Efficient Deep Learning Research Articles

An efficient deep learning framework for mixed-type wafer map defect pattern recognition

Defect detection on wafers holds immense significance in producing micro- and nano-semiconductors. As manufacturing processes grow in complexity, wafer maps may display a mixture of defect types, necessitating the utilization of more intricate deep learning models for effective feature learning. However, sophisticated models come with a demand for substantial computational resources. In this paper, we propose an efficient deep learning framework designed explicitly for mix-type wafer map defect pattern recognition. Our proposed model incorporates several crucial design elements, including lightweight convolutions, bottleneck residual connections, efficient channel attention mechanisms, and optimized activation functions, enabling it to learn spatial and channel features efficiently. We conduct evaluations on a real-world dataset containing 38 mixed-type defect patterns. The experimental results demonstrate that our framework maintains a high level of accuracy while possessing a compact parameter size and remarkably fast inference speed.

An accurate and efficient deep learning model for spatio-temporal prediction of interfacial flows

Aiming at fast and accurate spatiotemporal prediction of interfacial flow fields, a novel deep learning model combining Convolutional AutoEncoder (CAE) and long short-term memory with spatial and temporal attention (LSTM-STA) is proposed and named as CAE-LSTM-STA (hybrid model of CAE and LSTM-STA) in this article. To enable a fast calculation process, CAE is first utilized to compress the high-dimensional snapshots of flow fields into the low-dimensional latent space by its encoder. Then, the latent space serves as the input feature for LSTM-STA to temporally evolve the state of the low-dimensional latent space. Finally, the low-dimensional latent space at future time steps is fed back into the decoder of CAE to derive the full-order snapshots of the flow fields. The prediction performance of the proposed model is evaluated via two representative benchmark cases, including the dam break case and the rising bubble case. The capability of CAE in dimension reduction is found to be much better than that of the widely used Proper Orthogonal Decomposition, while LSTM-STA is observed that outperforms the original LSTM in multivariant temporal prediction. These promising results indicate that the proposed CAE-LSTM-STA model is able to effectively capture and advance the spatiotemporal characteristics of the interfacial flow fields, therefore making it an advanced surrogate model for fast and precise generation of the temporally continuous interfacial flow fields.

Efficient feature extraction and hybrid deep learning for early identification of uterine fibroids in ultrasound images

AbstractNon‐cancerous growths called uterine fibroids develop in the uterus. They can vary in size, location, and number, and can produce symptoms including excessive menstrual flow, pelvic discomfort, and reproductive problems. Early detection of uterine fibroids is important because it allows for timely intervention and appropriate management strategies. Extracting meaningful features from ultrasound (US) images requires robust and effective techniques. However, different feature extraction methods may yield varying results, and the choice of technique can influence the accuracy of fibroid detection. This paper presents an efficient approach for early detection of uterine fibroids in US images. Our proposed technique combines efficient feature extraction methods and a hybrid deep learning approach. Firstly, we employ the well‐known canny edge detection algorithm to accurately identify the edges of the fibroid region in the US images. This step helps in segmenting the target edge portion for further analysis. Additionally, we introduce an improved bird swarm optimization (IBSO) algorithm to extract a comprehensive set of features, utilizing both known features and newly obtained features. This approach enhances the accuracy of fibroid detection in uterine images. The proposed method uses a convolutional recurrent neural network (CRNN) to accurately detect and treat uterine fibroids, ensuring timely diagnosis and treatment. Our proposed IBSO‐CRNN model produces accuracy rates of 99.897% and 97.568% on the augmented and original datasets, respectively.

Automatic Liver Segmentation from Multiphase CT Using Modified SegNet and ASPP Module

Liver cancer is one of the dominant causes of cancer death worldwide. Computed Tomography (CT) is the commonly used imaging modality for diagnosing it. Computer-based liver cancer diagnosis systems can assist radiologists in image interpretation and improve diagnosis speed and accuracy. Since liver segmentation is crucial to such systems, researchers are relentlessly pursuing various segmentation approaches. A clinically viable computer-aided system requires examining multiphase CT images. However, most of the research focuses only on the portal venous phase. In this work, we developed an automatic and efficient Deep Learning (DL) method using SegNet, atrous spatial pyramid pooling module and leaky ReLU layers for liver segmentation from quadriphasic abdominal CT volumes. The proposed method was validated on two datasets, an internal institutional dataset consisting of multiphase CT and a public dataset of portal venous phase CT volumes. The Dice Coefficients (DC) obtained were greater than 96% for the latter dataset and the portal venous phase of the former. For arterial, delayed and plain CT phases of the former dataset, the DC achieved were 94.61%, 95.01% and 93.23%, respectively. Experiments showed that our model performed better than the other state-of-the-art DL models. Ablation studies have revealed that the proposed model leverages the strengths of all the three components that make it up. The promising performance of the proposed method suggests that it is appropriate for incorporation in hepatic cancer diagnosis systems.

Nucleus segmentation from the histopathological images of liver cancer through an efficient deep learning framework

Nucleus segmentation from the histopathological images of liver cancer through an efficient deep learning framework

Landet: an efficient physics-informed deep learning approach for automatic detection of anatomical landmarks and measurement of spinopelvic alignment

Purpose:An efficient physics-informed deep learning approach for extracting spinopelvic measures from X-ray images is introduced and its performance is evaluated against manual annotations.Methods:Two datasets, comprising a total of 1470 images, were collected to evaluate the model’s performance. We propose a novel method of detecting landmarks as objects, incorporating their relationships as constraints (LanDet). Using this approach, we trained our deep learning model to extract five spine and pelvis measures: Sacrum Slope (SS), Pelvic Tilt (PT), Pelvic Incidence (PI), Lumbar Lordosis (LL), and Sagittal Vertical Axis (SVA). The results were compared to manually labelled test dataset (GT) as well as measures annotated separately by three surgeons.Results:The LanDet model was evaluated on the two datasets separately and on an extended dataset combining both. The final accuracy for each measure is reported in terms of Mean Absolute Error (MAE), Standard Deviation (SD), and R Pearson correlation coefficient as follows: [SS∘:3.7(2.7),R=0.89]\\documentclass[12pt]{minimal} \\usepackage{amsmath} \\usepackage{wasysym} \\usepackage{amsfonts} \\usepackage{amssymb} \\usepackage{amsbsy} \\usepackage{mathrsfs} \\usepackage{upgreek} \\setlength{\\oddsidemargin}{-69pt} \\begin{document}$$[SS^\\circ : 3.7 (2.7), R = 0.89]$$\\end{document}, [PT∘:1.3(1.1),R=0.98],[PI∘:4.2(3.1),R=0.93],[LL∘:5.1(6.4),R=0.83],[SVA(mm):2.1(1.9),R=0.96]\\documentclass[12pt]{minimal} \\usepackage{amsmath} \\usepackage{wasysym} \\usepackage{amsfonts} \\usepackage{amssymb} \\usepackage{amsbsy} \\usepackage{mathrsfs} \\usepackage{upgreek} \\setlength{\\oddsidemargin}{-69pt} \\begin{document}$$[PT^\\circ : 1.3 (1.1), R = 0.98], [PI^\\circ : 4.2 (3.1), R = 0.93], [LL^\\circ : 5.1 (6.4), R=0.83], [SVA(mm): 2.1 (1.9), R = 0.96]$$\\end{document}. To assess model reliability and compare it against surgeons, the intraclass correlation coefficient (ICC) metric is used. The model demonstrated better consistency with surgeons with all values over 0.88 compared to what was previously reported in the literature.Conclusion:The LanDet model exhibits competitive performance compared to existing literature. The effectiveness of the physics-informed constraint method, utilized in our landmark detection as object algorithm, is highlighted. Furthermore, we addressed the limitations of heatmap-based methods for anatomical landmark detection and tackled issues related to mis-identifying of similar or adjacent landmarks instead of intended landmark using this novel approach.

Resource Efficient Deep Learning Hardware Watermarks with Signature Alignment

Deep learning intellectual properties (IPs) are high-value assets that are frequently susceptible to theft. This vulnerability has led to significant interest in defending the field's intellectual properties from theft. Recently, watermarking techniques have been extended to protect deep learning hardware from privacy. These technique embed modifications that change the hardware's behavior when activated. In this work, we propose the first method for embedding watermarks in deep learning hardware that incorporates the owner's key samples into the embedding methodology. This improves our watermarks' reliability and efficiency in identifying the hardware over those generated using randomly selected key samples. Our experimental results demonstrate that by considering the target key samples when generating the hardware modifications, we can significantly increase the embedding success rate while targeting fewer functional blocks, decreasing the required hardware overhead needed to defend it.

Human Activity Recognition Based On Video Summarization And Deep Convolutional Neural Network

Abstract In this technological era, human activity recognition (HAR) plays a significant role in several applications like surveillance, health services, Internet of Things, etc. Recent advancements in deep learning and video summarization have motivated us to integrate these techniques for HAR. This paper introduces a computationally efficient HAR technique based on a deep learning framework, which works well in realistic and multi-view environments. Deep convolutional neural networks (DCNNs) normally suffer from different constraints, including data size dependencies, computational complexity, overfitting, training challenges and vanishing gradients. Additionally, with the use of advanced mobile vision devices, the demand for computationally efficient HAR algorithms with the requirement of limited computational resources is high. To address these issues, we used integration of DCNN with video summarization using keyframes. The proposed technique offers a solution that enhances performance with efficient resource utilization. For this, first, we designed a lightweight and computationally efficient deep learning architecture based on the concept of identity skip connections (features reusability), which preserves the gradient loss attenuation and can handle the enormous complexity of activity classes. Subsequently, we employed an efficient keyframe extraction technique to minimize redundancy and succinctly encapsulate the entire video content in a lesser number of frames. To evaluate the efficacy of the proposed method, we performed the experimentation on several publicly available datasets. The performance of the proposed method is measured in terms of evaluation parameters Precision, Recall, F-Measure and Classification Accuracy. The experimental results demonstrated the superiority of the presented algorithm over other existing state-of-the-art methods.

Hybrid Deep Neural Networks with Multi-Tasking for Rice Yield Prediction Using Remote Sensing Data

Recently, data-driven approaches have become the dominant solution for prediction problems in agricultural industries. Several deep learning models have been applied to crop yield prediction in smart farming. In this paper, we proposed an efficient hybrid deep learning model that coordinates the outcomes of a classification model and a regression model in deep learning via the shared layers to predict the rice crop yield. Three statistical analyses on the features, including Pearson correlation coefficients (PCC), Shapley additive explanations (SHAP), and recursive feature elimination with cross-validation (RFECV), are proposed to select the most relevant ones for the predictive goal to reduce the model training time. The data preprocessing normalizes the features of the collected data into specific ranges of values and then reformats them into a three-dimensional matrix. As a result, the root-mean-square error (RMSE) of the proposed model in rice yield prediction has achieved 344.56 and an R-squared of 0.64. The overall performance of the proposed model is better than the other deep learning models, such as the multi-parametric deep neural networks (MDNNs) (i.e., RMSE = 370.80, R-squared = 0.59) and the artificial neural networks (ANNs) (i.e., RMSE = 550.03, R-squared = 0.09). The proposed model has demonstrated significant improvement in the predictive results of distinguishing high yield from low yield with 90% accuracy and 94% F1 score.

An efficient deep learning mechanisms for IoT/Non-IoT devices classification and attack detection in SDN-enabled smart environment

In recent years, the development of Internet of Things (IoT) applications has increased, resulting in higher demands for sufficient bandwidth, data rates, latency, and quality of service (QoS). In advanced communications, managing network resources for allocating IoT services and identifying the exact IoT devices connected to a network is a major concern. The existing studies have introduced various methods for classifying IoT devices in a network. However, the previous studies faced challenges like limited attributes, low efficiency, inappropriate features, and computational complexities. Also, the existing studies failed to concentrate on IoT/Non-IoT classification along with attack detection. Detecting attacks on IoT devices is critical for making network services more effective. Thus, the proposed study introduces an efficient IoT device classification and attack detection mechanism using software defined networking (SDN)-enabled fiber-wireless access networks internet of things (FiWi IoT) architecture. Initially, an effective resource allocation process is performed to mitigate the delay constraint issues by introducing a hybrid parallel neural network-based dynamic bandwidth allocation (DBA) method. Then, the input traffic information is gathered from the resource-efficient SDN-enabled FiWi IoT network, and the input data is pre-processed to eliminate unwanted noises using min-max normalization and standardization. Next, the essential attributes are extracted to attain enhanced classification performance. To reduce the feature dimensionality problem and thereby solve complexity issues, the most optimal features are selected by a new chaotic seagull optimization (CSO) approach. After that, IoT/non-IoT classification is performed using a transformer-driven deep intelligent model. Finally, the attacks are detected and classified by introducing a novel slice attention-based deep capsule autoencoder (SA_DCAE) model. For experimentation, the Python 3.7.0 tool is used in this work, and the performance of proposed classifiers is measured by evaluating varied matrices. Also, the comparison analysis proves the superiority of the proposed techniques to other existing methods.

Grassroots operator search for model edge adaptation using mathematical search space

Hardware-aware Neural Architecture Search (HW-NAS) is increasingly being used to design efficient deep learning architectures. An efficient and flexible search space is crucial to the success of HW-NAS. Current approaches focus on designing a macro-architecture and searching for the architecture’s hyperparameters based on a set of possible values. This approach is biased by the expertise of deep learning (DL) engineers and standard modeling approaches. In this paper, we present a Grassroots Operator Search (GOS) methodology. Our HW-NAS adapts a given model for edge devices by searching for efficient operator replacement. We express each operator as a set of mathematical instructions that capture its behavior. The mathematical instructions are then used as the basis for searching and selecting efficient replacement operators that maintain the accuracy of the original model while reducing computational complexity. Our approach is grassroots since it relies on the mathematical foundations to construct new and efficient operators for DL architectures. We demonstrate on various DL models, that our method consistently outperforms the original models on two edge devices, namely Redmi Note 7S and Raspberry Pi3, with a minimum of 2.2x speedup while maintaining high accuracy. Additionally, we showcase a use case of our GOS approach in pulse rate estimation on wristband devices, where we achieve state-of-the-art performance, while maintaining reduced computational complexity, demonstrating the effectiveness of our approach in practical applications.

An efficient deep learning architecture for effective fire detection in smart surveillance

The threat of fire is pervasive, poses significant risks to the environment, and may include potential fatalities, property devastation, and socioeconomic disruption. Successfully mitigating these risks relies on the prompt identification of fires, a process in which soft computing methodologies play a pivotal role. Although, these fire detection methodologies neglected to explore the relationships among fire-indicative features, which are important to enable a model to learn more representative and robust features in remote sensing scenarios. In the context of small fire detection from aerial view using satellite imagery or unmanned arial vehicle (UAVs) presents challenges to capture rich spatial detail, hinder the model ability for accurate fire scene classification. Furthermore, it is significant to manage model complexity effectively to facilitate deployment on UAVs for fast and accurate responses in an emergency situation. To cope with these challenges, we propose an advanced model integrated a modified soft attention mechanism (MSAM) and a 3D convolution operation with a MobileNet architecture to overcome obstacles related to optimising features and controlling model complexity. The MSAM enabling the model to selectively emphasise essential features during the training process which acts as a selective filter. This adaptive attention mechanism enhances sensitivity and allowing the model to prioritise relevant patterns for accurate fire detection. Concurrently, the integration of a 3D convolutional operation extends the model spatial awareness, to capture intricate details across multiple scales, and particularly in small regions observed from aerial viewpoints. Benchmark evaluations of the proposed model over the FD, DFAN, and ADSF datasets reveal superior performance with enhanced accuracy (ACR) compared to existing methods. Our model surpassed the state-of-the-art models with an average ACR improvement of 0.54%, 2.64%, and 1.20% on the FD, ADSF, and DFAN datasets, respectively. Furthermore, the use of an explainable artificial intelligence (XAI) technique enhances the validation of the model visual emphasis on critical regions of the image, providing valuable insights into its decision-making process.

A novel deep learning model for detection of inconsistency in e-commerce websites

On most e-commerce websites, there are two crucial factors that customers rely on to assess product quality and dependability: customer reviews provided online and related ratings. Reviews offer feedback to customers about the product’s merits, reasons for negative reviews, and feelings of satisfaction or dissatisfaction with the provided service. As for ratings, they express customer opinions about the product’s quality as numerical values from one to five (one or two for the worst opinion, three for the neutral opinion, and four or five for the best opinion). Usually, the customer reviews may be inconsistent with their relevant ratings; the customer may write the worst review despite providing a four- or five-star rating or write the best review with only a one- or two-star rating. Due to this inconsistency, customers may need help to identify relevant information. Therefore, it is required to develop a model that can classify reviews as either positive or negative, depending on the polarity of thoughts, to demonstrate if there is an inconsistency between customer reviews and their actual ratings by comparing them with the ratings resulting from the model. This paper proposes an efficient deep learning (DL) model for classifying customer reviews and assessing whether there is inconsistency. The recommended model’s performance and stability are examined on a large dataset of product reviews from Amazon e-commerce. The experimental findings showed that the proposed model dominates and significantly outperforms its peers regarding prediction accuracy and other performance measures.

Text Detection and Recognition for X-ray Weld Seam Images

X-ray weld seam images carry vital information about welds. Leveraging graphic–text recognition technology enables intelligent data collection in complex industrial environments, promising significant improvements in work efficiency. This study focuses on using deep learning methods to enhance the accuracy and efficiency of detecting weld seam information. We began by actively gathering a dataset of X-ray weld seam images for model training and evaluation. The study comprises two main components: text detection and text recognition. For text detection, we employed a model based on the DBNet algorithm and tailored post-processing techniques to the unique features of weld seam images. Through model training, we achieved efficient detection of the text regions, with 91% precision, 92.4% recall, and a 91.7% F1 score on the test dataset. In the text recognition phase, we introduced modules like CA, CBAM, and HFA to capture the character position information and global text features effectively. This optimization led to a remarkable text line recognition accuracy of 93.4%. In conclusion, our study provides an efficient deep learning solution for text detection and recognition in X-ray weld seam images, offering robust support for weld seam information collection in industrial manufacturing.

AM-SegNet for additive manufacturing in situ X-ray image segmentation and feature quantification

ABSTRACT Synchrotron X-ray imaging has been utilised to detect the dynamic behaviour of molten pools during the metal additive manufacturing (AM) process, where a substantial amount of imaging data is generated. Here, we develop an efficient and robust deep learning model, AM-SegNet, for segmenting and quantifying high-resolution X-ray images and prepare a large-scale database consisting of over 10,000 pixel-labelled images for model training and testing. AM-SegNet incorporates a lightweight convolution block and a customised attention mechanism, capable of performing semantic segmentation with high accuracy (∼96%) and processing speed (< 4 ms per frame). The segmentation results can be used for quantification and multi-modal correlation analysis of critical features (e.g. keyholes and pores). Additionally, the application of AM-SegNet to other advanced manufacturing processes is demonstrated. The proposed method will enable end-users in the manufacturing and imaging domains to accelerate data processing from collection to analytics, and provide insights into the processes’ governing physics.

Efficient-gastro: optimized EfficientNet model for the detection of gastrointestinal disorders using transfer learning and wireless capsule endoscopy images.

Gastrointestinal diseases cause around two million deaths globally. Wireless capsule endoscopy is a recent advancement in medical imaging, but manual diagnosis is challenging due to the large number of images generated. This has led to research into computer-assisted methodologies for diagnosing these images. Endoscopy produces thousands of frames for each patient, making manual examination difficult, laborious, and error-prone. An automated approach is essential to speed up the diagnosis process, reduce costs, and potentially save lives. This study proposes transfer learning-based efficient deep learning methods for detecting gastrointestinal disorders from multiple modalities, aiming to detect gastrointestinal diseases with superior accuracy and reduce the efforts and costs of medical experts. The Kvasir eight-class dataset was used for the experiment, where endoscopic images were preprocessed and enriched with augmentation techniques. An EfficientNet model was optimized via transfer learning and fine tuning, and the model was compared to the most widely used pre-trained deep learning models. The model's efficacy was tested on another independent endoscopic dataset to prove its robustness and reliability.

A blockchain-based deep learning approach for cyber security in next-generation medical cyber-physical systems

&lt;p&gt;Cyber-physical systems (CPSs) have been employed to seamlessly integrate numerous processes and physical components with integrated computing facilities and data storage, aiming to achieve a heightened level of effectiveness and efficiency across various qualitative and quantitative metrics, including technical and organizational aspects. The increased use of the web and the prospering network through IoT (Internet of things) have given a critical open door to CPS to prevail. While this innovation is as of now utilized in programmed pilot flying, advanced mechanics frameworks, clinical checking, modern control frameworks, and so forth, the headway of these frameworks should understand undeniable spotlight on making them proficient and secure. To work on the strength, reliability, and security of these frameworks, specialists can integrate blockchain innovation which has an inbuilt mix of consensual calculations, secure conventions, and circulated information capacity, with the CPS. This introduces an efficient deep learning approach based on blockchain for medical cyber-physical systems (CPS), consisting primarily of two components: a) a blockchain based security framework to protect the medical data and b) the extraction of quintessential features from these data to a classifier for performing the anomaly scans using deep learning. The experimental evaluation demonstrates that the suggested system outperforms existing models, achieving exceptional performance with an accuracy rate of 0.96 and a sensitivity score of 0.95.&lt;/p&gt;

Direct three-dimensional segmentation of prostate glands with nnU-Net.

In recent years, we and others have developed non-destructive methods to obtain three-dimensional (3D) pathology datasets of clinical biopsies and surgical specimens. For prostate cancer risk stratification (prognostication), standard-of-care Gleason grading is based on examining the morphology of prostate glands in thin 2D sections. This motivates us to perform 3D segmentation of prostate glands in our 3D pathology datasets for the purposes of computational analysis of 3D glandular features that could offer improved prognostic performance. To facilitate prostate cancer risk assessment, we developed a computationally efficient and accurate deep learning model for 3D gland segmentation based on open-top light-sheet microscopy datasets of human prostate biopsies stained with a fluorescent analog of hematoxylin and eosin (H&E). For 3D gland segmentation based on our H&E-analog 3D pathology datasets, we previously developed a hybrid deep learning and computer vision-based pipeline, called image translation-assisted segmentation in 3D (ITAS3D), which required a complex two-stage procedure and tedious manual optimization of parameters. To simplify this procedure, we use the 3D gland-segmentation masks previously generated by ITAS3D as training datasets for a direct end-to-end deep learning-based segmentation model, nnU-Net. The inputs to this model are 3D pathology datasets of prostate biopsies rapidly stained with an inexpensive fluorescent analog of H&E and the outputs are 3D semantic segmentation masks of the gland epithelium, gland lumen, and surrounding stromal compartments within the tissue. nnU-Net demonstrates remarkable accuracy in 3D gland segmentations even with limited training data. Moreover, compared with the previous ITAS3D pipeline, nnU-Net operation is simpler and faster, and it can maintain good accuracy even with lower-resolution inputs. Our trained DL-based 3D segmentation model will facilitate future studies to demonstrate the value of computational 3D pathology for guiding critical treatment decisions for patients with prostate cancer.

Joint resource allocation and beamforming design for RIS-aided symbiotic radio networks: A DRL approach

In this paper, we investigate a Reconfigurable Intelligent Surface (RIS)-assisted secure Symbiosis Radio (SR) network to address the information leakage of the primary transmitter (PTx) to potential eavesdroppers. Specifically, the RIS serves as a secondary transmitter in the SR network to ensure the security of the communication between the PTx and the Primary Receiver (PRx), and simultaneously transmits its information to the PTx concurrently by configuring the phase shifts. Considering the presence of multiple eavesdroppers and uncertain channels in practical scenarios, we jointly optimize the active beamforming of PTx and the phase shifts of RIS to maximize the secrecy energy efficiency of RIS-supported SR networks while satisfying the quality of service requirement and the secure communication rate. To solve this complicated non-convex stochastic optimization problem, we propose a secure beamforming method based on Proximal Policy Optimization (PPO), which is an efficient deep reinforcement learning algorithm, to find the optimal beamforming strategy against eavesdroppers. Simulation results show that the proposed PPO-based method is able to achieve fast convergence and realize the secrecy energy efficiency gain by up to 22% when compared to the considered benchmarks.

Deep learning-assisted Bayesian framework for real-time CO2 leakage locating at geologic sequestration sites

Accurate and efficient localization of CO2 leakage if occurred in subsurface formations, is of significant importance in achieving secure geological carbon sequestration (GCS) projects. However, this task is inherently challenging due to the considerable uncertainties in the subsurface. In this work, we develop a novel deep learning-assisted Bayesian framework for identifying potential CO2 leakage sites based on the reservoir pressure transient behavior measured at the wellbores of injection or observation wells. The method consists of two essential steps: 1) Deep learning surrogate: This step aims to effectively replace the intensive high-fidelity simulation with an efficient deep learning surrogate. 2) Bayesian inversion: In this step, the posterior distributions of potential CO2 leakage locations are inverted, in which the surrogate serves as the forward model. The above two processes are automated using Bayesian optimization instead of a labor-intensive trial-and-error approach. The proposed framework is verified using a 3D geological model simulating CO2 sequestration into a brine-filled reservoir. The results demonstrate the Bayesian-optimized surrogate could successfully capture the underlying process of subsurface CO2-brine flow. The Bayesian inversion algorithm enables localizing CO2 leakage with high accuracy. To our knowledge, the proposed Bayesian framework is implemented for the first time to locate multiple leakage sites at the field scale. The proposed workflow provides an accurate and efficient approach to detecting possible CO2 leakage locations in a real-time manner and has promising potential for field-scale GCS applications.