance of the developed system. Furthermore, the physical tests on a test rig are also carried out where a small WSN based on an unmanned aerial vehicle (UAV) is developed in our laboratory. The test results validate the feasibility and effectiveness of the method proposed.
Vojdan Kjorveziroski, Sonja Filiposka, Vladimir Trajkovik
Serverless computing is a new concept allowing developers to focus on the core functionality of their code, while abstracting away the underlying infrastructure. Even though there are existing commercial serverless cloud providers and open-source solutions, dealing with the explosive growth of new Internet of Things (IoT) devices requires more efficient bandwidth utilization, reduced latency, and data preprocessing closer to the source, thus reducing the overall data volume and meeting privacy regulations. Moving serverless computing to the edge of the network is a topic that is actively being researched with the aim of solving these issues. This study presents a systematic mapping review of current progress made to this effect, analyzing work published between 1 January 2015 and 1 September 2021. Using a document selection methodology which emphasizes the quality of the papers obtained through querying several popular databases with relevant search terms, we have included 64 entries, which we then further categorized into eight main categories. Results show that there is an increasing interest in this area with rapid progress being made to solve the remaining open issues, which have also been summarized in this paper. Special attention is paid to open-source efforts, as well as open-access contributions.
With advancements in computer-aided design, simulation of internal combustion engines has become a vital tool for product development and design innovation. Among the simulation software packages currently available, MATLAB/Simulink is widely used for automotive system simulations, but does not contain a comprehensive engine modeling toolbox. To leverage MATLAB/Simulink’s capabilities, a Simulink-based 1D flow engine modeling framework has been developed. The framework allows engine component blocks to be connected in a physically representative manner in the Simulink environment, reducing model build time. Each component block, derived from physical laws, interacts with other blocks according to block connection. In this Part 1 of series papers, a comprehensive gas dynamics model is presented and integrated in the engine modeling framework based on MATLAB/Simulink. Then, the gas dynamics model is validated with commercial engine simulation software by conducting a simple 1D flow simulation.
Robust goal programming (RGP) is an emerging field of research in decision-making problems with multiple conflicting objectives and uncertain parameters. RGP combines robust optimization (RO) with variants of goal programming techniques to achieve stable and reliable goals for previously unspecified aspiration levels of the decision-maker. The RGP model proposed in Kuchta (2004) and recently advanced in Hanks, Weir, and Lunday (2017) uses classical robust methods. The drawback of these methods is that they can produce optimal values far from the optimal value of the “nominal” problem. As a proposal for overcoming the aforementioned drawback, we propose light RGP models generalized for the budget of uncertainty and ellipsoidal uncertainty sets in the framework discussed in Schöbel (2014) and compare them with the previous RGP models. Conclusions regarding the use of different uncertainty sets for the light RGP are made. Most importantly, we discuss that the total goal deviations of the decision-maker are very much dependent on the threshold set rather than the type of uncertainty set used.
A new Wolter x-ray imager has been developed for the Z machine to study the emission of warm (>15 keV) x-ray sources. A Wolter optic has been adapted from observational astronomy and medical imaging, which uses curved x-ray mirrors to form a 2D image of a source with 5 × 5 × 5 mm3 field-of-view and measured 60-300-μm resolution on-axis. The mirrors consist of a multilayer that create a narrow bandpass around the Mo Kα lines at 17.5 keV. We provide an overview of the instrument design and measured imaging performance. In addition, we present the first data from the instrument of a Mo wire array z-pinch on the Z machine, demonstrating improvements in spatial resolution and a 350-4100× increase in the signal over previous pinhole imaging techniques.
We recently developed a one-dimensional imager of neutrons on the Z facility. The instrument is designed for Magnetized Liner Inertial Fusion (MagLIF) experiments, which produce D-D neutrons yields of ∼3 × 1012. X-ray imaging indicates that the MagLIF stagnation region is a 10-mm long, ∼100-μm diameter column. The small radial extents and present yields precluded useful radial resolution, so a one-dimensional imager was developed. The imaging component is a 100-mm thick tungsten slit; a rolled-edge slit limits variations in the acceptance angle along the source. CR39 was chosen as a detector due to its negligible sensitivity to the bright x-ray environment in Z. A layer of high density poly-ethylene is used to enhance the sensitivity of CR39. We present data from fielding the instrument on Z, demonstrating reliable imaging and track densities consistent with diagnosed yields. For yields ∼3 × 1012, we obtain resolutions of ∼500 μm.