Distributed Clustering Research Articles

FUZZY MODEL OF ELECTRICITY CONTROL WITH WIRELESS INFORMATION PROCESSED ON GPU

Methods of processing information transmitted through wireless networks with software development are investigated in the work. Innovative methods of data transmission such as optical technologies, quantum data transmission and wireless data transmission technologies are disclosed. It is noted that in the modern understanding, the concept of distributed computing defines the process of convergence (convergence) of distributed processing methods, such as GRID, cloud and fog computing, with the combination of virtual cluster systems (grid clusters, cloud clusters and fog clusters) into a single information communication and computing system . It is emphasized that, unlike cellular modems, ZigBee technology nodes have microcontrollers with a pre-installed operating system and flash memory, which allows solving simple computational tasks in real time before sending data. It is advisable to solve such tasks within the framework of a multi-agent approach, which will increase the efficiency of the use of sensor nodes and the entire sensor network. The advantages of the multi-agent technology of fog computing based on sensor nodes of the wireless network of the ZigBee standard are revealed. The method of multi-agent processing of sensory information and its main components are described. The architecture of the system of distributed sensor data processing is outlined, which includes 4 hardware and software levels: Terminal sensor nodes and controllers of measuring devices and automation devices that implement fuzzy calculations; Coordinators, sensor segment routers and cellular modems that collect, protect and transmit sensor data to the processing center; A data processing center that includes a cluster of servers for GRID calculations and a cloud data storage server; Client devices to access cloud storage, computing cluster servers, and distributed fog computing terminals. It is emphasized that indicators and forecast results can be stored on distributed sensor nodes or transmitted for accumulation in cloud storage for further extraction and intelligent processing in the GRID cluster of the data center.

GenesisRM: A state-driven approach to resource management for distributed JVM web applications

Reducing resource waste while maintaining end-to-end latency service-level objective (SLO) by simultaneously managing CPU bandwidth, memory allocation, and pod number of web applications running on Java virtual machine (JVM) is challenging. The challenges stem from the complexity of the multi-type resource allocation optimization problem, the high sensitivity of JVM performance to resource scaling actions, and the lack of low-level resource scaling mechanisms. We present GenesisRM, a resource management framework with a novel state-driven architecture. Specifically, we design a state control model for JVM web applications that encompasses seven pod states. This model serves as an abstraction layer, decoupling the centralized resource management system into a global state manager and distributed pod managers. The state manager controls the state transitions of the pods based on the overall workload, while the pod managers dynamically allocate resources for each pod according to the state and local workload.Then, we propose a multi-frequency control model with two predictive state controllers and a reactive state controller to manage the state of pods based on the state control model. In addition, GenesisRM brings new mechanisms to scale JVM pods’ low-level resources without damaging their performance. We evaluate our work using a real-world JVM web application benchmark in three different scale server clusters of Pengcheng Laboratory Developer Cloud, and the 21-day experimental results show that GenesisRM saves 31.70% CPU and 17.60% memory compared to the best-performing state-of-the-art solutions while guaranteeing the SLO imposed on end-to-end latency.

Distributed enhanced multi-objective evolutionary algorithm based on decomposition for cluster analysis in wireless sensor network

Conventional clustering algorithms do not recognize patterns and structures with contradicting objectives in large, distributed datasets. Distributed clustering leverages rapid processing capabilities to allow multiple nodes to work together. This paper proposes a Distributed clustering based on Multiobjective Evolutionary Algorithm by Decomposition (D-MOEA/d) to solve various multiobjective optimization problems in wireless sensor networks (WSNs). In MOEA/d, a multiobjective optimization problem decomposes into several scalar optimization subproblems, each focusing on a distinct objective. Each subproblem is expressed as a clustering problem that uses local data to perform distributed clustering. The proposed method has been extended to achieve improved accuracy in less time by using a smaller feature subset with less redundancy. The Distributed Enhanced MOEA/d (DE-MOEA/d) avoids local optima by achieving diversity in the population using fuzzy-based nearest neighbor selection, sparse population initialization, and evolved mutation operator. This integration improves the accuracy of the clustering process at WSN nodes, ensuring the attainment of well-balanced solutions across multiple optimization criteria in the distributed environment. Average Euclidean and total symmetrical deviations are the two cost functions used to minimize while clustering on the MOEA/d framework. Six real-life WSN datasets are used to assess the performance of the proposed technique: (1) the Delhi air pollution dataset, (2) the Canada weather station dataset, (3) the Thames River water quality dataset, (4) the Narragansett Bay water quality dataset, (5) the Cook Agricultural land dataset and 6) Gordon Soil dataset. The simulation results of both proposed algorithms are compared with Multiobjective distributed particle swarm optimization (DMOPSO) and Distributed K-means (DK-Means). The proposed algorithm DE-MOEA/d performs better in terms of the Silhouette index (SI), Dunn index (DI), Davies–Bouldin index (DBI), and Kruskal–Wallis (KW) statistical test.

Neuro‐fuzzy‐based cluster formation scheme for energy‐efficient data routing in IOT‐enabled WSN

SummaryInternet of things–enabled wireless sensor networks face challenges like inflexibility, poor scalability, suboptimal cluster head selection, and energy inefficiencies. This is due to the faster data transmission rates between cluster nodes during data packet routing. This creates unnecessary energy consumption burdens for those actively transmitting nodes. Conceptually, an effective cluster formation phase supports better data routing mechanisms, while sustaining the energy efficiency of individual nodes. This paper proposes a Neuro‐Fuzzy based Cluster Formation (NFCF) scheme to facilitate adaptive and energy‐efficient cluster topologies. NFCF utilizes fuzzy logic and neural networks to identify optimal super nodes for flexible cluster formations. This approach enables configurable cluster sizes along with inclusion/exclusion criteria for member nodes based on energy thresholds. Parameters evaluated for node selection include the degree of super node, expected energy per cluster, energy variance, and residual energy. Nodes not meeting the thresholds are excluded. The neural network updates fuzzy rules to guide optimal clustering decisions based on anticipated energy dynamics under different conditions. The performance of the proposed NFCF scheme is evaluated based on objective function changes related to data transmission, individual node energy variation, energy variance before and after transmissions, and averaged end‐to‐end delay across transmission cycles. Results are compared against genetic fuzzy clustering, fuzzy energy‐aware clustering, fuzzy‐based distributed clustering, fuzzy logic‐based multi‐hop clustering, and fuzzy weighted k‐means clustering.

MSCFS-RP: A Colored-Petri-Net-Based Analysis Model for Master–Slave Cloud File Systems with Replication Pipelining

As a typical information system, a cloud file system enables the storage, retrieval, and management of data on remote servers or server clusters. The reliable design of such systems is critical to ensure the security of data and availability of services. However, designing correct-by-construction systems is challenging due to the complexity of and concurrency inherent in cloud file systems. Further, existing works on cloud file system analysis often focus on specific systems or lack formal modeling and verification, leading to potential design flaws and security vulnerabilities. To address these issues, we propose MSCFS-RP, which is a formal analysis model based on colored Petri nets. Leveraging the strengths of colored Petri nets in representing diverse information types with colored tokens and defining explicit rules for concurrent interactions, our model captures the writing and reading processes of clients, meta servers, and clusters. With strong formalism and support for verification using CPN Tools, we rigorously evaluate key properties such as replication consistency under various scenarios. The results demonstrate that MSCFS-RP satisfies these properties, validating its effectiveness and trustworthiness in managing information within cloud storage systems.

Enhancing Big Data Performance Through Graph Coloring-Based Locality of Reference

Efficiency is a crucial factor when handling the retrieval and storage of data from vast amounts of records in a Big Data repository. These systems require a subset of data that can be accommodated within the combined physical memory of a cluster of servers. It becomes impractical to analyze all of the data if its size exceeds the available memory capacity. Retrieving data from virtual storage, primarily hard disks, is significantly slower compared to accessing data from main memory, resulting in increased access time and diminished performance. To address this, a proposed model aims to enhance performance by identifying the most suitable data locality structure within a big data set and reorganizing the data schema accordingly; by locality, it has been referred to as a particular access pattern. This allows transactions to be executed on data residing in the fastest memory layer, such as cache, main memory, or disk cache

Anomaly detection for key performance indicators by fusing self-supervised spatio-temporal graph attention networks

With the development of Artificial Intelligence for IT Operations (AIOps), numerous software and services are monitored by Key Performance Indicators (KPIs) collection components. Multivariate KPIs, as a type of time series data, are essential for effective management of the entity's service quality. In recent years, deep learning methods have made great improvements in the anomaly detection of multivariate time series; however, existing methods have not fully considered how to explicitly capture the correlation between multivariate time series in the feature dimension and temporal dimension, resulting in inevitable abnormal false positives. Therefore, this paper proposes a self-supervised multivariate KPIs anomaly detection method MAD-STA that combines graph structure learning and spatio-temporal GAT (Graph Attention Network). In the feature dimension, MAD-STA introduces a node embedding mechanism for graph structure learning and then uses the feature-oriented GAT layer to compute the graph attention coefficient to obtain the correlation between different KPIs. In the temporal dimension, MAD-STA uses the time-oriented GAT layer to compute attention weights between correlated timestamps, and the GRU-based VAE encoder captures long-term dependence to extract more comprehensive temporal feature representations. Finally, MAD-STA uses GRU-based VAE decoder to reconstruct the captured high-level features and achieves efficient anomaly detection and localization by calculating the anomaly score of multiple KPIs. Compared with the baseline methods on multiple data sets, the experimental results show that the anomaly detection accuracy of MAD-STA is better than that of the baseline method. Especially on the KPI data sets of the two server clusters of SMD and CKM, MAD-STA improves the performance and the F1 comprehensive index compared with the best baseline method. In addition, MAD-STA performs well on anomaly false positive rate and has excellent interpretability, which can be used to assist anomaly diagnosis and root cause index analysis.

Abstract PO-028: A high-throughput bone marrow 3D co-culture system to develop resistance to targeted agents in marginal zone lymphoma

Abstract Background. Small molecules targeting the B-cell receptor (BCR) signaling, including PI3K and BTK inhibitors, have represented a step forward for managing lymphoma patients. However, resistance often develops, negatively impacting the benefit of patients exposed to drugs. Thus, developing preclinical tools to identify modalities to overcome this phenomenon is important. The bone marrow is commonly involved in marginal zone lymphoma (MZL) patients, and it might represent a safe environment where lymphoma cells escape therapy. Bone marrow stromal cells (BMSCs) induce proliferation and drug resistance in several tumor entities, mainly through the release of cytokines. To study potential resistance mechanisms to BCR targeting agents, we aimed to develop a 3D co-culture system in which primary BMSCs were used to model the bone marrow tumor microenvironment while exposing an MZL cell line to the PI3K inhibitor copanlisib. Methods. The MZL cell line VL51 was cultured in 3D fibrin gel (10mg/mL), either alone or with primary BMSCs, in 96-well plates using the OT-2 liquid handler (Opentrons) for high-throughput seeding purposes. Copanlisib treatment started on day 3. The effects of the drug were assessed through morphological observations and quantitative measurements of images taken on days 3, 5, and 7 in the presence or absence of BMSCs. On day 7, supernatants were analyzed for 105 secreted factors using the Proteome Profiler Human XL Cytokine Array Kit (R&D Systems / Bio-Techne). Results. We observed differences in VL51 growth pattern in the different conditions. Most importantly, the co-culture with BMSCs significantly (p=0.025) decreased VL51 sensitivity to copanlisib. When not treated, VL51 cells formed significantly larger clusters (p<0.001) in the presence of BMSCs. However, upon copanlisib treatment, VL51 cells formed smaller and more distributed clusters comparable in size to those in the mono-culture. Subsequent analyses of the supernatants from the mono- or co-cultures following DMSO or copanlisib treatment identified 14 differentially secreted cytokines among the different conditions. Validation experiments were done using recombinant proteins in 2D cultures for 6 factors. Among these, recombinant IGFBP-3, serpin E1, and PTX-3 significantly decreased VL51 sensitivity to copanlisib in 2D cultures (p=0.05, 0.01, and 0.04, respectively). Validation experiments are ongoing with additional in vitro models and using BTK inhibitors. Conclusions. We created a high-throughput 3D co-culture MZL system and found that BMSCs reduce sensitivity to copanlisib in MZL models. Secretion of IGFBP-3, Serpin E1, and PTX-3 reduced response to copanlisib. This 3D-model will be used to assess resistance mechanisms to clinically relevant compounds and to evaluate alternative therapeutic approaches to improve the treatment of lymphoma patients. Citation Format: Alex Zadro, Alberto Arribas, Maria Vittoria Colombo, Eleonora Cannas, Filippo Spriano, Luciano Cascione, Chiara Arrigoni, Christian Candrian, Matteo Moretti, Francesco Bertoni. A high-throughput bone marrow 3D co-culture system to develop resistance to targeted agents in marginal zone lymphoma [abstract]. In: Proceedings of the Fourth AACR International Meeting on Advances in Malignant Lymphoma: Maximizing the Basic-Translational Interface for Clinical Application; 2024 Jun 19-22; Philadelphia, PA. Philadelphia (PA): AACR; Blood Cancer Discov 2024;5(3_Suppl):Abstract nr PO-028.

Optimized intuitionistic fuzzy enriched honey badger algorithm for cloud network-based work load scheduling

The difficulty of scheduling jobs or workloads increases due to the stochastic and transient characteristics of the cloud network. As a key prerequisite for establishing QoS, it asserts that effective work scheduling must be developed and executed. Maximum profit is made possible for cloud service providers by proper resource management. The most effective scheduling algorithm considers resources given by providers rather than the task set that users have accumulated. This paper developed a model that works in a two-level hierarchical model comprising global scheduling and local schedules to handle the heterogeneous type of request in real-time. These two levels of scheduling communicate with each other to produce an optimal scheduling scheme. Initially, all the requests are passed to the global scheduler, whose task is to categorize the type of request and pass it to the corresponding queue for assigning it to the related local scheduler using a parabolic intuitionistic fuzzy scheduler. In this work, the heterogeneous types of files are handled by maintaining different queues, in which each queue handles only a specific type of file like text doc, audio, image and video. Once the type of req is initiated by the clients, the global scheduler identifies the type of request and passes it to their relevant queue. In the next level, the local scheduler is assigned to each type of web server cluster. Once the work request is dispatched from the global workload scheduler, it is allocated to the local queue of the local scheduler, which allocates the resources of web servers by adapting the Quantum Honey Badger Algorithm, which searches the best-suited server for completing the assigned work based on the available resource parameters.

Study of data replication process using Raft replication algorithm to maintain consistency in server cluster

The article considers the issues of ensuring availability and fault tolerance of a server cluster. The analysis of methods and technologies for increasing system reliability and improving performance, such as data replication, automatic switching between servers and fast recovery after failures, is carried out. It is indicated that the key element in modern distributed systems is replication-based clustering. The types of replication are analyzed. It is shown that during the operation of the system, there is a choice between the stability of the received data and the speed or scaling limitation. It is shown that the Raft algorithm is one of the reliable mechanisms with a high level of availability for achieving consensus and data management. An analysis of the literature on the use of the Raft consensus algorithm is carried out. The replication process using the Raft algorithm is analyzed. It is substantiated that it is possible to ensure strong consistency and high availability of the cluster using the management and control mechanism on the Kubernetes platform while maintaining cluster configuration synchronization. The Raft consensus process is studied using the M/M/s QS model. The exponential distribution of time between client requests and servicing with a certain number of servers is considered. The program code for the mathematical model in Python has been developed. Modeling has been performed in the Visual code environment in Python. The following parameters have been studied: the number of servers that operate at a certain average speed, the expected time of a client request in the system, and the probability of message delay. It has been proven that as the arrival speed increases, the expected time of a client request increases. The analysis provides an understanding of how the Raft algorithm works in different contexts and is applicable to optimizing system design processes.

Variate Associated Domain Adaptation for Unsupervised Multivariate Time Series Anomaly Detection

Multivariate Time Series Anomaly Detection (MTS-AD) is crucial for the effective management and maintenance of devices in complex systems such as server clusters, spacecrafts and financial systems etc. However, upgrade or cross-platform deployment of these devices will introduce the issue of cross-domain distribution shift, which leads to the prototypical problem of Domain Adaptation for MTS-AD. Compared with general domain adaptation problems, MTS-AD domain adaptation presents two peculiar challenges: 1) The dimensions of data from the source domain and the target domain are usually different, so alignment without losing any information is necessary. 2) The association between different variates plays a vital role in the MTS-AD task, which is overlooked by traditional domain adaptation approaches. Aiming at addressing the above issues, we propose a V ariate A ssociated domai N a D aptation method combined with a Gr A ph Deviation Network (abbreviated as VANDA ) for MTS-AD, which includes two major contributions. First, we characterize the intra-domain variate associations of the source domain by a graph deviation network (GDN), which can share parameters across domains without dimension alignment. Second, we propose a sliding similarity to measure the inter-domain variate associations and perform joint training by minimizing the optimal transport distance between source and target data for transferring variate associations across domains. VANDA achieves domain adaptation by transferring both variate associations and GDN parameters from the source domain to the target domain. We construct two pairs of MTS-AD datasets from existing MTS-AD data and combine three domain adaptation strategies with six MTS-AD backbones as the benchmark methods for experimental evaluation and comparison. Extensive experiments demonstrate the effectiveness of our approach, which outperforms the benchmark methods, and significantly improves the AD performance of the target domain by effectively utilizing the source domain knowledge.

Memory and Communication Efficient Federated Kernel k-Means.

A federated kernel k -means (FedKKM) algorithm is developed in this article to conduct distributed clustering with low memory consumption on user devices. In FedKKM, a federated eigenvector approximation (FEA) algorithm is designed to iteratively determine the low-dimensional approximate vectors of the transformed feature vectors, using only low-dimensional random feature vectors. To maintain high communication efficiency in each iteration of FEA, a communication-efficient Lanczos algorithm (CELA) is further designed in FEA to reduce the communication cost. Based on the low-dimensional approximate vectors, the clustering result is obtained by leveraging a distributed linear k -means algorithm. A theoretical analysis shows that: 1) FEA has a convergence rate of O(1/T) , where T is the number of iterations; 2) the scalability of FedKKM is not affected by the dataset size since the communication cost of FedKKM is independent of the number of users' data; and 3) FedKKM is a (1+ϵ) approximation algorithm. The experimental results show that FedKKM achieves the comparable clustering quality to that of a centralized kernel k -means. Compared with state-of-the-art schemes, FedKKM reduces the memory consumption on user devices by up to 94% and also reduces the communication cost by more than 40%.

Power prediction of regional distributed photovoltaic clusters with incomplete information based on improved weighted fusion and transfer learning

AbstractThe rapid expansion of low‐voltage distributed photovoltaic (PV) systems with decentralized layouts poses significant data collection challenges. The scarcity of data amplifies prediction complexities, affecting the operational security and stability of distribution networks. To enhance predictive accuracy for distributed regional PV power generation, including unmonitored low‐voltage systems, this paper proposes a novel prediction approach that combines weight optimization and transfer learning. Firstly, to more accurately extrapolate from neighbouring monitored PV to estimate unmonitored PV outputs, a novel SSA‐MLP‐XGBoost model is proposed for monitored PV power plants, in which the hyperparameters of multilayer perceptron (MLP) and (extreme gradient boosting) XGBoost models are fine‐tuned using the sparrow search algorithm (SSA), and the optimized models are synergistically integrated to heighten prediction precision for monitored plants. Secondly, based on the predicted power curves of monitored PV power plants, a weight optimization model is presented to optimize the combined weights between unmonitored and monitored power stations. Furthermore, the power generation model from monitored facilities is transplanted to unmonitored plants for power generation estimation. Ultimately, the authors combine the weight optimization and transfer learning model to get more accurate and robust model. Validation across three regional distributed photovoltaic clusters demonstrates a noteworthy improvement in prediction accuracy—20%, 7.5%, and 25% for the respective clusters compared to the existing methods.

MapReduce Algorithm on a Serverless Platform

Using the MapReduce (MR) programming technique, large-scale data processing tasks can be divided into tasks that are easier to manage and independent. The serverless implementation's performance is better than the non-serverless MR model when the model parameters are altered in the experiment. Optimizing the use of the MR model on the serverless platform can be achieved by taking into account the relationship between implementation efficiency and platform settings. Additionally, it can act as a source of inspiration for future serverless hardware support configurations. Also, the serverless platform demonstrates how it improves the effectiveness of resource utilization in machine learning training. The intended result can then be achieved by combining the results of these concurrently operating jobs on server clusters. This work adapts MR, a popular big data processing framework, to the serverless platform, emphasizing realization simulation principles and services, and then uses the results in a word count experiment. The experiment uses a word count of about eleven thousand words to evaluate how well MR is implemented on Alibaba Cloud. It is validated for execution time on the platform with varying Central Processing Unit (CPU) core counts, memory configurations, and worker counts. By testing several platform configurations, it is found that the memory configuration has very little impact on the model's execution time, while the size of the CPU core has a considerable impact on reaction time relative to the number of workers.

Controllable non-uniformly distributed spiking cluster generation in broadband optoelectronic oscillator

An approach to achieve controllable non-uniformly distributed spiking cluster generation is proposed and demonstrated based on an externally-triggered broadband optoelectronic oscillator (OEO). The theory of controlling the distribution of the spiking pulses in a spiking cluster is established. Based on the theory, the dynamic and the distribution characteristics are analyzed and revealed in the stable spiking oscillation state under different externally-injected trigger signal voltages. The peak-voltage envelop of the cluster and the interval of the spiking pulses are demonstrated to have an approximate negative linearity relationship with the externally-injected trigger signal voltage in both the numerical simulation and the experiment, where a square waveform, a trapezoidal waveform, a parabola waveform, and a half-sinusoidal waveform are used as the externally-injected trigger signals. The results indicate that the spiking pulse distribution in the generated spiking cluster can be well controlled through tuning the externally-injected trigger signal voltage. The proposed scheme can be utilized in spiking encoding and reservoir computing.

Faster or Cheaper: A Q-learning based cost-effective mixed cluster scaling method for achieving low tail latencies

The cloud elasticity allows users to acquire resources and release useless resources as needed. This feature has attracted more and more web service providers to deploy their latency-crucial, user-oriented applications on cloud platforms. For web service providers, in the case of fluctuating workloads, scaling their server clusters over time could save system expenditures without service quality violations. Therefore, a lot of cloud platforms begin to offer automatic scaling strategies based on threshold-based rules for helping web service providers to save system expenditures. However, building threshold-based rules requires expertise, and such reactive scaling strategies cannot guarantee low and consistent tail latencies. For those proactive scaling strategies depending on predictions, random user behaviors lead to declines in prediction accuracy. In this paper, we propose a reinforcement learning based proactive strategy for scaling a mixed cluster, which is composed of a variety of cloud instances. Ensuring the availability and quality of reward signals is the main problem to be solved for algorithms based on standard RL. We design a reward function which could balance service cost, service quality and other parameters which can affect decision-making. We assign different weights to parameters based on their effects on decision-making. In order to avoid the explosion of the state space caused by fluctuating workloads and various server status, we discretize the continuous state space of our model. Experimental results based on TailBench show that our Q-learning based scaling method can maintain low and consistent tail latencies while achieving fewer costs than three common baselines.

Blind-Touch: Homomorphic Encryption-Based Distributed Neural Network Inference for Privacy-Preserving Fingerprint Authentication

Fingerprint authentication is a popular security mechanism for smartphones and laptops. However, its adoption in web and cloud environments has been limited due to privacy concerns over storing and processing biometric data on servers. This paper introduces Blind-Touch, a novel machine learning-based fingerprint authentication system leveraging homomorphic encryption to address these privacy concerns. Homomorphic encryption allows computations on encrypted data without decrypting. Thus, Blind-Touch can keep fingerprint data encrypted on the server while performing machine learning operations. Blind-Touch combines three strategies to efficiently utilize homomorphic encryption in machine learning: (1) It optimizes the feature vector for a distributed architecture, processing the first fully connected layer (FC-16) in plaintext on the client side and the subsequent layer (FC-1) post-encryption on the server, thereby minimizing encrypted computations; (2) It employs a homomorphic encryption-compatible data compression technique capable of handling 8,192 authentication results concurrently; and (3) It utilizes a clustered server architecture to simultaneously process authentication results, thereby enhancing scalability with increasing user numbers. Blind-Touch achieves high accuracy on two benchmark fingerprint datasets, with a 93.6% F1- score for the PolyU dataset and a 98.2% F1-score for the SOKOTO dataset. Moreover, Blind-Touch can match a fingerprint among 5,000 in about 0.65 seconds. With its privacy-focused design, high accuracy, and efficiency, Blind-Touch is a promising alternative to conventional fingerprint authentication for web and cloud applications.

Electrophysiological correlates of temporal numerosity adaptation.

Much research has revealed the human visual system is capable to estimate numerical quantities, rapidly and reliably, in both the spatial and the temporal domain. This ability is highly susceptible to short-term plastic phenomena related to previous exposure to visual numerical information (i.e., adaptation). However, while determinants of spatial numerosity adaptation have been widely investigated, little is known about the neural underpinnings of short-term plastic phenomena related to the encoding of temporal numerical information. In the present study we investigated the electrophysiological correlates of temporal numerosity adaptation. Participants were asked to estimate the numerosity of a test sequence of flashes after being exposed to either a high or low numerous adapting sequence. Behavioral results confirmed the expected underestimation of test stimulus when this was preceded by a high numerous sequence as compared to when preceded by a low numerous sequence. Electrophysiological data revealed that this behavior was tightly linked to the amplitude of the steady-state visual evoked (ssVEP) response elicited by the test stimulus. When preceded by a high numerous sequence, the test stimulus elicited larger ssVEP responses as compared to when preceded by a low numerous sequence with this pattern being robustly correlated with behavior. Finally, topographical maps showed that this difference was mostly evident across two antero-posterior distributed clusters of electrodes and correlated with changes in functional connectivity. Taken together, our results suggest that visual plastic phenomena related to the encoding of temporal numerosity information reflect changes in rhythmic evoked activity that are likely related to long range communications between distinct brain regions.

Optimization of artificial intelligence cloud computing in information management design

In order to promote the construction of enterprise informatization, the author studied the adaptive optimization method of Efficient Management Information System (EMIS). The author proposes to combine the fuzzy C-means algorithm to form a server clustering algorithm, and adds an improved Drosophila optimization algorithm to overcome the problems of slow Rate of convergence of GRNN and easy to fall into the minimum, and the cloud platform collects 23 performance indicators, the output results of the coordinated evolutionary algorithm are analyzed by the neighborhood rough set analysis of algorithms to select features to avoid the curse of dimensionality problem. The experimental results indicate that, compared with existing research results, the algorithm proposed by the author has increased its speed by 1.43, 3.22, and 3.72 times, respectively; In terms of convergence steps, they have also been reduced by 1.61, 5, and 6 times respectively, and when running the algorithm, the computer’s memory and CPU usage are controlled at around 50%, without affecting normal functionality. This proves that the task scheduling of the cloud platform is more balanced, and indirectly proves the accuracy of the algorithm’s clustering effect.

Exploiting Data Locality to Improve Performance of Heterogeneous Server Clusters

We consider load balancing in large-scale heterogeneous server systems in the presence of data locality that imposes constraints on which tasks can be assigned to which servers. The constraints are naturally captured by a bipartite graph between the servers and the dispatchers handling assignments of various arrival flows. When a task arrives, the corresponding dispatcher assigns it to a server with the shortest queue among [Formula: see text] randomly selected servers obeying these constraints. Server processing speeds are heterogeneous, and they depend on the server type. For a broad class of bipartite graphs, we characterize the limit of the appropriately scaled occupancy process, both on the process level and in steady state, as the system size becomes large. Using such a characterization, we show that imposing data locality constraints can significantly improve the performance of heterogeneous systems. This is in stark contrast to either heterogeneous servers in a full flexible system or data locality constraints in systems with homogeneous servers, both of which have been observed to degrade the system performance. Extensive numerical experiments corroborate the theoretical results. Funding: This work was partially supported by the National Science Foundation [CCF. 07/2021–06/2024].