Preprocessing Strategies Research Articles

Wearing myopia glasses on GANs: Mitigating bias for pre-trained Generative Adversarial Networks via online prior perturbation

Pre-trained Generative Adversarial Networks (GANs) can provide rich information and make various downstream tasks beneficial. However, the training process of GANs potentially learns pre-existing biases in the historical data, which are not only inherited, but also amplified to some extent when using the generated data for prediction. Most available methods for bias mitigation utilize pre-processing or in-processing strategies to retrain GANs, which needs to access the raw data and also increases the time and computational cost required to train a fully expressive GAN. Therefore, we propose a novel post-processing mechanism to achieve fairness outputs of GANs through online prior perturbation. Unlike traditional output-based offline processing, our method mitigates biases from model inputs by designing a prior perturbation network, called “prior perturber”, to form a combined network with the pre-trained GAN. Specifically, we introduce a bias prediction network, called “bias predictor”, for online adversarial training with the prior perturber, which can decouple the generated representations from the bias features and improve the fairness of the downstream prediction results. In addition, the training process is independent of specific target labels, and the generated fair representations have good transferability. Experimental evaluations on five real datasets validate the effectiveness of the proposed method and the best utility is obtained on the same fairness level as compared to other output-based baseline methods. The result on fairness-utility trade-offs significantly improves the fairness of original samples by 56.40%, while the utility is reduced by only 2.60%.

Combinatorial Order Pre-processing Search (COPS): A new pre-processing strategy for large-scale interpretable data analysis in process analytical technologies

Combinatorial Order Pre-processing Search (COPS), a novel approach for optimizing data pre-processing is proposed in this work. Unlike simultaneous hyperparameter optimization, COPS employs a priori optimization to reduce computational time while refining the search space for preprocessing sequences and combinations. It allows for setting a maximum number of pre-processing methods, while efficiently searching through combinations of methods with chemically relevant knowledge. In this work, 67 calibration datasets across various analytical techniques, including fluorescence spectroscopy, gas chromatography (GC), near-infrared spectroscopy (NIR), mid-infrared spectroscopy (MIR), visible-near-infrared spectroscopy (Vis-NIR), Raman spectroscopy, nuclear magnetic resonance (NMR) spectroscopy, and voltammetry were evaluated. COPS yielded significant improvements over existing methodologies based on design of experiment and compounded pre-processing approaches. The COPS outperformed the other methods, resulting in an average root mean square error of prediction (RMSEP) reduction of 31.7%, while also reduced the complexity (number of latent variables) of the model which allows for easier interpretation. This underscores the importance of combinatorial order set theory for the search of pre-processing method combinations (without fixing the sequence of pre-processing methods) to enhance model performance and interpretation. The novel COPS approach can be employed in process analytical technology (such as inline, online or at-line chemical sensing analytics) to enhance predictive accuracy and operational efficiency, fundamentally transforming the quality and reliability of chemical process monitoring and control.

Efficient On-Board Compression for Arbitrary-Shaped Cloud-Covered Remote Sensing Images via Adaptive Filling and Controllable Quantization

Due to the fact that invalid cloud-covered regions in remote sensing images consume a considerable quantity of coding bit rates under the limited satellite-to-ground transmission rate, existing image compression methods suffer from low compression efficiency and poor reconstruction quality, especially in cloud-free regions which are generally regarded as regions of interest (ROIs). Therefore, we propose an efficient on-board compression method for remote sensing images with arbitrary-shaped clouds by leveraging the characteristics of cloudy images. Firstly, we introduce two novel spatial preprocessing strategies, namely, the optimized adaptive filling (OAF) strategy and the controllable quantization (CQ) strategy. Specifically, the OAF strategy fills each cloudy region using the contextual information at its inner and outer edge to completely remove the information of cloudy regions and minimize their coding consumption, which is suitable for images with only thick clouds. The CQ strategy implicitly identifies thin and thick clouds and rationally quantifies the data in cloudy regions to alleviate information loss in thin cloud-covered regions, which can achieve the balance between coding efficiency and reconstructed image quality and is more suitable for images containing thin clouds. Secondly, we develop an efficient coding method for a binary cloud mask to effectively save the bit rate of the side information. Our method provides the flexibility for users to choose the desired preprocessing strategy as needed and can be embedded into existing compression framework such as JPEG2000. Experimental results on the GF-1 dataset show that our method effectively reduces the coding consumption of invalid cloud-covered regions and significantly improve the compression efficiency as well as the quality of decoded images.

Recognition of Chinese Electronic Medical Records for Rehabilitation Robots: Information Fusion Classification Strategy.

Named entity recognition is a critical task in the electronic medical record management system for rehabilitation robots. Handwritten documents often contain spelling errors and illegible handwriting, and healthcare professionals frequently use different terminologies. These issues adversely affect the robot's judgment and precise operations. Additionally, the same entity can have different meanings in various contexts, leading to category inconsistencies, which further increase the system's complexity. To address these challenges, a novel medical entity recognition algorithm for Chinese electronic medical records is developed to enhance the processing and understanding capabilities of rehabilitation robots for patient data. This algorithm is based on a fusion classification strategy. Specifically, a preprocessing strategy is proposed according to clinical medical knowledge, which includes redefining entities, removing outliers, and eliminating invalid characters. Subsequently, a medical entity recognition model is developed to identify Chinese electronic medical records, thereby enhancing the data analysis capabilities of rehabilitation robots. To extract semantic information, the ALBERT network is utilized, and BILSTM and MHA networks are combined to capture the dependency relationships between words, overcoming the problem of different meanings for the same entity in different contexts. The CRF network is employed to determine the boundaries of different entities. The research results indicate that the proposed model significantly enhances the recognition accuracy of electronic medical texts by rehabilitation robots, particularly in accurately identifying entities and handling terminology diversity and contextual differences. This model effectively addresses the key challenges faced by rehabilitation robots in processing Chinese electronic medical texts, and holds important theoretical and practical value.

SkinNet-14: a deep learning framework for accurate skin cancer classification using low-resolution dermoscopy images with optimized training time

The increasing incidence of skin cancer necessitates advancements in early detection methods, where deep learning can be beneficial. This study introduces SkinNet-14, a novel deep learning model designed to classify skin cancer types using low-resolution dermoscopy images. Unlike existing models that require high-resolution images and extensive training times, SkinNet-14 leverages a modified compact convolutional transformer (CCT) architecture to effectively process 32 × 32 pixel images, significantly reducing the computational load and training duration. The framework employs several image preprocessing and augmentation strategies to enhance input image quality and balance the dataset to address class imbalances in medical datasets. The model was tested on three distinct datasets—HAM10000, ISIC and PAD—demonstrating high performance with accuracies of 97.85%, 96.00% and 98.14%, respectively, while significantly reducing the training time to 2–8 s per epoch. Compared to traditional transfer learning models, SkinNet-14 not only improves accuracy but also ensures stability even with smaller training sets. This research addresses a critical gap in automated skin cancer detection, specifically in contexts with limited resources, and highlights the capabilities of transformer-based models that are efficient in medical image analysis.

A single-cell multimodal view on gene regulatory network inference from transcriptomics and chromatin accessibility data.

Eukaryotic gene regulation is a combinatorial, dynamic, and quantitative process that plays a vital role in development and disease and can be modeled at a systems level in gene regulatory networks (GRNs). The wealth of multi-omics data measured on the same samples and even on the same cells has lifted the field of GRN inference to the next stage. Combinations of (single-cell) transcriptomics and chromatin accessibility allow the prediction of fine-grained regulatory programs that go beyond mere correlation of transcription factor and target gene expression, with enhancer GRNs (eGRNs) modeling molecular interactions between transcription factors, regulatory elements, and target genes. In this review, we highlight the key components for successful (e)GRN inference from (sc)RNA-seq and (sc)ATAC-seq data exemplified by state-of-the-art methods as well as open challenges and future developments. Moreover, we address preprocessing strategies, metacell generation and computational omics pairing, transcription factor binding site detection, and linear and three-dimensional approaches to identify chromatin interactions as well as dynamic and causal eGRN inference. We believe that the integration of transcriptomics together with epigenomics data at a single-cell level is the new standard for mechanistic network inference, and that it can be further advanced with integrating additional omics layers and spatiotemporal data, as well as with shifting the focus towards more quantitative and causal modeling strategies.

Data-driven evaluation of building materials using Ground Penetrating Radar

Ground Penetrating Radar is a widely used technology in the nondestructive evaluation and monitoring of structures during repair and maintenance phases. Within this domain, a critical focus lies in assessing layered structures such as building envelopes. Evaluating attributes like thickness and material type in layered structures traditionally demands labor-intensive manual efforts involving calculations of dielectric properties and extensive signal processing. This study addresses these challenges by leveraging machine learning data-driven models that harness the entire recorded waveform, presenting a potential breakthrough for expediting diagnostic monitoring within the construction industry. We evaluate both supervised and unsupervised machine learning models for classification and regression tasks predicting the properties of common building materials. Experimental GPR A-scans serve as input for these models, with a thorough evaluation of both instance-based and parametric modeling approaches. A detection accuracy of 100% is achieved for identifying outlier scans, while material classification accuracy is 85%. Layer thickness predictions also had a high accuracy, with a typical error of 5%. Additionally, we explore the impact of feature learning and other preprocessing strategies on model performance. Our findings demonstrate the suitability of standard data-driven models for a spectrum of supervised learning tasks for layered structure diagnostics. However, we emphasize the importance of careful attention to the distribution of training data and its relevance to intended use in the field.

AI-driven generalized polynomial transformation models for unsupervised fundus image registration.

We introduce a novel AI-driven approach to unsupervised fundus image registration utilizing our Generalized Polynomial Transformation (GPT) model. Through the GPT, we establish a foundational model capable of simulating diverse polynomial transformations, trained on a large synthetic dataset to encompass a broad range of transformation scenarios. Additionally, our hybrid pre-processing strategy aims to streamline the learning process by offering model-focused input. We evaluated our model's effectiveness on the publicly available AREDS dataset by using standard metrics such as image-level and parameter-level analyzes. Linear regression analysis reveals an average Pearson correlation coefficient (R) of 0.9876 across all quadratic transformation parameters. Image-level evaluation, comprising qualitative and quantitative analyzes, showcases significant improvements in Structural Similarity Index (SSIM) and Normalized Cross Correlation (NCC) scores, indicating its robust performance. Notably, precise matching of the optic disc and vessel locations with minimal global distortion are observed. These findings underscore the potential of GPT-based approaches in image registration methodologies, promising advancements in diagnosis, treatment planning, and disease monitoring in ophthalmology and beyond.

Classification of osteoarthritic and healthy cartilage using deep learning with Raman spectra

Raman spectroscopy is a rapid method for analysing the molecular composition of biological material. However, noise contamination in the spectral data necessitates careful pre-processing prior to analysis. Here we propose an end-to-end Convolutional Neural Network to automatically learn an optimal combination of pre-processing strategies, for the classification of Raman spectra of superficial and deep layers of cartilage harvested from 45 Osteoarthritis and 19 Osteoporosis (Healthy controls) patients. Using 6-fold cross-validation, the Multi-Convolutional Neural Network achieves comparable or improved classification accuracy against the best-performing Convolutional Neural Network applied to either the raw or pre-processed spectra. We utilised Integrated Gradients to identify the contributing features (Raman signatures) in the network decision process, showing they are biologically relevant. Using these features, we compared Artificial Neural Networks, Decision Trees and Support Vector Machines for the feature selection task. Results show that training on fewer than 3 and 300 features, respectively, for the disease classification and layer assignment task provide performance comparable to the best-performing CNN-based network applied to the full dataset. Our approach, incorporating multi-channel input and Integrated Gradients, can potentially facilitate the clinical translation of Raman spectroscopy-based diagnosis without the need for laborious manual pre-processing and feature selection.

Light Reflections Detection And Correction For Robotic Volumetric PIV

Laser light reflection mitigation in Particle Image Velocimetry (PIV) is crucial for accurate flow field measurements. While numerous methods exist for planar PIV, fewer have been developed for volumetric PIV systems, and in particular for coaxial setups like robotic volumetric PIV. Light reflections in volumetric PIV experiments result in high-intensity regions that corrupt particle detection and analysis. This study presents a novel approach for treating light reflections in robotic volumetric PIV experiments. The proposed method uses image filtering and masking techniques in the wavenumber space to separate particle images from reflection regions. The process involves decomposing the image signal into low- and high-wavenumber components using the 2D discrete Fourier transform (DFT) to then use a high-pass filter to attenuate the intensity of the reflection regions. Finally, a step of automated adaptive masking is applied to remove residual reflection areas that the filtering is not able to eliminate. The proposed approach is tested on experimental data obtained from experiments performed using robotic volumetric PIV on two different geometries: a side-view mirror and a rotating two-blade propeller. Comparison between raw and pre-processed images, as well as particle tracking results, is presented. The results from this data comparison show successful removal of reflection-induced artifacts in instantaneous images by using the spatial Fourier filter automated masking approach. The developed image pre-processing strategy effectively removes unsteady light reflection regions in robotic volumetric PIV images, preventing the appearance of spurious particle tracks and improving the accuracy of flow field measurements. The spatial gaps introduced by the masking procedure can be easily filled in via measurements from multiple directions, which are promptly achieved via the robotic volumetric PIV approach.

MRI-Based Brain Tumor Classification Using a Dilated Parallel Deep Convolutional Neural Network

Brain tumors are frequently classified with high accuracy using convolutional neural networks (CNNs) to better comprehend the spatial connections among pixels in complex pictures. Due to their tiny receptive fields, the majority of deep convolutional neural network (DCNN)-based techniques overfit and are unable to extract global context information from more significant regions. While dilated convolution retains data resolution at the output layer and increases the receptive field without adding computation, stacking several dilated convolutions has the drawback of producing a grid effect. This research suggests a dilated parallel deep convolutional neural network (PDCNN) architecture that preserves a wide receptive field in order to handle gridding artifacts and extract both coarse and fine features from the images. This article applies multiple preprocessing strategies to the input MRI images used to train the model. By contrasting various dilation rates, the global path uses a low dilation rate (2,1,1), while the local path uses a high dilation rate (4,2,1) for decremental even numbers to tackle gridding artifacts and to extract both coarse and fine features from the two parallel paths. Using three different types of MRI datasets, the suggested dilated PDCNN with the average ensemble method performs best. The accuracy achieved for the multiclass Kaggle dataset-III, Figshare dataset-II, and binary tumor identification dataset-I is 98.35%, 98.13%, and 98.67%, respectively. In comparison to state-of-the-art techniques, the suggested structure improves results by extracting both fine and coarse features, making it efficient.

Comparing preprocessing strategies for 3D-Gene microarray data of extracellular vesicle-derived miRNAs

BackgroundExtracellular vesicle-derived (EV)-miRNAs have potential to serve as biomarkers for the diagnosis of various diseases. miRNA microarrays are widely used to quantify circulating EV-miRNA levels, and the preprocessing of miRNA microarray data is critical for analytical accuracy and reliability. Thus, although microarray data have been used in various studies, the effects of preprocessing have not been studied for Toray’s 3D-Gene chip, a widely used measurement method. We aimed to evaluate batch effect, missing value imputation accuracy, and the influence of preprocessing on measured values in 18 different preprocessing pipelines for EV-miRNA microarray data from two cohorts with amyotrophic lateral sclerosis using 3D-Gene technology.ResultsEighteen different pipelines with different types and orders of missing value completion and normalization were used to preprocess the 3D-Gene microarray EV-miRNA data. Notable results were suppressed in the batch effects in all pipelines using the batch effect correction method ComBat. Furthermore, pipelines utilizing missForest for missing value imputation showed high agreement with measured values. In contrast, imputation using constant values for missing data exhibited low agreement.ConclusionsThis study highlights the importance of selecting the appropriate preprocessing strategy for EV-miRNA microarray data when using 3D-Gene technology. These findings emphasize the importance of validating preprocessing approaches, particularly in the context of batch effect correction and missing value imputation, for reliably analyzing data in biomarker discovery and disease research.

Exploring the potential of representation and transfer learning for anatomical neuroimaging: Application to psychiatry

The perspective of personalized medicine for brain disorders requires efficient learning models for anatomical neuroimaging-based prediction of clinical conditions. There is now a consensus on the benefit of deep learning (DL) in addressing many medical imaging tasks, such as image segmentation. However, for single-subject prediction problems, recent studies yielded contradictory results when comparing DL with Standard Machine Learning (SML) on top of classical feature extraction. Most existing comparative studies were limited in predicting phenotypes of little clinical interest, such as sex and age, and using a single dataset. Moreover, they conducted a limited analysis of the employed image pre-processing and feature selection strategies. This paper extensively compares DL and SML prediction capacity on five multi-site problems, including three increasingly complex clinical applications in psychiatry namely schizophrenia, bipolar disorder, and Autism Spectrum Disorder (ASD) diagnosis. To compensate for the relative scarcity of neuroimaging data on these clinical datasets, we also evaluate three pre-training strategies for transfer learning from brain imaging of the general healthy population: self-supervised learning, generative modeling and supervised learning with age. Overall, we find similar performance between randomly initialized DL and SML for the three clinical tasks and a similar scaling trend for sex prediction. This was replicated on an external dataset. We also show highly correlated discriminative brain regions between DL and linear ML models in all problems. Nonetheless, we demonstrate that self-supervised pre-training on large-scale healthy population imaging datasets (N≈10k), along with Deep Ensemble, allows DL to learn robust and transferable representations to smaller-scale clinical datasets (N≤1k). It largely outperforms SML on 2 out of 3 clinical tasks both in internal and external test sets. These findings suggest that the improvement of DL over SML in anatomical neuroimaging mainly comes from its capacity to learn meaningful and useful abstract representations of the brain anatomy, and it sheds light on the potential of transfer learning for personalized medicine in psychiatry

An effective multi-level memetic search with neighborhood reduction for the clustered team orienteering problem

The Clustered Team Orienteering Problem (CluTOP) extends the classic Clustered Orienteering Problem by considering the use of multiple vehicles. The problem is known to be NP-hard and can be used to formulate many real-life applications. This work presents a highly effective multi-level memetic search for CluTOP that combines a backbone-based edge assembly crossover to generate promising offspring solutions with an effective bilevel synergistic local search procedure at both cluster and customer levels to improve offspring solutions. Other novel features of the proposed approach include a joint use of three specific hash functions to identify the tabu status of candidate solutions at the cluster level, a multi-neighborhood search with inter-route and intra-route optimization at the customer level, a pre-processing neighborhood reduction strategy to avoid examining non-promising candidate solutions, and a strategy for controlled exploration of infeasible solutions. Extensive experimental results on 1848 benchmark instances convincingly demonstrate high competitiveness of the approach in terms of both solution quality and computational time, compared to the state-of-the-art heuristics from the literature. In particular, the proposed algorithm improves upon the existing best-known solutions for 294 instances, while matching the previous best-known results for all but 3 of the remaining instances. To gain further insights into the algorithm’s performance, additional experiments are conducted to analyze its main components.

A multidimensional analysis of self-esteem and individualism: A deep learning-based model for predicting elementary school students' academic performance

The current research descends into the discrete connection between self-esteem, individuality and academic achievement among primary school learners. It addresses the essential need for predictive algorithms to analyze and forecast academic results. Self-esteem and identity, as diverse psychological variables, have been highlighted as crucial determinants in creating academic success. The fundamental goal is to develop an effective prediction model for elementary school children's academic performance by including self-esteem and individuality in the analytical framework. We suggested an Owl Search Optimized Dynamic Deep Neural Network (OSO-DDNN) based model for making more precise predictions. The sample size for the research was 147,210 elementary school students who took the LegiLexi exam between 2016 and 2021. To standardize the data and improve model performance, the suggested methodology uses a pre-processing strategy based on the Min-Max scalar approach. Furthermore, Independent Component Analysis is used to detect and extract essential factors related to self-esteem, individuality and academic achievement. The architecture and parameters of the suggested approaches are optimized using the OSO algorithm to provide a prediction model that remains accurate and efficient. We quantify the effectiveness of the assessments using the following metrics: precision, recall, accuracy and F1-score to develop a design using conventional methods. The findings demonstrate that the model's resilience and promise for anticipating primary school pupils' academic success are based on psychological aspects. The OSO-DDNN is an essential instrument for educators and policymakers in identifying at-risk pupils and implementing targeted interventions, supporting a comprehensive approach to education that combines psychological well-being with academic performance.

An Optimized Semantic Segmentation Framework for Human Skin Detection

The study incorporating optimization strategy in semantic segmentation is underexplored in dermatology. Existing approaches used complex and various heuristic designs of image processing algorithms and deep models customized for skin detection problems. This paper demonstrates Particle Swarm Optimization (PSO)-incorporated AlexNet framework for the skin segmentation task. The results from testing the trained model are promising. The model produced satisfactory performances even with a strict split of 50 %, confirming the high efficiency of the proposed framework. The mean Jaccard index and Dice similarity measures evaluated between the annotated and predicted mask ranged from 0.80 to 0.93 in the binary classification of pixels as “skin” versus “background”. This work identified that the location and color variability of skin pixels in the training data are crucial to obtaining a good skin segmentation performance. Further works that can be explored in this area include adopting a robust preprocessing strategy to increase data variability and improve model generalization or implementing an optimization-enhanced strategy on the existing segmentation models for comparison.

Chasing Fairness in Graphs: A GNN Architecture Perspective

There has been significant progress in improving the performance of graph neural networks (GNNs) through enhancements in graph data, model architecture design, and training strategies. For fairness in graphs, recent studies achieve fair representations and predictions through either graph data pre-processing (e.g., node feature masking, and topology rewiring) or fair training strategies (e.g., regularization, adversarial debiasing, and fair contrastive learning). How to achieve fairness in graphs from the model architecture perspective is less explored. More importantly, GNNs exhibit worse fairness performance compared to multilayer perception since their model architecture (i.e., neighbor aggregation) amplifies biases. To this end, we aim to achieve fairness via a new GNN architecture. We propose Fair Message Passing (FMP) designed within a unified optimization framework for GNNs. Notably, FMP explicitly renders sensitive attribute usage in forward propagation for node classification task using cross-entropy loss without data pre-processing. In FMP, the aggregation is first adopted to utilize neighbors' information and then the bias mitigation step explicitly pushes demographic group node presentation centers together. In this way, FMP scheme can aggregate useful information from neighbors and mitigate bias to achieve better fairness and prediction tradeoff performance. Experiments on node classification tasks demonstrate that the proposed FMP outperforms several baselines in terms of fairness and accuracy on three real-world datasets. The code is available at https://github.com/zhimengj0326/FMP.

Dual-View Whitening on Pre-trained Text Embeddings for Sequential Recommendation

Recent advances in sequential recommendation models have demonstrated the efficacy of integrating pre-trained text embeddings with item ID embeddings to achieve superior performance. However, our study takes a unique perspective by exclusively focusing on the untapped potential of text embeddings, obviating the need for ID embeddings. We begin by implementing a pre-processing strategy known as whitening, which effectively transforms the anisotropic semantic space of pre-trained text embeddings into an isotropic Gaussian distribution. Comprehensive experiments reveal that applying whitening to pre-trained text embeddings in sequential recommendation models significantly enhances performance. Yet, a full whitening operation might break the potential manifold of items with similar text semantics. To retain the original semantics while benefiting from the isotropy of the whitened text features, we propose a Dual-view Whitening method for Sequential Recommendation (DWSRec), which leverages both fully whitened and relaxed whitened item representations as dual views for effective recommendations. We further examine the advantages of our approach through both empirical and theoretical analyses. Experiments on three public benchmark datasets show that DWSRec outperforms state-of-the-art methods for sequential recommendation.

Guiding a Harsh-Environments Robust Detector via RAW Data Characteristic Mining

Consumer-grade cameras capture the RAW physical description of a scene and then process the image signals to obtain high-quality RGB images that are faithful to human visual perception. Conventionally, dense prediction scenes require high-precision recognition of objects in RGB images. However, predicting RGB data to exhibit the expected adaptability and robustness in harsh environments can be challenging. By capitalizing on the broader color gamut and higher bit depth offered by RAW data, in this paper, we demonstrate that RAW data can significantly improve the accuracy and robustness of object detectors in harsh environments. Firstly, we propose a general Pipeline for RAW Detection (PRD), along with a preprocessing strategy tailored to RAW data. Secondly, we design the RAW Corruption Benchmark (RCB) to address the dearth of benchmarks that reflect realistic scenarios in harsh environments. Thirdly, we demonstrate the significant improvement of RAW images in object detection for low-light and corrupt scenes. Specifically, our experiments indicate that PRD (using FCOS) outperforms RGB detection by 13.9mAP on LOD-Snow without generating restored images. Finally, we introduce a new nonlinear method called Functional Regularization (FR), which can effectively mine the unique characteristics of RAW data. The code is available at https://github.com/DreamerCCC/RawMining.

A multi-fidelity deep operator network (DeepONet) for fusing simulation and monitoring data: Application to real-time settlement prediction during tunnel construction

Ground settlement prediction during mechanized tunneling is of paramount importance and remains a challenging research topic. Typically, two paradigms are existing: a physics-driven approach utilizing numerical simulation models for prediction, and a data-driven approach employing machine learning techniques to learn mappings between influencing factors and the settlement. To integrate the advantages of both approaches and to assimilate the data from different sources, we propose a multi-fidelity deep operator network (DeepONet) framework, leveraging the recently developed operator learning methods. The presented framework comprises of two components: a low-fidelity subnet that captures the fundamental ground settlement patterns obtained from finite element simulations, and a high-fidelity subnet that learns the nonlinear correlation between numerical models and real engineering monitoring data. A pre-processing strategy for causality is adopted to consider the spatio-temporal characteristics of the settlement during tunnel excavation. The results show that the proposed method can effectively capture the physical information provided by the numerical simulations and accurately fit measured data (R2 around 0.9) as well. Notably, even when dealing with very limited noisy monitoring data (with a 50% error), the proposed model remains robust, achieving satisfactory results with R2>0.8. In comparison, the R2 score obtained by pure simulation-based prediction is only 0.2. The utilization of transfer learning significantly reduces the training time from 20 min to within 30 s, showcasing the potential of our method for real-time settlement prediction during tunnel construction.