Client-server Research Articles

RFL-LSU: A Robust Federated Learning Approach with Localized Stepwise Updates

Distributed intelligence enables the widespread deployment of AI technology, greatly promoting the development of AI. Federated learning is a widely used distributed intelligence technology that allows iterative optimization of global model while protecting user data privacy. Currently, federated learning faces some security threats, as its open architecture provides attackers with opportunities to disrupt the learning process by submitting malicious updates or inserting backdoors. In this article, we propose a robust federated learning method to defend against potential malicious attacks. Specifically, we enhance the algorithm’s performance and stability by implementing localized stepwise updates on the client side and element-wise anomaly detection on the server side. We conducted experiments in a more realistic non-i.i.d. scenario and compared the results with other typical federated learning methods. Our results demonstrate that our approach exhibits strong robustness under non-i.i.d. data distributions, outperforming other methods in terms of test accuracy and resilience to attacks.

Image Spam Detection through Pattern Recognition with IP Tracing Technique

Image spam is a kind of E-mail spam where the message text of the spam is presented as a picture in an image file. The basic rationale behind image spam is difficult to detect using text spam filtering methods which is designed to detect patterns in text in the plain-text E-mail body or attachments. A new trend in email spam is the emergence of image spam. The most previous works of image spam detection focused on filtering the image spam on the client side. This proposed system considered more desirable comprehensive solution which embraces the both server side filtering and client side detection methods. The spectral clustering algorithm is introduced to similarity measure for cluster analysis of spam images to filter the attack activities of spammers and fast trace back the spam sources. The active learning algorithm is limited where the learner guides the users to label as few images as possible while maximizing the classification accuracy.

Automatically Injecting Robustness Statements into Distributed Applications

When developing a distributed application, several issues need to be handled, and software components should include some mechanisms to make their execution resilient when network faults, delays, or tampering occur. For example, synchronous calls represent a too-tight connection between a client requesting a service and the service itself, whereby potential network delays or temporary server overloads would keep the client side hanging, exposing it to a domino effect. The proposed approach assists developers in dealing with such issues by providing an automatic tool that enhances a distributed application using simple blocking calls and makes it robust in the face of adverse events. The proposed devised solution consists in automatically identifying the parts of the application that connect to remote services using simple synchronous calls and substituting them with a generated customized snippet of code that handles potential network delays or faults. To accurately perform the proposed transformation, the devised tool finds application code statements that are data-dependent on the results of the original synchronous calls. Then, for the dependent statements, a solution involving guarding code, proper synchronization, and timeouts is injected. We experimented with the analysis and transformation of several applications and report a meaningful example, together with the analysis of the results achieved.

Networked microcontrollers for accessible, distributed spatial audio

State-of-the-art systems for spatial and immersive audio are typically very costly, being reliant on specialist audio hardware capable of performing computationally intensive signal processing and delivering output to many tens, if not hundreds, of loudspeakers. Centralised systems of this sort suffer from limited accessibility due to their inflexibility and expense. Building on the research of the past few decades in the transmission of audio data over computer networks, and the emergence in recent years of increasingly capable, low-cost microcontroller-based development platforms with support for both networking and audio functionality, we present a prototype decentralised, modular alternative. Having previously explored the feasibility of running a microcontroller device as a networked audio client, here we describe the development of a client-server system with improved scalability via multicast data transmission. The system operates on ubiquitous, commonplace computing and networking equipment, with a view to it being a simple, versatile, and highly-accessible platform, capable of granting users the freedom to explore audio spatialisation approaches at vastly reduced expense. Though faced by significant technical challenges, particularly with regard to maintaining synchronicity between distributed audio processors, the system produces perceptually plausible results. Findings are commensurate with a capability, with further development and research, to disrupt and democratise the fields of spatial and immersive audio.

IS-USG: INTELLIGENCE SYSTEM ULTRASONOGRAPHY Machine Learning Implementation for Recognition of Pregnant Sheep in Gumukmas Multifarm Jember

In 2023, the government's program related to food independence emphasizes the importance of meat independence, including goat meat. CV. Gumukmas Multi Farm Jember faces several challenges in managing pregnant sheep, such as early detection of pregnancy, preventing abortion, and managing sheep health history. Therefore, the Intelligence System Ultrasonography (IS-USG) system was developed to detect sheep pregnancy more accurately and integrate medical data with the help of machine learning. This system allows automatic monitoring of pregnancy conditions and fetal health, and recommends interventions such as probiotic feeding. IS-USG is implemented using a web and mobile-based client-server architecture, with Laravel and MySQL backends, and RESTful API integration for real-time data synchronization. Testing was conducted to evaluate the effectiveness of the system, and initial results showed that this system can improve sheep reproductive management, accelerate pregnancy detection, and minimize the risk of birth failure.

Event log extraction methodology for Ethereum applications

The adoption of smart contracts in decentralized blockchain-based applications enables reliable and certified audits. These audits allow the extraction of valuable information from blockchains, which can be used to reconstruct the execution of the application and facilitate advanced analyses. One of the most commonly used techniques in this context is process mining, which leverages event logs to trace and accurately represent the process execution of applications. However, extracting execution data from blockchains poses significant challenges, and the current methodologies developed have some limitations. Most approaches are tailored to specific use cases, requiring that analysis techniques are defined during the smart contract’s development. Other techniques are applied a posteriori, relying on blockchain events that often lack a standardized format. This absence of standardization requires complex processing steps to correlate logs with the executed actions and such approaches are not universally applicable to all smart contracts on the blockchain, further limiting their scope. Lastly, none of the existing techniques can extract information from event logs embedded in internal transactions of smart contracts.To address these limitations, we propose EveLog an application-agnostic methodology that can be applied to any EVM-compatible application without predefined constraints. Its primary goal is to extract information from smart contracts, capturing both public and internal transactions, and organizing the results into a structured XES event log. The EveLog methodology consists of five key steps: (i) extraction of data from smart contract transactions, (ii) decoding raw data, (iii) selection of sorting criteria, (iv) construction of traces, and (v) generation of the XES event log. EveLog has been implemented in a client–server application and tested on existing solutions, specifically the CryptoKitties application, a blockchain-based game on the Ethereum blockchain. The study was conducted using 12,996 blocks, including over 8000 real transactions from the Ethereum mainnet.

A complete client–server architecture providing prevention from insider attack using homogeneous multi-core RSA crypto-processor

A complete client–server architecture providing prevention from insider attack using homogeneous multi-core RSA crypto-processor

Smart-KG: Partition-Based Linked Data Fragments for querying knowledge graphs

RDF and SPARQL provide a uniform way to publish and query billions of triples in open knowledge graphs (KGs) on the Web. Yet, provisioning of a fast, reliable, and responsive live querying solution for open KGs is still hardly possible through SPARQL endpoints alone: while such endpoints provide a remarkable performance for single queries, they typically can not cope with highly concurrent query workloads by multiple clients. To mitigate this, the Linked Data Fragments (LDF) framework sparked the design of different alternative low-cost interfaces such as Triple Pattern Fragments (TPF), that partially offload the query processing workload to the client side. On the downside, such interfaces still come with the expense of unnecessarily high network load due to the necessary transfer of intermediate results to the client, leading to query performance degradation compared with endpoints. To address this problem, in the present work, we investigate alternative interfaces, refining and extending the original TPF idea, which also aims at reducing server-resource consumption, by shipping query-relevant partitions of KGs from the server to the client. To this end, first, we align formal definitions and notations of the original LDF framework to uniformly present existing LDF implements and such “partition-based” LDF approaches. These novel LDF interfaces retrieve, instead of the exact triples matching a particular query pattern, a subset of pre-materialized, compressed, partitions of the original graph, containing all answers to a query pattern, to be further evaluated on the client side. As a concrete representative of partition-based LDF, we present smart-KG+, extending and refining our prior work (In WWW ’20: The Web Conference 2020 (2020) 984–994 ACM / IW3C2) in several respects. Our proposed approach is a step forward towards a better-balanced share of the query processing load between clients and servers by shipping graph partitions driven by the structure of RDF graphs to group entities described with the same sets of properties and classes, resulting in significant data transfer reduction. Our experiments demonstrate that the smart-KG+ significantly outperforms existing Web SPARQL interfaces on both pre-existing benchmarks for highly concurrent query execution as well as an accustomed query workload inspired by query logs of existing SPARQL endpoints.

Efficient federated learning for pediatric pneumonia on chest X-ray classification

According to the World Health Organization (WHO), pneumonia kills about 2 million children under the age of 5 every year. Traditional machine learning methods can be used to diagnose chest X-rays of pneumonia in children, but there is a privacy and security issue in centralizing the data for training. Federated learning prevents data privacy leakage by sharing only the model and not the data, and it has a wide range of application in the medical field. We use federated learning method for classification, which effectively protects data security. And for the data heterogeneity phenomenon existing in the actual scenario, which will seriously affect the classification effect, we propose a method based on two-end control variables. Specifically, based on the classical federated learning FedAvg algorithm, we modify the loss function on the client side by adding a regular term or a penalty term, and add momentum after the average aggregation on the server side. The federated learning approach prevents the data privacy leakage problem compared to the traditional machine learning approach. In order to solve the problem of low classification accuracy due to data heterogeneity, our proposed method based on two-end control variables achieves an average improvement of 2% and an accuracy of 98% on average, and 99% individually, compared to the previous federated learning algorithms and the latest diffusion model-based method. The classification results and methodology of this study can be utilized by clinicians worldwide to improve the overall detection of pediatric pneumonia.

Web-платформа для організації хмарних обчислень у нейрофізіологічних дослідженнях на основі даних айтрекінгу

The purpose of the work is to develop the architecture and web version of the software complex based on the proposed new concept of cloud computing organization, which allows to expand the diagnostic capabilities of the tools of model-oriented information technology for assessing the neurophysiological state of a person using methods of nonlinear dynamic identification of the oculomotor system based on eye tracking data. The concept of cloud computing is proposed, which is based on the combination of PaaS and SaaS services as part of the developed software complex, due to which the cross-platform nature of cloud computing is ensured, the productivity and efficiency of scientific research increases. The developed architecture allows you to easily expand the functionality of the software complex and adapt it to different application conditions. The key feature of the complex is its undemanding hardware on the client side thanks to the use of cloud computing and its modular structure, which allows for easy scaling of the complex, as well as the isolation of the script code execution process in the cloud computing environment to increase the level of security when interpreting the script code on the server . Compared to other similar services, the complex has several advantages: it provides effective work in research and educational applications, supports Python and JavaScript programming languages, and also allows the use of software-implemented identification methods through specially developed GUI interfaces. In addition, the complex offers social opportunities and a high level of abstraction, which allows to optimize the research process.

Federated influencer learning for secure and efficient collaborative learning in realistic medical database environment.

Enhancing deep learning performance requires extensive datasets. Centralized training raises concerns about data ownership and security. Additionally, large models are often unsuitable for hospitals due to their limited resource capacities. Federated learning (FL) has been introduced to address these issues. However, FL faces challenges such as vulnerability to attacks, non-IID data, reliance on a central server, high communication overhead, and suboptimal model aggregation. Furthermore, FL is not optimized for realistic hospital database environments, where data are dynamically accumulated. To overcome these limitations, we propose federated influencer learning (FIL) as a secure and efficient collaborative learning paradigm. Unlike the server-client model of FL, FIL features an equal-status structure among participants, with an administrator overseeing the overall process. FIL comprises four stages: local training, qualification, screening, and influencing. Local training is similar to vanilla FL, except for the optional use of a shared dataset. In the qualification stage, participants are classified as influencers or followers. During the screening stage, the integrity of the logits from the influencer is examined. If the integrity is confirmed, the influencer shares their knowledge with the others. FIL is more secure than FL because it eliminates the need for model-parameter transactions, central servers, and generative models. Additionally, FIL supports model-agnostic training. These features make FIL particularly promising for fields such as healthcare, where maintaining confidentiality is crucial. Our experiments demonstrated the effectiveness of FIL, which outperformed several FL methods on large medical (X-ray, MRI, and PET) and natural (CIFAR-10) image dataset in a dynamically accumulating database environment, with consistently higher precision, recall, Dice score, and lower standard deviation between participants. In particular, in the PET dataset, FIL achieved about a 40% improvement in Dice score and recall.

Software tools for organizing cloud computing in psychophysiological research based on eyetracking data

The architecture and web version of the software complex have been developed, which significantly expands the diagnostic capabilities of model-oriented information technologies for the assessment of the neurophysiological state. The complex provides cross-platform cloud computing, increases the productivity and efficiency of scientific research, using methods of non-parametric identification of the oculomotor system based on eyetracking data, which is achieved thanks to a new concept of cloud computing organization. Cloud computing technology has been further developed thanks to the proposed concept that combines the principles of PaaS (Platform as a Service) and SaaS (Software as a Service). The key feature of this software complex is its undemanding hardware on the client side thanks to the use of cloud computing, and its modular structure, which allows it to be easily scaled. Compared to other similar services, the complex has several advantages: it provides effective work in research and educational areas, supports several programming languages for improving algorithms, and also allows the use of ready-made identification methods through specially developed GUI interfaces. In addition, it offers social capabilities and a high level of abstraction that optimizes the research process.

Design of software system architecture for reading and publishing amateur literary works

The article discusses the functionality and key requirements for a web application for reading and publishing amateur literary works. Against the backdrop of current changes in the book industry caused by the development of technology and the emergence of self-publishing, the need to create platforms that provide authors with the opportunity to publish content independently is emphasized. One of the main problems that authors face when publishing is the loss of the original formatting of their works when transferring them to text editors on web platforms. This paper proposes a solution to this problem by developing a web application with the function of importing the content of text files while preserving their original formatting. The article also contains a comparative analysis of popular web-based platforms for publishing works, such as Archive of Our Own, Wattpad, and Arkush, which helped to highlight the advantages and disadvantages of each of them. In particular, the difficulties associated with storing and processing unstructured data, such as images and text files, were investigated. To solve the problem of storing such data, a comparative analysis of three approaches was conducted: storing data in relational databases, local file systems, and cloud storage. Taking into account the results of this analysis, the web application under development was chosen to integrate with the Amazon Web Services cloud provider and use Amazon S3 cloud storage, which provides scalability, reliability, and convenience in managing large amounts of unstructured data. It was also decided to use the Amazon CloudFront CDN service to improve the speed of data access and increase overall system performance. In addition, the article describes the choice of a technology stack for implementing the web application, which includes Java, Spring Framework, MySQL, ORM Hibernate for the server side and HTML, CSS, React.js for the client side. The article highlights important aspects of designing a web application architecture aimed at scalability and resilience to large amounts of data and an increase in the number of active users, and provides a basis for further system improvement.

Bidirectional Corrective Model-Contrastive Federated Adversarial Training

When dealing with non-IID data, federated learning confronts issues such as client drift and sluggish convergence. Therefore, we propose a Bidirectional Corrective Model-Contrastive Federated Adversarial Training (BCMCFAT) framework. On the client side, we design a category information correction module to correct biases caused by imbalanced local data by incorporating the local client’s data distribution information. Through local adversarial training, more robust local models are obtained. Secondly, we propose a model-based adaptive correction algorithm in the server that leverages a self-attention mechanism to handle each client’s data distribution information and introduces learnable aggregation tokens. Through the self-attention mechanism, model contrast learning is conducted on each client to obtain aggregation weights of corrected client models, thus addressing the issues of accuracy degradation and slow convergence caused by client drift. Our algorithm achieves the best natural accuracy on the CIFAR-10, CIFAR-100, and SVHN datasets and demonstrates excellent adversarial defense performance against FGSM, BIM, and PGD attacks.

Real-Time Document Collaboration—System Architecture and Design

This article explores the world of dependable systems, specifically focusing on system design, software solutions, and architectural decisions that facilitate collaborative work on shared text documents across multiple users in near real time. It aims to dive into the intricacies of designing robust and effective document collaboration software focusing on understanding the requirements of such a system, the working principle of collaborative text editing, software architecture, technology stack selection, and tooling that can sustain such a system. To examine the pros and cons of the proposed system, the paper will detail how collaborative text editing software can benefit from such an architecture regarding availability, elasticity, and scaling. The intricate nature of this system renders this paper a valuable resource for prospective investigations within the domain of dependable systems and distributed systems. This research first examines the requirements of a real-time collaboration system and the necessary core features. Then, it analyzes the design, the application structure, and the system organization while also considering key architectural requirements as the necessity of scaling, the usage of microservices, cross-service communications, and client–server communication. For the technology stack of the implementation, this research considers the alternatives at each layer, from client to server. Once these decisions are made, it follows system development while examining possible improvements for the issues previously encountered. To validate the architecture, a testing strategy is developed, to examine the key capabilities of the system, such as resource consumption and throughput. The conclusions review the combination of modern and conventional application development principles needed to address the challenges of conflict-free document replication, decoupled and stateless event-driven architecture, idempotency, and data consistency. This paper not only showcases the design and implementation process but also sets a foundation for future research and innovation in dependable systems, collaborative technologies, sustainable solutions, and distributed system architecture.

FRAST: TFHE-Friendly Cipher Based on Random S-Boxes

A transciphering framework, also known as hybrid homomorphic encryption, is a practical method of combining a homomorphic encryption (HE) scheme with a symmetric cipher in the client-server model to reduce computational and communication overload on the client side. As a server homomorphically evaluates a symmetric cipher in this framework, new design rationales are required for “HE-friendly” ciphers that take into account the specific properties of the HE schemes. In this paper, we propose a new TFHE-friendly cipher, dubbed FRAST, with a TFHE-friendly round function based on a random S-box to minimize the number of rounds. The round function of FRAST can be efficiently evaluated in TFHE by a new optimization technique, dubbed double blind rotation. Combined with our new WoP-PBS method, the double blind rotation allows computing multiple S-box calls in the round function of FRAST at the cost of a single S-box call. In this way, FRAST enjoys 2.768 (resp. 10.57) times higher throughput compared to Kreyvium (resp. Elisabeth) for TFHE keystream evaluation in the offline phase of the transciphering framework at the cost of slightly larger communication overload.

Federated Learning with Multi-Method Adaptive Aggregation for Enhanced Defect Detection in Power Systems

The detection and identification of defects in transmission lines using computer vision techniques is essential for maintaining the safety and reliability of power supply systems. However, existing training methods for transmission line defect detection models predominantly rely on single-node training, potentially limiting the enhancement of detection accuracy. To tackle this issue, this paper proposes a server-side adaptive parameter aggregation algorithm based on multi-method fusion (SAPAA-MMF) and formulates the corresponding objective function. Within the federated learning framework proposed in this paper, each client executes distributed synchronous training in alignment with the fundamental process of federated learning. The hierarchical difference between the global model, aggregated using the improved joint mean algorithm, and the global model from the previous iteration is computed and utilized as the pseudo-gradient for the adaptive aggregation algorithm. This enables the adaptive aggregation to produce a new global model with improved performance. To evaluate the potential of SAPAA-MMF, comprehensive experiments were conducted on five datasets, involving comparisons with several algorithms. The experimental results are analyzed independently for both the server and client sides. The findings indicate that SAPAA-MMF outperforms existing federated learning algorithms on both the server and client sides.

Privacy-SF: An encoding-based privacy-preserving segmentation framework for medical images

Deep learning is becoming increasingly popular and is being extensively used in the field of medical image analysis. However, the privacy sensitivity of medical data limits the availability of data, which constrains the advancement of medical image analysis and impedes collaboration across multiple centers. To address this problem, we propose a novel encoding-based framework, named Privacy-SF, aimed at implementing privacy-preserving segmentation for medical images. Our proposed segmentation framework consists of three CNN networks: 1) two encoding networks on the client side that encode medical images and their corresponding segmentation masks individually to remove the privacy features, 2) a unique mapping network that analyzes the content of encoded data and learns the mapping from the encoded image to the encoded mask. By sequentially encoding data and optimizing the mapping network, our approach ensures privacy protection for images and masks during both the training and inference phases of medical image analysis. Additionally, to further improve the segmentation performance, we carefully design augmentation strategies specifically for encoded data based on its sequence nature. Extensive experiments conducted on five datasets with different modalities demonstrate excellent performance in privacy-preserving segmentation and multi-center collaboration. Furthermore, the analysis of encoded data and the experiment of model inversion attacks validate the privacy-preserving capability of our approach.

Real-Time CNN Training and Compression for Neural-Enhanced Adaptive Live Streaming.

We propose a real-time convolutional neural network (CNN) training and compression method for delivering high-quality live video even in a poor network environment. The server delivers a low-resolution video segment along with the corresponding CNN for super resolution (SR), after which the client applies the CNN to the segment in order to recover high-resolution video frames. To generate a trained CNN corresponding to a video segment in real-time, our method rapidly increases the training accuracy by promoting the overfitting property of the CNN while also using curriculum-based training. In addition, assuming that the pretrained CNN is already downloaded on the client side, we transfer only residual values between the updated and pretrained CNN parameters. These values can be quantized with low bits in real time while minimizing the amount of loss, as the distribution range is significantly narrower than that of the updated CNN. Quantitatively, our neural-enhanced adaptive live streaming pipeline (NEALS) achieves higher SR accuracy and a lower CNN compression loss rate within a constrained training time compared to the state-of-the-art CNN training and compression method. NEALS achieves 15 to 48% higher quality of the user experience compared to state-of-the-art neural-enhanced live streaming systems.

Predictive Prefetching in Client–Server Systems: A Navigational Behavior Modeling Approach

Predictive Prefetching in Client–Server Systems: A Navigational Behavior Modeling Approach