Computational Network Research Articles

Forward genetics combined with unsupervised classifications identified zebrafish mutants affecting biliary system formation

Impaired formation of the biliary network can lead to congenital cholestatic liver diseases; however, the genes responsible for proper biliary system formation and maintenance have not been fully identified. Combining computational network structure analysis algorithms with a zebrafish forward genetic screen, we identified 24 new zebrafish mutants that display impaired intrahepatic biliary network formation. Complementation tests suggested these 24 mutations affect 24 different genes. We applied unsupervised clustering algorithms to unbiasedly classify the recovered mutants into three classes. Further computational analysis revealed that each of the recovered mutations in these three classes has a unique phenotype on node-subtype composition and distribution within the intrahepatic biliary network. In addition, we found most of the recovered mutations are viable. In those mutant fish, which are already good animal models to study chronic cholestatic liver diseases, the biliary network phenotypes persist into adulthood. Altogether, this study provides unique genetic and computational toolsets that advance our understanding of the molecular pathways leading to biliary system malformation and cholestatic liver diseases.

Deep Reinforcement Learning-based Mining Task Offloading Scheme for Intelligent Connected Vehicles in UAV-aided MEC

The convergence of unmanned aerial vehicle (UAV)-aided mobile edge computing (MEC) networks and blockchain transforms the existing mobile networking paradigm. However, in the temporary hotspot scenario for intelligent connected vehicles (ICVs) in UAV-aided MEC networks, deploying blockchain-based services and applications in vehicles is generally impossible due to its high computational resource and storage requirements. One possible solution is to offload part of all the computational tasks to MEC servers wherever possible. Unfortunately, due to the limited availability and high mobility of the vehicles, there is still lacking simple solutions that can support low-latency and higher reliability networking services for ICVs. In this article, we study the task offloading problem of minimizing the total system latency and the optimal task offloading scheme, subject to constraints on the hover position coordinates of the UAV, the fixed bonuses, flexible transaction fees, transaction rates, mining difficulty, costs and battery energy consumption of the UAV. The problem is confirmed to be a challenging linear integer planning problem, we formulate the problem as a constrained Markov decision process. Deep Reinforcement Learning (DRL) has excellently solved sequential decision-making problems in dynamic ICVs environment, therefore, we propose a novel distributed DRL-based P-D3QN approach by using Prioritized Experience Replay strategy and the dueling double deep Q-network (D3QN) algorithm to solve the optimal task offloading policy effectively. Finally, experiment results show that compared with the benchmark scheme, the P-D3QN algorithm can bring about 26.24% latency improvement and increase about 42.26% offloading utility.

A Novel Approach of Cloud Computing Network for Authentication and Security Enhancement of IoT Enabled Cancer Forecasting System

Recently, a variety of approaches have been employed to address a broad spectrum of real-world issues; these methodology cover a variety of areas, including healthcare systems. Previous researchers concentrated on health-care monitoring software. It had several shortcomings, such as poor health-care data storage, time, expense, and processing complexity. This paper proposes a unique IoT-enabled and secured clinical monitoring paradigm to address these issues. Initially, implant several sensors to gather information on vital indicators like body temperature fluctuation. Phone numbers, marital status, heart rate deviation, residence, name, age, and blood pressure are among the patient's health information. The IoT medical sensor dataset is used in this investigation. During pre-processing, extra unnecessary attributes are removed, resulting in data size reduction and normalization. A convolutional neural network supporting the classification of cancer sickness that is based on the improved teaching-learning optimization (CNN-ITLO) method. The CNN-ITLO model ascertains whether or not the patient is cancer-prone based on the sensor input. The management of the hospital receives the gathered data after which it is analyzed. Lastly, the homomorphic encryption approach is used to encrypt and store the patient's data on the cloud. The Java platform will be used to carry out the recommended work. According to the experiment results, the suggested method performed better than current cutting-edge practices.

Joint Optimization of Measurement Point Intelligent Selection and End-to-End Network Traffic Calculation in Datacenters

The effective design and management of data centers needs to follow the end-to-end traffic characteristics of data center networks (DCNs). However, directly measuring the end-to-end traffic of the network requires huge software and hardware costs. Since the particularity of the structure of DCNs, the flow estimation method used in the traditional computer network cannot be applied to existing DCNs. In this paper, we study the end-to-end traffic calculation of cloud computing DCNs. We propose <bold xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">LLS-TC</b> , which is an intelligent end-to-end traffic inference algorithm based on network tomography. Only using SNMP (simple network management protocol) data generally supported by switches, end-to-end traffic information can be calculated quickly and accurately. <bold xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">LLS-TC</b> first devices a network traffic measurement point intelligent selection scheme based on node weighting. It first assigns weight to nodes through node criticality, and then uses node weighted incidence matrix approximation algorithm to calculate initial solution. <bold xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">LLS-TC</b> then designs a network tomography method suitable for cloud computing network traffic calculation. It uses the time correlation of data center traffic to model the algorithm problem into a linear state space model, and finally calculates the traffic carried by each path through the improved Kalman filtering algorithm. Our evaluation and analysis demonstrate that <bold xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">LLS-TC</b> can effectively use the extracted coarse-grained traffic characteristics, and greatly improve the accuracy of the calculation on the premise of ensuring the computational efficiency.

Satellite-Based ITS Data Offloading &amp; Computation in 6G Networks: A Cooperative Multi-Agent Proximal Policy Optimization DRL With Attention Approach

Satellite-Based ITS Data Offloading &amp; Computation in 6G Networks: A Cooperative Multi-Agent Proximal Policy Optimization DRL With Attention Approach

Collaborative Computing in Non-Terrestrial Networks: A Multi-Time-Scale Deep Reinforcement Learning Approach

Collaborative Computing in Non-Terrestrial Networks: A Multi-Time-Scale Deep Reinforcement Learning Approach

Implementing Optical Analog Computing and Electrooptic Hopfield Network by Silicon Photonic Circuits

Implementing Optical Analog Computing and Electrooptic Hopfield Network by Silicon Photonic Circuits

Large-scale experiment in STEM education for high school students using artificial intelligence kit based on computer vision and Python

This study proposes an artificial intelligence (AI) kit for high school students in science, technology, engineering, and mathematics (STEM). The AI kit includes an edge AI machine and electronic components. A compact, purpose-built kit resembling a laptop was designed for ease of replication and portability. Utilizing pre-trained convolutional neural network models and computer vision algorithms, five Thai schools participated in on-site instructions. A quasi-experimental study assessed the students' learning outcomes using a paired sample t-test. Results revealed improved knowledge and reduced score variation. Additionally, gender analysis confirmed that both male and female students met the learning criteria. The students expressed satisfaction with the distinctive hardware and learning method employed during the class activities. Notably, the test results demonstrated that the AI kit enhanced students’ enthusiasm and facilitated comprehension.

Energy Efficient Transmission Strategy for Mobile Edge Computing Network in UAV-Based Patrol Inspection System

Energy Efficient Transmission Strategy for Mobile Edge Computing Network in UAV-Based Patrol Inspection System

Adaptive Training and Aggregation for Federated Learning in Multi-Tier Computing Networks

Adaptive Training and Aggregation for Federated Learning in Multi-Tier Computing Networks

Impact of white noise in artificial neural networks trained for classification: Performance and noise mitigation strategies.

In recent years, the hardware implementation of neural networks, leveraging physical coupling and analog neurons has substantially increased in relevance. Such nonlinear and complex physical networks provide significant advantages in speed and energy efficiency, but are potentially susceptible to internal noise when compared to digital emulations of such networks. In this work, we consider how additive and multiplicative Gaussian white noise on the neuronal level can affect the accuracy of the network when applied for specific tasks and including a softmax function in the readout layer. We adapt several noise reduction techniques to the essential setting of classification tasks, which represent a large fraction of neural network computing. We find that these adjusted concepts are highly effective in mitigating the detrimental impact of noise.

Scalable Cloud Architectures for Deploying AI Applications

The integration of Artificial Intelligence (AI) into cloud computing represents a pivotal shift in the technological landscape, offering unprecedented opportunities for innovation across industries. Deploying AI applications at scale on the cloud poses unique challenges, necessitating the development of scalable cloud architectures tailored to meet the computational demands and data processing needs inherent to AI. This article explores the foundational aspects of cloud computing architectures, emphasizing the principles of scalability essential for AI applications. It delves into the core challenges faced when deploying AI on the cloud, such as computational requirements, data management, network constraints, and security concerns. Further, it presents various architectural models that facilitate scalability, including containerization, serverless computing, and cloud-native AI services, drawing from realworld case studies to illustrate effective strategies and best practices. Additionally, the article examines performance optimization techniques, security considerations, and the future directions of cloud-based AI deployments, highlighting the role of emerging technologies such as quantum computing and edge AI. By providing a comprehensive overview of scalable cloud architectures for AI applications, this article aims to guide researchers, practitioners, and organizations in leveraging cloud computing to its full potential, thereby enabling more efficient, secure, and scalable AI solutions.

Internet of Things (IoT) Data Communication Challenges, and Solutions

IoT has become an integral part of our lives, and data communication is one of the most important aspects of IoT. IoT devices can exchange data with each other via secure communication channels. This study aimed to review the IoT data communication challenges and solutions This paper attempts to address the key challenges and solutions for effective IoT data communication, starting by exploring the current challenges of IoT data communication related to scalability, reliability, and security, potential solutions to these challenges, such as using a mesh network, centralized servers, and edge computing. In addition, the paper discusses the importance of proper authentication and authorization techniques to ensure secure communication between IoT devices. The potential of IoT data communication is vast and has the power to fundamentally transform our society. It is an exciting time to be involved in this field, and the future looks promising for the advancement of IoT data communication.

Verification of Neural Networks’ Global Robustness

Neural networks are successful in various applications but are also susceptible to adversarial attacks. To show the safety of network classifiers, many verifiers have been introduced to reason about the local robustness of a given input to a given perturbation. While successful, local robustness cannot generalize to unseen inputs. Several works analyze global robustness properties, however, neither can provide a precise guarantee about the cases where a network classifier does not change its classification. In this work, we propose a new global robustness property for classifiers aiming at finding the minimal globally robust bound, which naturally extends the popular local robustness property for classifiers. We introduce VHAGaR, an anytime verifier for computing this bound. VHAGaR relies on three main ideas: encoding the problem as a mixed-integer programming and pruning the search space by identifying dependencies stemming from the perturbation or the network's computation and generalizing adversarial attacks to unknown inputs. We evaluate VHAGaR on several datasets and classifiers and show that, given a three hour timeout, the average gap between the lower and upper bound on the minimal globally robust bound computed by VHAGaR is 1.9, while the gap of an existing global robustness verifier is 154.7. Moreover, VHAGaR is 130.6x faster than this verifier. Our results further indicate that leveraging dependencies and adversarial attacks makes VHAGaR 78.6x faster.

Optimizing microelectronic module cooling under magnetic fields through hybrid nanofluid: a computational fluid dynamics-artificial neural network approach

Optimizing microelectronic module cooling under magnetic fields through hybrid nanofluid: a computational fluid dynamics-artificial neural network approach

Optimisation Methods Based on Soft Computing for Improving Power System Stability

In order to keep electrical grids running reliably and efficiently, power system stability is essential. The optimisation of power systems is fraught with complexity and uncertainty, yet soft computing approaches have been shown to be an excellent tool for dealing with these issues. In order to improve the reliability of the power system, this article presents a summary of optimisation techniques that rely on soft computing techniques. , the methods of evolutionary computing, such as particle swarm optimisation (PSO) and genetic algorithms (GA), are discussed. By efficiently searching for optimum solutions in huge solution spaces, these algorithms are used to optimise power system parameters and control techniques. In general, optimisation techniques based on soft computing provide effective and flexible ways to enhance power system stability. These technologies improve grid operation by using swarm intelligence, evolutionary computing, fuzzy logic, and neural networks to tackle the issues of power systems, which are dynamic and unpredictable.

A Comparison of Madaline and Perceptron Algorithms on Classification with Quantum Computing Approach

The fundamental problem in this research is to explore a more profound understanding regarding both performance and efficiency in quantity computing. Successful implementation of algorithms in computational computing environments can unlock the potential for significant improvements in information processing and neural network modelling. This research focuses on developing the Madaline and Perceptron algorithm using a quantum approach. This study compares the two algorithms regarding the accuracy and epoch of the test results. The dataset used in this study is using a lens dataset. There are four attributes: 1) patient age: young, pre-presbyopia, presbyopia 2) eyeglass prescription: myopia, hypermetropia 3) astigmatic: no, yes. 4) tear production rate: reduced, normal. There are three classes: 1) patients must have hard contact lenses installed, 2) patients must have soft contact lenses installed, and 3) patients cannot have contact lenses installed. The number of data is 24 data. The result of this research is the development of the Madaline and Perceptron algorithms with a quantum computing approach. Comparing these algorithms shows the best accuracy is the Perceptron algorithm, namely 100%. In comparison, Madaline is 62.5%, and the smallest epoch is the Madaline algorithm, namely 4 epochs, while the smallest Perceptron epoch is 317. This research significantly contributes to developing computing and neural networks, with potential applications extending from data processing to more accurate modelling in artificial intelligence, data analysis and understanding complex patterns.

Minimizing Task Age upon Decision for Low-Latency MEC Networks Task Offloading with Action-Masked Deep Reinforcement Learning.

In this paper, we consider a low-latency Mobile Edge Computing (MEC) network where multiple User Equipment (UE) wirelessly reports to a decision-making edge server. At the same time, the transmissions are operated with Finite Blocklength (FBL) codes to achieve low-latency transmission. We introduce the task of Age upon Decision (AuD) aimed at the timeliness of tasks used for decision-making, which highlights the timeliness of the information at decision-making moments. For the case in which dynamic task generation and random fading channels are considered, we provide a task AuD minimization design by jointly selecting UE and allocating blocklength. In particular, to solve the task AuD minimization problem, we transform the optimization problem to a Markov Decision Process problem and propose an Error Probability-Controlled Action-Masked Proximal Policy Optimization (EMPPO) algorithm. Via simulation, we show that the proposed design achieves a lower AuD than baseline methods across various network conditions, especially in scenarios with significant channel Signal-to-Noise Ratio (SNR) differences and low average SNR, which shows the robustness of EMPPO and its potential for real-time applications.

Generalized coefficients of clustering in (un)directed and (un)weighted networks: An application to systemic risk quantification for cryptocoin markets

A coefficient of clustering is one of the important network parameters to describe clustering characteristics around a node in a network. It accounts for the number of cycles of length 3 (triangles) relative to the number of paths of length 2 in which this node participates as their inner node. In practice, the number of longer paths and cycles may be larger than the number of 2-paths and 3-cycles, thereby requiring its generalized version. In this paper, we aim at proposing the so-called λ-clustering coefficient to characterize the node clustering tendency at level λ≥3. We define it by counting the number of λ-cycles relative to the number of (λ−1)-paths containing a fixed inner node. In addition, we introduce the total clustering coefficient by considering the total number of all cycles relative to the total number of all paths consisting of this inner node. We formulate these generalized clustering coefficients for each of the following cases: undirected and unweighted networks, undirected but weighted networks, directed but unweighted networks, and directed and weighted networks. We then implement them for correlation- and conditional risk measure-based networks that represent conventional cryptocurrency markets and stablecoin markets. They can be viewed as new measures of systemic risk. The larger their values, the larger the tendency of risk transmission from one market to other subsequent markets forming longer cycles, and therefore the higher the systemic risk level. The systemic risk quantification is appropriate when they fulfill certain properties. In summary, the novelty of this study lies in the mathematical formulation of the generalized clustering coefficients and their computation for the cryptocoin networks with the purpose of appropriately managing systemic risk in the represented cryptocoin markets.

Computational neural network for processing light-reflective spectra of plants and remote phytosanitary monitoring of potatoes

The article is devoted to studying the possibility of using the WaveLetNN artificial neural network to analyze the results of remote phytosanitary monitoring of early detection of plants in potato plantings affected by late blight. Various methods for analyzing the spectral characteristics of plant reflection are considered, including the classification method. To detect plants infected with late blight, the WaveLetNN neural network analyzes the light reflective characteristics of potato plants obtained as a result of research (in the range of 300–1100 nm) and calculates the cognitive significance index (CSI = 0...10), which characterizes the intensity of biochemical processes inside plants aimed at countering phytopathogenic microflora. It was found that a significant increase in the CSI index signals infection of plants by phytopathogenic microflora and activation of protective biochemical processes on the part of plants. To reliably indicate infected plants, the WaveLetNN neural network underwent test training on a large number of light reflectance spectra of uninfected plants and plants artificially infected with late blight. The spectral reflectance characteristics of infected and uninfected plants were measured during 3, 4, 7 and 8 days after infection. Processing the obtained spectra using the WaveLetNN neural network made it possible to identify significant differences between the second- and third-order spectral characteristics of uninfected and late blight infected plants on the third day after infection. Moreover, for infected plants the CSI index values were 6.1...6.7, and CSI for healthy plants – 1.9...2.5. The Wave-LetNN neural network eliminates the influence on the light reflectance spectra of the spatial arrangement of plant leaves, unevenness of the soil surface and shading of individual sections of the field, normalizing the spectra to the total intensity of light reflected from the leaves. Thus, the WaveLetNN neural network can be used as the software core of online systems for remote phytosanitary monitoring of potato plants.