Intelligence Techniques Research Articles

Explainable AI reveals Clever Hans effects in unsupervised learning models

Abstract Unsupervised learning has become an essential building block of artifical intelligence systems. The representations it produces, for example, in foundation models, are critical to a wide variety of downstream applications. It is therefore important to carefully examine unsupervised models to ensure not only that they produce accurate predictions on the available data but also that these accurate predictions do not arise from a Clever Hans (CH) effect. Here, using specially developed explainable artifical intelligence techniques and applying them to popular representation learning and anomaly detection models for image data, we show that CH effects are widespread in unsupervised learning. In particular, through use cases on medical and industrial inspection data, we demonstrate that CH effects systematically lead to significant performance loss of downstream models under plausible dataset shifts or reweighting of different data subgroups. Our empirical findings are enriched by theoretical insights, which point to inductive biases in the unsupervised learning machine as a primary source of CH effects. Overall, our work sheds light on unexplored risks associated with practical applications of unsupervised learning and suggests ways to systematically mitigate CH effects, thereby making unsupervised learning more robust.

Use of artificial intelligence techniques in characterization of vibration signals for application in agri-food engineering

Bottling machinery is a critical component in agri-food industries, where maintaining operational efficiency is key to ensuring productivity and minimizing economic losses. Early detection of faulty conditions in this equipment can significantly improve maintenance procedures and overall system performance. This research focuses on health monitoring of gripping pliers in bottling plants, a crucial task that has traditionally relied on analyzing raw vibration signals or using narrowly defined, application-specific features. However, these methods often face challenges related to limited robustness, high computational costs, and sensitivity to noise. To address these limitations, we propose a novel approach based on generic features extracted through basic signal processing techniques applied to vibration signals. These features are then classified using a random forest algorithm, enabling an effective analysis of health states. The proposed method is evaluated against traditional approaches and demonstrates clear advantages, including higher accuracy in detecting and classifying faulty conditions, greater robustness against random perturbations, and a reduced computational cost. Additionally, the method requires fewer training instances to achieve reliable performance. This study highlights the potential of artificial intelligence and signal processing techniques in predictive maintenance, offering a scalable and efficient solution for fault detection in manufacturing processes, particularly within the agri-food sector.

Explainable Artificial Intelligence Approach to Heart Attack Risk Prediction

This study examines the feasibility of explainable artificial intelligence (XAI) techniques for analyzing and accurately classifying heart attack risks. Given the complexity of heart attack risk factors, traditional machine learning models often do not provide the transparency needed for clinical decision-making. This research addresses this gap by incorporating XAI techniques, specifically SHAP (SHapley Additive exPlanations), to reveal model predictions. In this retrospective study, multiple databases were searched, and data on eight risk factors of 1319 patients were obtained. Prediction models have been developed using six different machine learning algorithms for heart attack classification. In heart attack risk classification, the XGBoost (eXtreme Gradient Boosting) model achieved the best predictive values with 91.28% Accuracy, 90% Precision, 92% Recall, and 91% F1-score. In addition, the model algorithms were evaluated according to AUC, and again, the XGBoost model achieved the best result 0.91. In the Random Forest Feature importance evaluation, troponin was the most critical variable affecting the diagnosis. SHAP graphs showed that troponin (+4.19) was the most critical risk factor. This research highlights the potential of XAI to bridge the gap between complex AI models and clinical applicability and suggests that future studies move in a promising direction to refine further and validate AI-powered healthcare solutions.

Heat Transfer Study of a Ternary Hybrid Nanofluid Flow in a Rotating Horizontal Annulus Using Artificial Intelligence Computing Techniques: Application of Particle Swarm Optimization and Artificial Neural Network

This study investigates the influence of an axial magnetic field on the flow and Heat Transfer (HT) of a Ternary Hybrid Nanofluid in a rotating horizontal annulus between two coaxial tubes, considering the impact of viscous dissipation and thermal radiation. The study examines the use of spherical aluminum oxide (Al2O[Formula: see text] nanoparticles, platelet-shaped graphene and Cylindrical Carbon Nanotubes (CNTs). The velocity and temperature fields were transformed to get a system of Ordinary Differential Equations (ODEs) representing the flow and HT governing equations. A dataset is produced by systematically varying key parameters using the bvp4c function in MATLAB. Following this, two advanced artificial intelligence techniques, Artificial Neural Networks (ANNs) and Fuzzy Particle Swarm Optimization (FPSO), are utilized to forecast the Nusselt number. The effects of various parameters, including the radiation parameter, Reynolds number and Hartmann number on HT and flow characteristics were analyzed. Analysis of data presents that the Nusselt number increases with higher radiation parameter and falls with higher Hartmann number (Ha), Reynolds number and Eckert number. Furthermore, Nusselt number achieves higher values for Quadratic Thermal Radiation (QTR) case in comparison to Linear Thermal Radiation (LTR). It is observed that when value of radiation parameter is increased from 5.5 to 11.5, the Nusselt number values rises by 216.19% (when [Formula: see text]), 231.34% (when [Formula: see text]) and 249.82% (when [Formula: see text]) for LTR, whereas the Nusselt number values rises by 153.68% (when [Formula: see text]), 157.01% (when [Formula: see text]) and 160.70% (when [Formula: see text]) for QTR. The applications of the nanofluid flow inside a rotating horizontal annulus are seen in heat exchangers, gas turbines, rotating machinery, cylindrical solar collectors, medical devices like centrifuge and rotating drum dryers in food processing.

Evaluating techniques of artificial intelligence for social robots

Introduction. Based on techniques of artificial intelligence (AI) robots equipped with algorithms are becoming more social and intelligent as they enter society. As a relatively unexplored topic of research the current study evaluated whether the perception of robot intelligence was influenced by different techniques of AI. Method. In an online study participants viewed two versions of a humanoid robot which varied by their surface colour and stated AI abilities. For each image participants rated the perceived intelligence of the robot. Results. Using an online survey, the results found no statistically significant effect for robot surface colour on judgments of robot intelligence but that robot voice enablement and the ability to detect a user’s face and emotions added significantly to the perception of robot intelligence. In addition, amongst the AI techniques evaluated, text used for human-robot communication was the least effective method for conveying the perception of intelligence for a humanoid robot. Conclusion. As tentative conclusions, the perception of robot intelligence can be based on the specific AI technique used to design the robot, and it appears that the more the human-like AI ability of the robot, the more likely that users will view the robot as intelligent.

Optimization of coded modulation theory and algorithm for optical fiber communication incorporating biomechanical signal transduction mechanism

Optical fiber communication coding and modulation techniques play a key role in high-speed and high-capacity transmission but are still limited by problems such as signal attenuation, nonlinear effects and increasing bit error rate. In order to optimize the performance of optical communication systems, this study draws on the biomechanical signal conduction mechanism to construct an optical fiber modulation scheme that integrates pulse time coding, adaptive modulation and redundancy coding. The experimental results show that this method significantly reduces the Bit Error Rate (BER), improves the signal-to-noise ratio, and enhances the signal robustness at different transmission distances. Compared with the conventional Quadrature Phase Shift Keying (QPSK) and 16-QAM modulation, the proposed scheme reduces the BER by about 37.5% and improves the Signal-to-Noise Ratio (SNR) by 2.1 dB at 150 km transmission, which verifies its advantages in terms of interference immunity and energy utilization. The research results provide a novel optimization strategy for optical fiber communication systems and lay a theoretical foundation for the research of next-generation intelligent modulation techniques.

Fluid-structure interaction aspects of the shape-morphing structures and robots

Abstract Advancement in intelligent materials for the past few decades enabled the development of functional morphing structures and robots operating in fluid environments. Fluid-structure interaction problems for functional morphing structures and robots were naturally accentuated. In this paper, the recent advancements in shape-morphing robots and structures in fluid flow across different Reynolds number scales are reviewed and summarized, from microrobots with Reynolds number of much lower than 1 to deformable aircrafts in turbulent flows. With the quest to improve the design and functionality of the morphing structures and robots, we discuss modeling methods, experiments, and materials for the morphing structures over a vast range of Reynolds number. The importance of understanding fluid-structure interaction to design morphing structures and robots are emphasized. Following up with several critical future questions to address, the potential applications of artificial intelligence and machine learning (AI/ML) techniques in improving the design of shape-morphing structures and robots are discussed. These shape-morphing structures are expected to have a significant impact in enhancing sustainable solutions for today's challenges, and exploring unknown of deeper oceans and outer space.

ATP1A3 Acts as a Potential Anti-oncogene in Glioblastoma via the Antagonizing Interaction with Small Nuclear Ribonucleoprotein Polypeptide G.

The sodium pump α3 subunit (ATP1A3) is associated with various brain's physiological and pathological mechanisms. However, its molecular mechanisms and cellular targets in glioblastoma (GBM) are poorly understood. Bioinformatics and phosphor-proteomics analysis, target fishing experiment, confocal immunofluorescence, molecular cloning, and western blot techniques were carried out to elucidate probable downstream signaling pathways. Then GBM xenografts were established to assess potential molecular mechanisms of ATP1A3 associated with its in vivo anti-glioma impacts. The mechanistic analyses indicated that the antagonism between ATP1A3 and small nuclear ribonucleoprotein polypeptide G (SNRPG) could suppress GBM growth. ATP1A3 inhibits SNRPGinduced GBM epithelial-mesenchymal transition, and SNRPG decreases ATP1A3 by increasing phosphorylation at S643. As a negative feedback loop, ATP1A3 overexpression causes a reduction of SNRPG-induced invasion-metastasis cascades via regulating KLF9. Furthermore, by using artificial intelligence (AI) techniques, we have also exerted the design and application of a synthetic peptide (ATP1A3-S643 peptide), which could be the potential inhibitor of ATP1A3 phosphorylation. To better explore the anti-glioma effect of ATP1A3 activation, a bioengineering nanomedicine capable of ondemand ATP1A3 activator delivery to the brain for GBM has also been developed in this work, which exhibited an improved therapeutic efficacy in the ATP1A3-targeted treatment of glioma. ATP1A3 is a potential anti-glioma treatment target, and its activation critically depends on its antagonizing interaction with SNRPG.

Early Detection and Classification of Cancer Histology Images Using Artificial Intelligence Techniques: A Review

Early Detection and Classification of Cancer Histology Images Using Artificial Intelligence Techniques: A Review

Dissecting a social bot powered by generative AI: anatomy, new trends and challenges

The rise of social networks has transformed communication, information sharing and entertainment, but it has also facilitated the rise of harmful activities such as the spread of misinformation, often through the use of social bots. These automated accounts that mimic human behaviour have been implicated in significant events, including political interference and market manipulation. In this paper, we provide a comprehensive review of recent advances in social bot detection, with a particular focus on the role of generative AI and large language models. We present a new categorisation scheme for bots that aims to reduce class overlap while maintaining generality. In addition, we analyse the most commonly used datasets and state-of-the-art classification techniques, and through user profile-based measures, we use Explainable Artificial Intelligence (XAI) and data mining techniques to uncover factors that contribute to bot misclassification. Our findings contribute to the development of more robust detection methods, which are essential for mitigating the impact of malicious bots on online platforms.

Adaptive AI Framework for Anomaly Detection and DDoS Mitigation in Distributed Systems

Distributed Denial of Service (DDoS) attacks remain a major challenge for distributed systems, given the traditional detection mechanisms often fail to address the scalability and complexity of modern network architecture. This paper explores the integration of Artificial Intelligence (AI) techniques for anomaly detection and the proposed approach leverages machine learning models to detect and respond to malicious traffic patterns in real-time. This study presents a comprehensive review of AI based techniques, their application within distributed systems, and comparison against conventional methods. This study also demonstrates that AI driven anomaly detection offers effective defense against evolving DDoS threats.

Enhancing Kidney Disease Diagnosis Using ACO-Based Feature Selection and Explainable AI Techniques

Kidney disease is a global health concern, impacting a substantial part of the overall population and contributing to high morbidity and mortality rates. The initially diagnosed phases of kidney disease are often present without noticeable indications, leading to delayed diagnosis and treatment. Therefore, early detection is crucial to reducing complications and improving the lives of those impacted. However, the performance of previous automated approaches has often been hindered by suboptimal feature selection and algorithms’ “black-box” nature, which adversely affect their interpretability and clinical applicability. This paper aims to address these limitations by creating an effective machine-learning-based approach that integrates ant colony metaheuristic optimization algorithms for feature selection and explainable artificial intelligence techniques such as SHAP and LIME for model interpretation. The ant colony optimization method identified the most relevant feature subsets using a clinical dataset, reducing model complexity while preserving predictive accuracy. Performance evaluation shows that the extra trees classifier, when using optimized selected features, achieved the highest performance with an accuracy of 97.70% and an area under the curve of 99.55%, outperforming previous models trained on raw and complete processed feature sets. To enhance interpretability, the SHAP and LIME explainable techniques were employed, providing detailed insights into the contribution of key features such as TimeToEventMonths, HistoryDiabetes, and Age. This comprehensive framework, combining advanced feature selection with explainable models, improves clinical decision-making and fosters trust in machine learning applications for healthcare.

Deep learning for global horizontal irradiation estimation from MSG SEVIRI data

ABSTRACT Predicting solar irradiance using remote sensing is an important topic in renewable energy generation. Remote sensing offers the possibility of predicting solar irradiation from any location by using satellite images, consequently improving the planning and operation of photovoltaic systems. A deep Convolutional Neural Network CNN approaches are proposed to estimate a Global Horizontal Irradiation GHI from MSG SEVIRI (Meteosat Second Generation Spinning Enhanced Visible and Infrared Imager) images. In the first approach, two solar channels (0.8 and 1.6 µm) and two thermal infrared channels (3.9 and 10.8 µm) were used for this aim, and for the second one, the number in this case is increased to nine channels (0.6, 0.8, 1.6, 3.9, 6.2, 9.7, 10.8, 12, and 13.4 µm). The proposed methods were tested and evaluated using 13,760 MSG images from four channels and 30,960 images from nine channels taken at different times and dates. To give more support to our results, we made a comparison between the proposed approaches and the GisTel, which is a physical model based on the relationship between the clarity index determined from the MSG images and the global radiation that reaches the ground in a clear sky. Interesting margins of error were obtained by the use of nine channels (solar, water vapour and infrared channels). In the present work, we were able to improve the results obtained by the Gistel model by an average MAE error of 21.77 W m−2 and an RMSE of 28.88 W m−2. This proposed method leads to good GHI estimation and demonstrates its effectiveness and robustness, as well as the utility and benefit of using the artificially intelligent techniques in remote sensing imagery applications.

Preterm Birth Prediction Using an Explainable Machine Learning Approach

Pregnancy and childbirth are critical periods, with complications like preterm birth presenting serious risks to both mother and baby. Preterm birth is defined as delivery before 37 weeks of gestation. It poses serious health risks to newborns, including respiratory complications, developmental delays, and long-term disabilities. Early prediction of preterm birth can enable timely interventions to reduce these risks and improve maternal-fetal health outcomes. This paper proposes a machine learning-based framework for preterm birth prediction using a blended approach. This approach combines multiple classifiers into blended model to enhance the prediction accuracy. Feature selection techniques such as variance threshold, Pearson’s correlation, and mutual information are applied exclusively to the training set to enhance model performance. The proposed blended model consistently outperforms the standalone base models. It achieves the highest accuracy of 74.61%, precision of 71.83%, recall of 70.94%, F1 score of 71.38%, and an area under the curve (AUC) of 84.56% during 5-fold cross-validation. For holdout testing, it also maintains a superior performance with an accuracy of 73.82% and AUC of 83.91%. Additionally, explainable artificial intelligence techniques are applied to interpret the model’s predictions. It identifies that abnormal amniotic fluid, prenatal care, and household income have a substantial impact on predicting both preterm and normal birth outcomes. The proposed approach shows strong potential for real-world healthcare applications to offer clinicians valuable insights for early preterm birth intervention. Received: 10 October 2024 | Revised: 9 December 2024 | Accepted: 18 February 2025 Conflicts of Interest The authors declare that they have no conflicts of interest to this work. Data Availability Statement The data that support the findings of this study are openly available in Midwifery at https://doi.org/10.1016/j.midw.2020.102670, reference number [40]. Author Contribution Statement Ahmad Hassan: Conceptualization, Methodology, Software, Validation, Formal analysis, Investigation, Resources, Writing – original draft, Writing – review & editing, Visualization, Supervision, Project administration. Saba Nawaz: Validation, Formal analysis, Investigation, Resources, Writing – original draft, Writing – review & editing, Project administration. Sidra Tahira: Validation, Investigation, Data curation, Writing – review & editing, Visualization, Project administration. Arslan Ahmed: Investigation, Data curation, Writing – review & editing, Supervision, Project administration.

Research on the optimization of mortar mix proportion based on neural network models and genetic algorithm

Strength and durability of concrete are critical performance indicators for the safety and service life of building structures. These properties are significantly influenced by the material proportions and their microstructure. Traditional methods for designing concrete mix ratios have certain limitations when dealing with complex multivariable relationships. Therefore, intelligent mix optimization techniques have become a key focus of current research. This paper presents an optimization approach for mortar mix design based on a multi-output neural network model with a multi-head attention mechanism, combined with the genetic algorithm. Firstly, a neural network model based on the multi-head attention mechanism is developed to establish a nonlinear mapping relationship between material proportions and performance. The genetic algorithm is then applied to optimize the model’s predictions, yielding the optimal mix design. Finally, by converting the optimized mix design data into element ion ratios parameters, the correlation between these microscopic factors and cementitious materials durability is analyzed. Results show that the neural network model effectively captures complex nonlinear relationships, with the predicted strength and durability closely aligning with experimental data. The mix ratio optimized by the genetic algorithm significantly improves the strength and durability of the mortar. Furthermore, the study of ion content provides new theoretical support for enhancing concrete durability. This research not only offers an innovative solution for the intelligent optimization of concrete mix design but also lays a theoretical foundation for concrete material design and performance enhancement.

IMPLEMENTATION OF DATA SCIENCE IN PLANT ENGINEERING

ABSTRACT Effective risk and credit management are the keys to the financial health of plant engineering firms. Companies can counter market uncertainties and fluctuations in demand and complex supply chains through tailored credit risk analysis. Such analysis will help firms detect early financial vulnerabilities so that they can identify the appropriate time to take preventive measures against loss, defaults, or insolvency. More critically, sophisticated fraud detection and risk predictions are playing an important role in financial stability in plant engineering. Advanced and sophisticated business intelligence techniques can identify patterns of transactions that may represent potential fraud. With this feature, firms are in a better position to respond promptly to financial and operational risks while enhancing stability and resilience. Credit monitoring systems further give another level of security as the credit health can be put into proper perspective through integrating real-time analytics with historic financial data. Data Mining(DM) in such systems maximizes the precision of predicting risks by classifying risk levels and creating tailored credit solutions for plant engineering clients. This integration of these systems with enterprise platforms helps the firms make decisions seamlessly across departments, ultimately supporting overall efforts in risk management. Use of data science in plant engineering has improved the possibility of forecasting risks, detection of frauds, and monitoring of credit toward financial security and operational resilience. As the technology gets more integrative to the industry, it supports growth sustainability for the plant engineering companies handling finance complexities. KEYWORDS: Benchmarks, Data Science, Finance, Plant Engineering, Risk Management

Explainable artificial intelligence for botnet detection in internet of things

The proliferation of internet of things (IoT) devices has led to unprecedented connectivity and convenience. However, this increased interconnectivity has also introduced significant security challenges, particularly concerning the detection and mitigation of botnet attacks. Detecting botnet activities in IoT environments is challenging due to the diverse nature of IoT devices and the large-scale data generated. Artificial intelligence and machine learning based approaches showed great potential in IoT botnet detection. However, as these approaches continue to advance and become more complex, new questions are opened about how decisions are made using such technologies. Integrating an explainability layer into these models can increase trustworthy and transparency. This paper proposes the utilization of explainable artificial intelligence (XAI) techniques for improving the interpretability and transparency of the botnet detection process. It analyzes the impact of incorporating XAI in the botnet detection process, including enhanced model interpretability, trustworthiness, and potential for early detection of emerging botnet attack patterns. Three different XAI based techniques are presented i.e. rule extraction and distillation, local interpretable model-agnostic explanations (LIME), Shapley additive explanations (SHAP). The experimental results demonstrate the effectiveness of the proposed approach, providing valuable insights into the inner workings of the detection model and facilitating the development of robust defense mechanisms against IoT botnet attacks. The findings of this study contribute to the growing body of research on XAI in cybersecurity and offer practical guidance for securing IoT ecosystems against botnet threats.

Enhanced Model for Gestational Diabetes Mellitus Prediction Using a Fusion Technique of Multiple Algorithms with Explainability

High glucose levels during pregnancy cause Gestational Diabetes Mellitus (GDM). The risks include cesarean deliveries, long-term type 2 diabetes, fetal macrosomia, and infant respiratory distress syndrome. These risks highlight the need for accurate GDM prediction. This research proposes a novel fusion model for early GDM prediction. It uses conventional Machine Learning (ML) and advanced Deep Learning (DL) algorithms. Subsequently, it combines the strengths of both ML and DL algorithms using various ensemble techniques. It incorporates a meta-classifier that further reinforces its robust prediction performance. The dataset is split into training and testing sets in a 70/30 ratio. The initial steps involve exploratory analysis and data preprocessing techniques such as iterative imputation and feature engineering. Subsequently, oversampling is applied to the training set to address class imbalance which ensures the model learns effectively. The testing set remains imbalanced to maintain the credibility of the model’s performance evaluation. The fusion model achieves an accuracy of 98.21%, precision of 97.72%, specificity of 98.64%, recall of 97.47%, F1 score of 97.59%, and an Accuracy Under the Curve (AUC) of 99.91%. The model exhibits efficiency with an average processing time of 0.06 s to predict GDM. These results outperform the previous studies using the same GDM prediction dataset and demonstrate the model's superior performance. Additionally, Explainable Artificial Intelligence (XAI) techniques are utilized to interpret the model’s decisions. They highlight the most influential features in GDM prediction and ensures transparency. The proposed fusion model can facilitate proactive GDM prediction to elevate GDM management and maternal–fetal health outcomes.

Comparing machine learning classifier models in discriminating cognitively unimpaired older adults from three clinical cohorts in the Alzheimer’s disease spectrum: demonstration analyses in the COMPASS-ND study

BackgroundResearch in aging, impairment, and Alzheimer’s disease (AD) often requires powerful computational models for discriminating between clinical cohorts and identifying early biomarkers and key risk or protective factors. Machine Learning (ML) approaches represent a diverse set of data-driven tools for performing such tasks in big or complex datasets. We present systematic demonstration analyses to compare seven frequently used ML classifier models and two eXplainable Artificial Intelligence (XAI) techniques on multiple performance metrics for a common neurodegenerative disease dataset. The aim is to identify and characterize the best performing ML and XAI algorithms for the present data.MethodWe accessed a Canadian Consortium on Neurodegeneration in Aging dataset featuring four well-characterized cohorts: Cognitively Unimpaired (CU), Subjective Cognitive Impairment (SCI), Mild Cognitive Impairment (MCI), and AD (N = 255). All participants contributed 102 multi-modal biomarkers and risk factors. Seven ML algorithms were compared along six performance metrics in discriminating between cohorts. Two XAI algorithms were compared using five performance and five similarity metrics.ResultsAlthough all ML models performed relatively well in the extreme-cohort comparison (CU/AD), the Super Learner (SL), Random Forest (RF) and Gradient-Boosted trees (GB) algorithms excelled in the challenging near-cohort comparisons (CU/SCI). For the XAI interpretation comparison, SHapley Additive exPlanations (SHAP) generally outperformed Local Interpretable Model agnostic Explanation (LIME) in key performance properties.ConclusionThe ML results indicate that two tree-based methods (RF and GB) are reliable and effective as initial models for classification tasks involving discrete clinical aging and neurodegeneration data. In the XAI phase, SHAP performed better than LIME due to lower computational time (when applied to RF and GB) and incorporation of feature interactions, leading to more reliable results.

Axial-shunted Spatial-temporal Conversation for Change Detection

Benefitting from the maturing of intelligence techniques and advanced sensors, recent years have witnessed the full flourishing of change detection (CD) on multi-temporal remote sensing images. However, extraneous interference caused by normal tempor evolution and the extreme sparsity of spatial changes still plague the detection accuracy. To counteract this dilemma, a lightweight axial-shunted spatial-temporal conversation network (ASCNet) is proposed, which models the intrinsic representations in dually augmented images with a parallel treatment of convolutions and attentions. Specifically, for the features of weakly-augmented bi-temporal image pairs from siamese CNN, a roundtable attention-based and intra-scale axial-shunted interaction, with linear complexity, is presented. By splitting horizontally or vertically into multiple chunks and then performing axial-squeeze operation, axial-shunted scheme can achieve fine-grained attention while maintaining linear complexity. Moreover, roundtable attention pursues efficient bi-temporal modeling by incorporating both self-attention and cross-attention in a single attentional computation, while imposing change guiding and difference gating for focusing on changes. Simultaneously, a video transformer is introduced for the modeling of strongly-augmented sequences, followed by an inter-scale spatial-temporal alignment to recalibrate the feature responses. ASCNet demonstrates state-of-the-art performance on four publicly available CD datasets while maintaining superior computational efficiency. The source code is available at: https://github.com/fengyuchao97/ASCNet .