Earticle

Vertical Bell Lab Layered Space-Time is a spatial multiplexing (SM) multi-input multi- output (MIMO) scheme that provides high spectral efficiency with low computational complexity. Ordered Successive Interference Cancellation (OSCI) is known as most representative detection algorithm for SM MIMO system. Error propagation is a major drawback constraining the overall performance of OSIC detection significantly. To solve this problem, we propose a new detection algorithm, which improves the performance of the conventional detection algorithm with reasonable increase of computational complexity. Simulation results show that the proposed algorithm outperforms the conventional detection algorithm under spatially uncorrelated as well as correlated channels.

Email spoofing represents a common cybersecurity risk that abuses the weaknesses in the email protocols to falsify the sender addresses and trick the recipients into providing sensitive information or performing malicious requests. Conventional rulebased detection systems have been found ineffective against more advanced forms of spoofing. This research project suggests using machine learning to identify email spoofing using a range of features derived through email header, content, and metadata. Various algorithms, such as Support Vector Machines (SVM), Random Forest, and Logistic Regression, are tested to find the most suitable in terms of identifying legitimate emails and spoofed ones. On the dataset, preprocessing is done through tokenization, feature encoding, and vectorization to increase the model accuracy. The evaluation of the performance is performed in terms of precision, recall, F1-score, or ROC-AUC. Through experiments, it has been shown that machine learning models, especially ensemble based methods, greatly exceed traditional methods in accuracy and low false positive rate in detecting spoofed emails. This work has demonstrated the promise of intelligent systems when used to reinforce email security and has also offered a scalable implementation of real-time detection of spoofing.

음성 데이터 라벨링은 음성 인식 및 자연어 처리 기술의 성능을 결정하는 중요한 과정이다. 그러나 전통적인 수작업 라벨링 방식은 대규모 데이터 처리에 있어 많은 시간과 비용이 소요되며, 라벨링 오류의 가능성이 높다. 본 논문에서 는 이러한 문제를 해결하기 위해 효율적인 음성 데이터 라벨링 자동화 시스템을 설계하였다. 제안된 시스템은 음성 인식 기술과 기계 학습 알고리즘을 통합하여 음성 데이터를 자동으로 분석하고 라벨링하도록 설계되었다. 이 시스템은 다양한 음성 데이터 유형을 처리할 수 있으며, 사용자 인터페이스를 통해 라벨링 작업의 직관성과 효율성을 향상시킨다. 본 설계는 음성 데이터 처리의 효율성을 높이고, 다양한 응용 분야에서 음성 인식 기술의 활용을 촉진할 수 있는 잠재력을 가진다.

Speech data labeling is a crucial process that determines the performance of speech recognition and natural language processing technologies. However, traditional manual labeling methods are time-consuming, costly, and prone to labeling errors, especially when dealing with large-scale data. This paper presents the design of an efficient automated speech data labeling system to address these challenges. The proposed system integrates the latest speech recognition technologies and machine learning algorithms to automatically analyze and label speech data. It is designed to handle various types of speech data and enhances the intuitiveness and efficiency of the labeling process through a user-friendly interface. This design improves the efficiency of speech data processing and has the potential to facilitate the application of speech recognition technologies across various domains.

The appearance of Generative Artificial Intelligence (GAI) technologies is revolutionising global e-commerce. Redefining how consumers engage with online shopping platforms is reshaping e-commerce by influencing consumer behaviour across all decision-making stages. This research-in-progress applies EKB model and SEM to examine how AI-driven influences key consumer decision factors—namely Price of Product, Quality of Product, Services, Perceived Risk and Information Search—across five major markets: Japan, Korea, China, India, USA. By comparing traditional e-commerce models (with/without AI/ML) against (non-conversational/conversational) GAI platforms, the research provides nuanced insights into AI's impact at each stage of the consumer decision-making journey. Giving particular attention to the purchase decision stage, AI’s role in dynamically supporting consumer choices is critically assessed. Survey data are yet to be gathered and analysed to test the proposed model, offering new theoretical, practical implications for AI’s interactive capabilities will be hypothesised to alter consumer outcomes across culturally diverse settings.

The SLA 3d printer is the first of the commercial 3D printer. The 3D printed output is printed hanging on the bed that move to the upper position. Sandblasted bed is used to prevent layer shift. If sandblasting is wrong, the 3D printed output is layer shifted. For this reason, 3D printer manufacturing companies inspect the bed surface. However, the sandblasted surface has variety of irregular shapes and craters, so it is difficult to establish a quality control standard. To solve problems, this paper presents a standardized sandblasting histogram and threshold. We present a filter that can increase the classification rate.

현실 세계에서는 기존에 알려지지 않은 새로운 유형의 변종 공격이 끊임없이 등장하고 있지만, 인공신경망과 지도학 습을 통해 개발된 공격 트래픽 분류모델은 학습을 실시하지 않은 새로운 유형의 공격을 제대로 탐지하지 못한다. 기존 연구들 대부분은 이러한 문제점을 간과하고 인공신경망의 구조 개선에만 집중한 결과, 다수의 새로운 공격을 정상 트래 픽으로 분류하는 현상이 빈번하게 발생하여 공격 트래픽 분류성능이 심각하게 저하되었다. 한편, 다중분류 문제에서 각 클래스에 대한 분류가 정답일 확률을 결과값으로 출력하는 소프트맥스(softmax) 함수도 학습하지 않은 새로운 유형의 공격 트래픽에 대해서는 소프트맥스 점수를 제대로 산출하지 못하여 분류성능의 신뢰도 또는 정확도를 제고하는데 한계 를 노출하고 있다 . 이에 본 논문에서는 소프트맥스 함수의 이러한 특성을 활용하여 모델이 일정 수준 이하의 확률로 판단한 트래픽을 공격으로 분류함으로써 새로운 유형의 공격에 대한 탐지성능을 향상시키는 방안을 제안하고, 실험을 통해 효율성을 입증한다.

In the real world, new types of attacks or variants are constantly emerging, but attack traffic classification models developed through artificial neural networks and supervised learning do not properly detect new types of attacks that have not been trained. Most of the previous studies overlooked this problem and focused only on improving the structure of their artificial neural networks. As a result, a number of new attacks were frequently classified as normal traffic, and attack traffic classification performance was severly degraded. On the other hand, the softmax function, which outputs the probability that each class is correctly classified in the multi-class classification as a result, also has a significant impact on the classification performance because it fails to calculate the softmax score properly for a new type of attack traffic that has not been trained. In this paper, based on this characteristic of softmax function, we propose an efficient method to improve the classification performance against new types of attacks by classifying traffic with a probability below a certain level as attacks, and demonstrate the efficiency of our approach through experiments.

In this work, we implemented a fatigue detection system using machine learning and evaluated its performance. The proposed system is mainly based on the Viola-Jones face detection algorithm and the convolutional neural network (CNN). Viola-Jones object detection framework is mainly focused on the detection of the face and facial features. CNN is extended to DenseNets and made fully convolution to tackle the problem semantic image segmentation. The main idea behind the DenseNets is to capture the dense blocks that perform iterative concatenation of feature maps. The proposed system is implemented on many different video sequences and observed that its average accuracy is 99.18% and the detection rate is 99.71% out of approximately 100 image frames. The system shows high accuracy in segmentation, low error rate, and quick processing of input data distinguishes from the existing similar systems. Finally, if this system is implemented, it can minimize the number of accidents caused by drivers' fatigue.

본 논문은 와이브로 시스템 환경에서 MIMO 이동통신을 STC 방법을 이용해 성능분석하였다. 본 논문에서는 3개의 알고리즘, SM, ML 그리고 ZF 방식을 적용하여 성능 분석 하였다. 다양한 성능분석을 수행한 결과 ZF 방식이 다른 두가지 방식인 STC, ML 방식보다 성능이 우수함을 확인할 수 있었다.

This paper shows various perfermance analysis utilizing STC(Space Time Coding) in MIMO mobile communication by WiBro system environments. In this paper, 3 algorithm which are SM method, ML (Maximum Likelihood) and ZF(Zero Forcing) algorithm use for perfermance analysis. From the various simulation result, it is confirm that ZF method is superior compare than STC and ML method.

집단 비교 연구 시 측정동일성의 성립 여부는 집단 간 의미 있는 비교를 하기 위한 필수 요건으로 제시되고 있다. 이를 위해 일반적으로 다집단 확인적요인분석(MG CFA)이 널리 사용되어 왔으나, MG CFA는 비교집단이 소수일 경우에 적합한 것으로, 많은 집단을 비교하게 되는 국가 비교 연구에서는 그 한계가 제기된다. 따라서 본 연구에서는 10개 이상의 많은 수의 국가(혹은 집단)를 비교하기에 적절한 방법론인 다층 확인적요인분석(ML CFA)과 다층 요인혼합모형(ML FMM)을 이용한 분석 방법론을 기술하였다. ML CFA는 절편만을 임의효과로 추정하는 임의절편모형과 요인계수도 임의효과로 추정하는 임의절편 및 임의요인계수모형으로 구분하여 장단점을 기술하였다. 구체적으로 각 방법론에 대한 이론적 모형과 측정동일성 검정 절차를 제시하고, 기존의 MG CFA에 비해 지니는 이점 및 적용 시 유의해야 할 사항을 서술하였다. 또한 이러한 방법론을 적용한 예시로서, PISA 2015 자료를 활용하여 학생이 인식한 과학의 도구적 동기 및 즐거움에 대해 국가별 측정동일성 검정 절차를 분석하고 국가별 잠재평균을 추정하였다. 마지막으로 본 연구의 향후 연구 및 의의에 대해 논의하였다.

In multi-group analysis, measurement invariance is a requirement for meaningful comparisons between groups. Multi-group confirmatory factor analysis (MG CFA) has been widely used for group comparisons. However, MG CFA is appropriate for comparisons with a small number of groups and is limited for a large number of groups, in particular, in cross-cultural comparative studies. To overcome the limitation of MG CFA, this study described alternative approaches: multilevel confirmatory factor analysis (ML CFA) and multilevel factor mixture modeling (ML FMM), which are effective for comparing more than 10 groups. In ML CFA, its advantages and disadvantages were described by introducing two models: random intercept models that estimate only intercepts as random effects and random intercept and loading models that estimate intercepts and factor loadings as random effects. Specifically, this study presented theoretical models for the two methods and procedures for testing measurement invariances. In addition, this study discussed advantages of ML CFA, relative to those of MG CFA, and several points that should be considered when applying ML CFA. And, as an example of applying ML CFA, this study conducted latent means analysis across countries for instrumental motivation of science and enjoyment perceived by students using the PISA 2015 data. Finally, implications of this study and future research directions were discussed.

It is now well established that the bactericidal activity of vancomycin is not temporally stationary. As a result, clinical pharmacokinetics consultation service (CPCS) become necessary for the maintenance of the trough concentration of vancomycin to minimize adverse effect while maximizing therapeutic effect. In particular, CPCS was essential in geriatric, pediatric patients or patients over moderate stage kidney disease patients with creatinine clearance (CLcr) < 60ml/min (by National Kidney Foundation standards) for successful vancomycin therapy. However, in previous studies, CPSS was claimed to be not clinically relevant in adult patients with CLcr ≥ 60ml/min. Therefore, the objective of the present study was to evaluate the merit of CPCS and the factor(s) influencing the results of CPCS in adult patients with CLcr ≥ 60ml/min. Accordingly, the patients, from 18 to 65 years of age, with calculated CLcr values above of 60ml/min by Cockroft-Gault equation received vancomycin CPCS between January and June 2008 in Seoul National University Bundang Hospital; The electronic medical records (EMRs) of the patients were retrospectively reviewed. The patients were first classified into the maintenance dose group, the increased dose group or the reduced dose group according to CPCS results. A number of factors, such as gender, age, body weight, CLcr, duration of therapy, indication, dosage, combination of nephrotoxic drugs and occurrence of adverse effects, were studied for the potential impact on the results of CPCS. Among 132 cases, 83 cases were categorized as the maintenance dose group, 30 cases as the increased dose group, and 19 cases as the reduced dose group; The rate of patients having the dose adjustment was 37.1%. A significant (44.7%) number of patients received medications that potentially affected renal function (e.g., diuretics, aminoglycosides, amphotericin B). Approximately 6.1% of the patients experienced adverse effects (e.g., renal failure and thrombocytopenia) related to vancomycin administration. Statistical analysis indicate that age and CLcr show statistical significance between the three groups, suggesting vancomycin dose adjustment is necessary in the younger patients and/or patients having higher CLcr. When vancomycin was empirically administered in adult patients under 65 years of age with CLcr ≥ 60ml/min, the therapeutic trough range was not achieved depending on the age and the renal function, suggesting that CPCS is vital in the dose adjustment for the drug in the patient group.

구조방정식 모형을 추정할 때 일반적으로 사용되는 최대우도 방법은 자료가 정규분포를 따른다는 가정에 기반하고 있다. 그러나 심리학을 포함한 사회과학 분야에서 정규성 가정이 위배되는 사례가 빈번히 보고되고 있으며, 이러한 상황은 추정 결과에 편향을 초래하여 통계적 추론의 타당성을 저하시킨다. 이에 정규성 가정이 위배된 상황에서도 신뢰할 수 있는 결과를 주는 여러 대안적인 방법들이 탐색되어 왔으나, 방법별 수행도가 연구마다 일관적이지 않아 적절한 추정 방법을 선택하기 위한 기준이 명확하지 않은 상황이다. 따라서 본 연구는 정규성 가정 위배 시 발생하는 문제에 대응할 수 있는 대안적인 방법을 정리하고, 방법별 수행도를 비교․제시하기 위해 지난 30여 년간의 관련 연구를 통합하여 연구자들이 실질적으로 참고할 수 있는 지침을 제안하고자 한다. 먼저, 최대우도 방법에서 정규성 가정의 의미와 가정 위배가 추정 결과에 미치는 영향을 설명한다. 다음으로, 정규성 가정이 위배된 상황에서 활용 가능한 다양한 방법들을 소개하고, 이들 방법이 비정규성에 대응하는 원리를 논의한다. 나아가, 기존 연구들을 체계적으로 탐색한 후 연구 결과를 조건별로 분류하고, 이를 표와 그림으로 시각화하여 각 방법의 수행도를 비교하고 논의한다. 마지막으로, 위에서 논의된 내용을 종합한 가이드라인을 제공하면서 본 연구의 의의와 한계에 관해 논한다.

Maximum likelihood (ML), which is commonly used to estimate structural equation models, is based on the assumption of normality in the data. However, violations of the normality assumption are frequently reported in psychology and the social sciences, which can lead to biased estimation results and undermine the validity of statistical inferences. Although alternative methods that can provide reliable results under non-normal conditions have been explored, the performance of these methods has shown inconsistent patterns across studies, making it difficult to establish clear criteria for selecting appropriate methods. This study aims to address the problems posed by violations of the normality assumption and to explore alternative methods for dealing effectively with such violations. By integrating studies from the last 30 years of research, the study attempts to provide practical guidelines for researchers confronted with non-normality in their data. It first discusses the importance of the normality assumption in ML and examines the impact of its violation on estimation results. It then presents several alternative methods that are applicable under non-normal conditions and analyses the principles by which these methods deal with non-normality. Furthermore, previously published studies are systematically reviewed and categorized according to specific conditions, with the results visualized through tables and figures to compare the performance of different methods. Finally, the study integrates these discussions to propose guidelines for researchers and highlight their implications and limitations.

Artificial Intelligence (AI) and Machine Learning (ML) technologies are becoming more important in wireless telecommunications networks, especially in the transition from 5G to 6G, a more advanced AI networking environment. While in 5G networks AI is used basically to get better performance from the individual tasks, in 6G, AI will be a model that is used at each layer of the system design-from the physical retransmission of the signals right through to the management of the services. The paper will examine the advanced AI technologies of Deep Learning, Reinforcement Learning, Generative Models, and Federated Learning, and their impact on core processes in the networking framework like beamforming, channel estimation, spectrum access, and anomaly detection which are evaluated against core metrics of accuracy, latency, power consumption, privacy, and comprehensibility. In the process of going beyond technical detail, the review situates AI-based wireless innovations in different fields including autonomous vehicles, telesurgery, industrial IoT, and smart cities. It also points out the persistent challenges, such as data scarcity, real-time inference, edge deployment, and ethical concerns, and presents some promising future research directions, including digital twins, AI?quantum convergence, and regulatory frameworks. This work presents a strategic roadmap to achieve scalable, secure, and intelligent 6G networks by providing a cross-layer and cross-domain synthesis.

구조적 MRI 영상은 여러 단 변량과 다변량 방법을 위해 그레이 메터 (GM), 화이트 메터 (WM), 뇌척수액 (CSF) 세션화 과정을 하고 난후 형태계측학적 특징을 추출하기 위해 사용한다. 새로운 접근 방법은 매우 가벼운알츠하이머 병에서 가벼운 알츠하이머병의 진단을 위해 적용된다. 간이정신상태검사에 따른 형태계측학적 특징과가우시안 복합 모델 파라미터를 결합하여 정상인으로부터 알츠하이머 병 환자로 분류하는 방법을 제안한다. 결합한특징은 주성분 분석 기법을 이용한 고차원의 저주를 제거한 후 다중 커널 SVM 분류기에 공급한다. 제안한 진단방법의 실험적 결과는 90%이상의 특성도와 고민감도에 따라 다중 커널 SVM을 가진 층화 정확도가 96%까지 최대산출한다.

Structural MRI(sMRI) imaging is used to extract morphometric features after Grey Matter(GM), White Matter(WM) for several univariate and multivariate method, and Cerebro-spinal Fluid (CSF) segmentation. A new approach is applied for the diagnosis of very mild to mild AD. We propose the classification method of Alzheimer disease patients from normal controls by combining morphometric features and Gaussian Mixture Models parameters along with MMSE (Mini Mental State Examination) score. The combined features are fed into Multi-kernel SVM classifier after getting rid of curse of dimensionality using principal component analysis. The experimenral results of the proposed diagnosis method yield up to 96% stratification accuracy with Multi-kernel SVM along with high sensitivity and specificity above 90%.

양자 컴퓨터의 발전으로 기존 공개키 암호체계의 안전성이 위협받음에 따라 이에 대응할 수 있는 양자내성암호(PQC) 기술 이 주목받고 있다. 본 논문에서는 NIST PQC 표준화의 주요 후보인 ML-KEM 알고리즘을 대상으로 ARM 아키텍처의 NEON 과 RISC-V의 Vector Extension(RVV) 기반 병렬 최적화 구현 동향을 분석하였다. 각 아키텍처의 SIMD 연산 구조와 특성이 ML-KEM의 핵심 연산 최적화에 어떻게 다르게 적용되는지를 유형별로 분석하여 아키텍처별 최적화 전략 프레임워크를 제시 한다. 분석 결과, 고정 길이 SIMD를 사용하는 NEON은 마이크로 아키텍처 수준의 미세 조정이, 가변 길이 벡터를 지원하는 RVV는 알고리즘의 데이터 흐름에 맞는 적응형 최적화가 효과적임을 확인했다. 본 연구는 향후 다양한 환경에서 PQC를 구현 하고 최적화하려는 연구자들에게 실질적인 가이드라인을 제공한다는 점에서 의의를 가진다.

With the advancement of quantum computers, the security of existing public-key cryptosystems is being threatened, leading to increased attention on quantum-resistant cryptography (PQC) technologies that can address this issue. This paper analyzes the trends in parallel optimization implementations of the ML-KEM algorithm, a major candidate for NIST PQC standardization, based on the NEON of the ARM architecture and the Vector Extension (RVV) of RISC-V. We categorize and analyze how the distinct SIMD characteristics of each architecture are differently applied to optimize ML-KEM's core operations. Based on this synthesis, we present an architectural optimization strategy framework. The analysis confirms that NEON, with its fixed-length SIMD, benefits from microarchitectural fine-tuning, while RVV, supporting variable-length vectors, is effective with adaptive optimizations tailored to algorithmic data flow. This study is significant in that it provides a practical guideline for researchers implementing and optimizing PQC on diverse hardware platforms.

The question of how to analyze unbalanced hierarchical data generated from structural equation models has been a common problem for researchers and analysts. Among difficulties plaguing statistical modeling are removing estimation bias due to measurement error and incorporating variability associated with the social milieu in which individuals are situated. This paper presents empirical Bayes estimation by means of the EM algorithm in the context of unbalanced sampling designs. The EM algorithm is particularly useful when the analytic expressions exist for the conditional expectations of the missing data given complete data and for the maximum likelihood estimators (MLE) of the model parameters. The accuracy of the algorithm was tested using a set of artificial data. The numerical results suggest that this new methodology is a useful mean for studying hypothesized relations among latent variables varying at two levels of hierarchy.

본 연구는 인공지능(머신러닝·딥러닝) 기법을 아이폰(iOS) 디지털 포렌식 절차에 적용하여 삭제·손상 데이터의 복구 효율성과 정확도를 향상시키는 방안을 제안한다. iOS의 강력한 보안 구조는 전통적인 수동 분석 및 단순 복구 방식의 한계를 점점 확대시키고 있으며, 이에 따라 보다 정밀하고 신뢰성 있는 복구 기법이 요구되고 있다. 이를 해결하기 위해 본 연구는 ML 트리아지 기반의 자동 분류 · 우선순위화와 DL 인페인팅 기반의 시각적 복원을 결합한 AI 융합 복구 파이프라인을 제시하고, 그 파이프라인에서 ML 트리아지 기법이 가지는 설명 가능성 부족, 데이터 편향, 후보 선정 불안정성 등의 한계를 개선하기 위해 XAI(Explainable AI) 적용, 데이터 다양성 확보 및 정기적 재학습, Top-k 우선순위화 알고리즘 고도화 방안을 포함하였다. 또한 APFS의 기술적 제약과 복구 가능성을 분석하고, 표준화된 AI 복구 절차 및 평가지표를 제시함으로써 처리 시간을 단축하고, 복구 정확도 및 법적 증거 확보의 신뢰성을 강화하고자 한다. 본 연구의 결과는 향후 아이폰 디지털 포렌식 자동화 및 고도화에 기여할 것으로 기대된다.

This study proposes an advanced approach to enhance the efficiency and accuracy of recovering deleted or damaged data in iPhone (iOS) digital forensics by integrating artificial intelligence techniques, specifically machine learning (ML) and deep learning (DL). The robust security architecture of iOS has increasingly amplified the limitations of conventional manual and simple recovery methods, creating the need for more precise and reliable forensic solutions. To address these challenges, this research presents an AI-integrated recovery pipeline that combines ML triage-based automatic classification and prioritization with DL inpainting-based visual restoration. In addition, to mitigate the shortcomings of conventional ML triage—such as insufficient explainability, data bias, and instability in Top-k candidate selection—this study incorporates Explainable AI (XAI), ensures data diversity through periodic retraining, and introduces an enhanced hybrid Top-k prioritization algorithm. Furthermore, by analyzing the technical constraints and recovery potential of the Apple File System (APFS), the study proposes standardized AI-driven recovery procedures and evaluation metrics. The proposed framework aims to reduce processing time, improve recovery accuracy, and reinforce the legal reliability and transparency of digital evidence in iOS forensic analysis.

Driven by stringent global CO2 emission regulations, the demand for offshore wind energy is rapidly expanding, necessitating structural materials with higher durability and strength for larger installations. Following our previous evaluation of S355ML steel welds (Part I), this study investigates the mechanical performance of Flux-Cored Arc Welding (FCAW) on S420ML steel. This higher-strength TMCP steel is increasingly preferred for large-scale offshore substructures to achieve enhanced structural efficiency and significant weight reduction. In this research, we evaluated the weldability and mechanical integrity of S420ML joints under optimized FCAW conditions. The experimental results demonstrated that the welded joints achieved a tensile strength of 598 MPa, exceeding the minimum requirements of the S420ML base metal and ensuring a joint efficiency of over 102%. Notably, Charpy V-notch impact tests at -50°C exhibited an average absorbed energy of 112.9 J, which significantly surpasses the standard requirement for offshore structures and demonstrates excellent low-temperature toughness. These findings provide critical technical validation for the application of S420ML steel in the next generation of offshore wind foundations.

As global greenhouse gas reduction regulations are strengthened and the demand for eco-friendly energy increases, renewable energies, including offshore wind power, are growing rapidly. Unlike onshore wind power generation, offshore wind power is located in the ocean. As a result, the offshore wind power substructure is exposed to low temperatures, corrosion, and continuous fatigue loads. Therefore, selecting appropriate materials and welding techniques is crucial for durability. In this study, FCAW welding was performed on S355ML steel (EN10025) for offshore wind power applications. After the welding process, the mechanical properties of the welded joint were evaluated through tensile, low-temperature impact, and hardness tests to assess the welding condition. The study revealed that the tensile and yield strength of the welded joint were superior to those of the base material. Additionally, the impact strength at low temperatures was confirmed to exceed the standard.

Fouling in heat exchangers (HXs) affects various industries by lowering efficiency and increasing costs. Traditional fouling-prediction models often do not reflect important mechanistic information and thus become very complex and less reliable. The applications of artificial intelligence (AI), machine learning (ML), and deep learning (DL) open new frontiers, as these techniques can model complex correlations and work with large volumes of data. This review synthesizes 51 articles published between 2005 and June 2025, covering important trends, research shortcomings, and proposals. Models such as artificial neural networks (ANNs)/deep neural networks (DNNs) and Gaussian process regression (GPR) deliver the finest results in terms of accurate prediction.