Heterogeneous Datasets Research Articles

Asterics: a simple tool for the ExploRation and Integration of omiCS data

BackgroundThe rapid development of omics acquisition techniques has induced the production of a large volume of heterogeneous and multi-level omics datasets, which require specific and sometimes complex analyses to obtain relevant biological information. Here, we present ASTERICS (version 2.5), a publicly available web interface for the analyses of omics datasets.ResultsASTERICS is designed to make both standard and complex exploratory and integration analysis workflows easily available to biologists and to provide high quality interactive plots. Special care has been taken to provide a comprehensive documentation of the implemented analyses and to guide users toward sound analysis choices regarding some specific omics data. Data and analyses are organized in a comprehensive graphical workflow within ASTERICS workspace to facilitate the understanding of successive data editions and analyses leading to a given result.ConclusionASTERICS provides an easy to use platform for omics data exploration and integration. The modular organization of its open source code makes it easy to incorporate new workflows and analyses by external contributors. ASTERICS is available at https://asterics.miat.inrae.fr and can also be deployed using provided docker images.

BI-FedGNN: Federated graph neural networks framework based on Bayesian inference

The development of the Industrial Internet of Things (IIoT) in recent years has resulted in an increase in the amount of data generated by connected devices, creating new opportunities to enhance the quality of service for machine learning in the IIoT through data sharing. Graph neural networks (GNNs) are the most popular technique in machine learning at the moment because they can learn extremely precise node representations from graph-structured data. Due to privacy issues and legal restrictions of clients in industrial IoT, it is not permissible to directly concentrate vast real-world graph-structured datasets for training on GNNs. To resolve the aforementioned difficulties, this paper proposes a federal graph learning framework based on Bayesian inference (BI-FedGNN) that performs effectively in the presence of noisy graph structure information or missing strong relational edges. BI-FedGNN extends Bayesian Inference (BI) to the process of Federal Graph Learning (FGL), adding random samples with weights and biases to the client-side local model training process, improving the accuracy and generalization ability of FGL in the training process by rendering the graph structure data involved in GNNs training more similar to the graph structure data existing in the real world. Through extensive experimental tests, the results show that BI-FedGNN has about 0.5%–5.0% accuracy improvement over other baselines of federal graph learning. In order to expand the applicability of BI-FedGNN, experiments are carried out on heterogeneous graph datasets, and the results indicate that BI-FedGNN can also have at least 1.4% improvement in classification accuracy.

Neuroimaging-based classification of PTSD using data-driven computational approaches: A multisite big data study from the ENIGMA-PGC PTSD consortium

BackgroundRecent advances in data-driven computational approaches have been helpful in devising tools to objectively diagnose psychiatric disorders. However, current machine learning studies limited to small homogeneous samples, different methodologies, and different imaging collection protocols, limit the ability to directly compare and generalize their results. Here we aimed to classify individuals with PTSD versus controls and assess the generalizability using a large heterogeneous brain datasets from the ENIGMA-PGC PTSD Working group. MethodsWe analyzed brain MRI data from 3,477 structural-MRI; 2,495 resting state-fMRI; and 1,952 diffusion-MRI. First, we identified the brain features that best distinguish individuals with PTSD from controls using traditional machine learning methods. Second, we assessed the utility of the denoising variational autoencoder (DVAE) and evaluated its classification performance. Third, we assessed the generalizability and reproducibility of both models using leave-one-site-out cross-validation procedure for each modality. ResultsWe found lower performance in classifying PTSD vs. controls with data from over 20 sites (60 % test AUC for s-MRI, 59 % for rs-fMRI and 56 % for d-MRI), as compared to other studies run on single-site data. The performance increased when classifying PTSD from HC without trauma history in each modality (75 % AUC). The classification performance remained intact when applying the DVAE framework, which reduced the number of features. Finally, we found that the DVAE framework achieved better generalization to unseen datasets compared with the traditional machine learning frameworks, albeit performance was slightly above chance. ConclusionThese results have the potential to provide a baseline classification performance for PTSD when using large scale neuroimaging datasets. Our findings show that the control group used can heavily affect classification performance. The DVAE framework provided better generalizability for the multi-site data. This may be more significant in clinical practice since the neuroimaging-based diagnostic DVAE classification models are much less site-specific, rendering them more generalizable.

Deep learning on photoacoustic tomography to remove image distortion due to inaccurate measurement of the scanning radius.

Photoacoustic tomography (PAT) is a non-invasive, non-ionizing hybrid imaging modality that holds great potential for various biomedical applications and the incorporation with deep learning (DL) methods has experienced notable advancements in recent times. In a typical 2D PAT setup, a single-element ultrasound detector (USD) is used to collect the PA signals by making a 360° full scan of the imaging region. The traditional backprojection (BP) algorithm has been widely used to reconstruct the PAT images from the acquired signals. Accurate determination of the scanning radius (SR) is required for proper image reconstruction. Even a slight deviation from its nominal value can lead to image distortion compromising the quality of the reconstruction. To address this challenge, two approaches have been developed and examined herein. The first framework includes a modified version of dense U-Net (DUNet) architecture. The second procedure involves a DL-based convolutional neural network (CNN) for image classification followed by a DUNet. The first protocol was trained with heterogeneous simulated images generated from three different phantoms to learn the relationship between the reconstructed and the corresponding ground truth (GT) images. In the case of the second scheme, the first stage was trained with the same heterogeneous dataset to classify the image type and the second stage was trained individually with the appropriate images. The performance of these architectures has been tested on both simulated and experimental images. The first method can sustain SR deviation up to approximately 6% for simulated images and 5% for experimental images and can accurately reproduce the GTs. The proposed DL-approach extends the limits further (approximately 7% and 8% for simulated and experimental images, respectively). Our results suggest that classification-based DL method does not need a precise assessment of SR for accurate PAT image formation.

Adaptive Neighbor Graph Aggregated Graph Attention Network for Heterogeneous Graph Embedding

Graph attention network can generate effective feature embedding by specifying different weights to different nodes. The key of the research on heterogeneous graph embedding is the way to combine its rich structural information with semantic relations to aggregate the neighborhood information. Most of the existing heterogeneous graph representation learning methods guide the selection of neighbors by defining various meta-paths on heterogeneous graphs. However, these models only consider the information contained in the nodes under different paths and ignore the potential semantic relationships of nodes in different neighbor graph structures, which leads to the underutilization of graph structure information. In this article, we propose a novel adaptive framework named Neighbor Graph Aggregated Graph Attention Network (NGGAN) to fully exploit graph topological details in heterogeneous graph, and aggregates their information to obtain an effective embedding. The key idea is to use different levels of sampling methods to define neighborhood, and use neighbor graphs to represent the complex structural interaction between nodes. In this way, the high-order relationship between nodes and the latent semantics of neighbor graphs can be fully explored. Afterward, a hierarchical attention mechanism is applied to adaptively learn the importance of different objects, including node information, path information, and neighbor graph information. Multiple downstream tasks are performed on four real-world heterogeneous graph datasets, and the experimental results demonstrate the effectiveness of NGGAN.

Universal adversarial perturbations for multiple classification tasks with quantum classifiers

Abstract Quantum adversarial machine learning is an emerging field that studies the vulnerability of quantum learning systems against adversarial perturbations and develops possible defense strategies. Quantum universal adversarial perturbations are small perturbations, which can make different input samples into adversarial examples that may deceive a given quantum classifier. This is a field that was rarely looked into but worthwhile investigating because universal perturbations might simplify malicious attacks to a large extent, causing unexpected devastation to quantum machine learning models. In this paper, we take a step forward and explore the quantum universal perturbations in the context of heterogeneous classification tasks. In particular, we find that quantum classifiers that achieve almost state-of-the-art accuracy on two different classification tasks can be both conclusively deceived by one carefully-crafted universal perturbation. This result is explicitly demonstrated with well-designed quantum continual learning models with elastic weight consolidation method to avoid catastrophic forgetting, as well as real-life heterogeneous datasets from hand-written digits and medical MRI images. Our results provide a simple and efficient way to generate universal perturbations on heterogeneous classification tasks and thus would provide valuable guidance for future quantum learning technologies.

Deep Learning-Based Semantic Segmentation of Urban Areas Using Heterogeneous Unmanned Aerial Vehicle Datasets

Deep learning techniques have recently shown remarkable efficacy in the semantic segmentation of natural and remote sensing (RS) images. However, these techniques heavily rely on the size of the training data, and obtaining large RS imagery datasets is difficult (compared to RGB images), primarily due to environmental factors such as atmospheric conditions and relief displacement. Unmanned aerial vehicle (UAV) imagery presents unique challenges, such as variations in object appearance due to UAV flight altitude and shadows in urban areas. This study analyzed the combined segmentation network (CSN) designed to train heterogeneous UAV datasets effectively for their segmentation performance across different data types. Results confirmed that CSN yielded high segmentation accuracy on specific classes and can be used on diverse data sources for UAV image segmentation. The main contributions of this study include analyzing the impact of CSN on segmentation accuracy, experimenting with structures with shared encoding layers to enhance segmentation accuracy, and investigating the influence of data types on segmentation accuracy.

Novel deep learning models with novel integrated activation functions for autism screening: AutiNet and MinAutiNet

Deep Learning (DL) particularly CNN has the potential to outplay conventional Machine Learning (ML) procedures in the realm of computer vision. With the advent of DL, profound research on autism utilizing fMRI data has declared advancements in autism screening. Yet, DL models are inadequate for heterogeneous datasets encompassing complex fMRI data. To resolve these problems, two cutting-edge AutiNet and MinAutiNet models are proposed with a minimum number of layers and neurons in classifying the autistic subjects. Both AutiNet and MinAutiNet models are the two refined forms of the traditional LeNet-5 model. AutiNet model includes two convolutional layers, two average pooling layers, a flatten layer, and one fully connected layer. One convolutional layer, one average pooling layer, one flatten layer, and one fully connected layer make up the MinAutiNet model. Pre-processing is a pre-requisite phase in the AutiNet and MinAutiNet models where denoising of raw fMRI data is done with slice time correction, realignment, min–max normalization and one hot encoding to obtain numerical values. As inspired by the promising results of non-handcrafted feature extraction, the proposed DL models have been used for feature extraction. Two novel integrated activation functions namely Li-ReLU and S-RReLU are developed and incorporated into the proposed DL models towards improving the performance of autism screening. Both AutiNet and MinAutiNet models surpass the standard accuracy in classifying autistic subjects with minimum number of layers and neurons. Experimental results reveal that the proposed AutiNet attains the maximum accuracy of 77.78% with Caltech and KKI datasets and the proposed MinAutiNet achieves the maximum accuracy of 88.89% with SBL dataset. The proposed models are compared with other state-of-the-art works to ensure its superiority.

Data integration: “Seamless data harmony: The art and science of effective data integration”

The idea of data integration has evolved as a key strategy in today's data-driven environment, as data is supplied from various and heterogeneous sources. This article explores the relevance, methodology, difficulties, and transformative possibilities of data integration, delving into its multidimensional world. Data integration serves as the cornerstone for well-informed decision-making by connecting heterogeneous datasets and fostering unified insights. This article gives readers a sneak preview of the in-depth investigation into data integration, illuminating its technical complexities and strategic ramifications for companies and organizations looking to maximize the value of their data as-sets.

VNAGT: Variational Non-Autoregressive Graph Transformer Network for Multi-Agent Trajectory Prediction

Accurately predicting the trajectory of road agents in complex traffic scenarios is challenging because the movement patterns of agents are complex and stochastic, not only depending on their own past trajectories but also being closely related to the social interaction with other surrounding agents. Besides accuracy, efficient prediction with low inference latency is also a highly desirable feature for the practical application of trajectory prediction. To address these issues, we propose the VNAGT model, a variational non-autoregressive graph transformer that adopts the framework of conditional variational autoencoder and incorporates the non-autoregressive approach such that diverse trajectories and low prediction latency can be achieved simultaneously. In order to capture the social and temporal interaction, we put forward a unified graph attention-based module that is applicable for homogeneous and heterogeneous multi-agents such that the class information can be seamlessly integrated when it is available. Non-autoregressive decoding is combined with variational learning to produce multiple plausible predictions with low latency. We train and validate the model on two real-world homogeneous and heterogeneous trajectory datasets. The experimental results demonstrate the superior performance of our model in comparison with the state-of-the-art methods.

A Novel Domain Knowledge-Informed Machine Learning Approach for Modeling Solid Waste Management Systems.

Sustainability challenges, such as solid waste management, are usually scientifically complex and data scarce, which makes them not amenable to science-based analytical forms or data-intensive learning paradigms. Deep integration between data science and sustainability science in highly complementary manners offers new opportunities for tackling these conundrums. This study develops a novel hybrid neural network (HNN) model that imposes the holistic decision-making context of solid waste management systems (SWMS) on a traditional neural network (NN) architecture. Equipped with adaptable hybridization designs of hand-crafted model structure, constrained or predetermined parameters, and a customized loss function, the HNN model is capable of learning various technical, economic, and social aspects of SWMS from a small and heterogeneous data set. In comparison, the versatile HNN model not only outperforms traditional NN models in convergence rates, which leads to a 22% lower mean testing error of 0.20, but also offers superior interpretability. The HNN model is capable of generating insights into the enabling factors, policy interventions, and driving forces of SWMS, laying a solid foundation for data-driven decision making.

Reliable and accurate prediction of basic pK_a values in nitrogen compounds: the pK_a shift in supramolecular systems as a case study

This article presents a quantitative structure–activity relationship (QSAR) approach for predicting the acid dissociation constant (pK_a) of nitrogenous compounds, including those within supramolecular complexes based on cucurbiturils. The model combines low-cost quantum mechanical calculations with QSAR methodology and linear regressions to achieve accurate predictions for a broad range of nitrogen-containing compounds. The model was developed using a diverse dataset of 130 nitrogenous compounds and exhibits excellent predictive performance, with a high coefficient of determination (R^2) of 0.9905, low standard error (s) of 0.3066, and high Fisher statistic (F) of 2142. The model outperforms existing methods, such as Chemaxon software and previous studies, in terms of accuracy and its ability to handle heterogeneous datasets. External validation on pharmaceutical ingredients, dyes, and supramolecular complexes based on cucurbiturils confirms the reliability of the model. To enhance usability, a script-like tool has been developed, providing a streamlined process for users to access the model. This study represents a significant advancement in pK_a prediction, offering valuable insights for drug design and supramolecular system optimization.Graphical

Electron microscopy mapping of the DNA-binding sites of monomeric, dimeric, and multimeric KSHV RTA protein.

Kaposi's sarcoma-associated herpesvirus (KSHV) is a human herpesvirus associated with several human cancers, typically in patients with compromised immune systems. Herpesviruses establish lifelong infections in hosts in part due to the two phases of infection: the dormant and active phases. Effective antiviral treatments to prevent the production of new viruses are needed to treat KSHV. A detailed microscopy-based investigation of the molecular interactions between viral protein and viral DNA revealed how protein-protein interactions play a role in DNA-binding specificity. This analysis will lead to a more in-depth understanding of KSHV DNA replication and serve as the basis for anti-viral therapies that disrupt and prevent the protein-DNA interactions, thereby decreasing spread to new hosts.

COMO: a pipeline for multi-omics data integration in metabolic modeling and drug discovery.

Identifying potential drug targets using metabolic modeling requires integrating multiple modeling methods and heterogeneous biological datasets, which can be challenging without efficient tools. We developed Constraint-based Optimization of Metabolic Objectives (COMO), a user-friendly pipeline that integrates multi-omics data processing, context-specific metabolic model development, simulations, drug databases and disease data to aid drug discovery. COMO can be installed as a Docker Image or with Conda and includes intuitive instructions within a Jupyter Lab environment. It provides a comprehensive solution for the integration of bulk and single-cell RNA-seq, microarrays and proteomics outputs to develop context-specific metabolic models. Using public databases, open-source solutions for model construction and a streamlined approach for predicting repurposable drugs, COMO enables researchers to investigate low-cost alternatives and novel disease treatments. As a case study, we used the pipeline to construct metabolic models of B cells, which simulate and analyze them to predict metabolic drug targets for rheumatoid arthritis and systemic lupus erythematosus, respectively. COMO can be used to construct models for any cell or tissue type and identify drugs for any human disease where metabolic inhibition is relevant. The pipeline has the potential to improve the health of the global community cost-effectively by providing high-confidence targets to pursue in preclinical and clinical studies. The source code of the COMO pipeline is available at https://github.com/HelikarLab/COMO. The Docker image can be pulled at https://github.com/HelikarLab/COMO/pkgs/container/como.

Developing Clinical Phenotype Data Collection Standards for Research in Africa.

Modern biomedical research is characterised by its high-throughput and interdisciplinary nature. Multiproject and consortium-based collaborations requiring meaningful analysis of multiple heterogeneous phenotypic datasets have become the norm; however, such analysis remains a challenge in many regions across the world. An increasing number of data harmonisation efforts are being undertaken by multistudy collaborations through either prospective standardised phenotype data collection or retrospective phenotype harmonisation. In this regard, the Phenotype Harmonisation Working Group (PHWG) of the Human Heredity and Health in Africa (H3Africa) consortium aimed to facilitate phenotype standardisation by both promoting the use of existing data collection standards (hosted by PhenX), adapting existing data collection standards for appropriate use in low- and middle-income regions such as Africa, and developing novel data collection standards where relevant gaps were identified. Ultimately, the PHWG produced 11 data collection kits, consisting of 82 protocols, 38 of which were existing protocols, 17 were adapted, and 27 were novel protocols. The data collection kits will facilitate phenotype standardisation and harmonisation not only in Africa but also across the larger research community. In addition, the PHWG aims to feed back adapted and novel protocols to existing reference platforms such as PhenX.

Machine and Deep Learning Algorithms for COVID-19 Mortality Prediction Using Clinical and Radiomic Features

Aim: Machine learning (ML) and deep learning (DL) predictive models have been employed widely in clinical settings. Their potential support and aid to the clinician of providing an objective measure that can be shared among different centers enables the possibility of building more robust multicentric studies. This study aimed to propose a user-friendly and low-cost tool for COVID-19 mortality prediction using both an ML and a DL approach. Method: We enrolled 2348 patients from several hospitals in the Province of Reggio Emilia. Overall, 19 clinical features were provided by the Radiology Units of Azienda USL-IRCCS of Reggio Emilia, and 5892 radiomic features were extracted from each COVID-19 patient’s high-resolution computed tomography. We built and trained two classifiers to predict COVID-19 mortality: a machine learning algorithm, or support vector machine (SVM), and a deep learning model, or feedforward neural network (FNN). In order to evaluate the impact of the different feature sets on the final performance of the classifiers, we repeated the training session three times, first using only clinical features, then employing only radiomic features, and finally combining both information. Results: We obtained similar performances for both the machine learning and deep learning algorithms, with the best area under the receiver operating characteristic (ROC) curve, or AUC, obtained exploiting both clinical and radiomic information: 0.803 for the machine learning model and 0.864 for the deep learning model. Conclusions: Our work, performed on large and heterogeneous datasets (i.e., data from different CT scanners), confirms the results obtained in the recent literature. Such algorithms have the potential to be included in a clinical practice framework since they can not only be applied to COVID-19 mortality prediction but also to other classification problems such as diabetic prediction, asthma prediction, and cancer metastases prediction. Our study proves that the lesion’s inhomogeneity depicted by radiomic features combined with clinical information is relevant for COVID-19 mortality prediction.

Forensic analysis of manipulation chains: A deep residual network for detecting JPEG-manipulation-JPEG

Due to the proliferation of image editing software, the problem of identifying doctored photos has risen to prominence in information forensics. Even though existing image processing history detection techniques could provide reasonably acceptable outcomes on publicly available databases, most of them generally fail to accomplish this when the doctored images are lossy compressed, which is commonly accomplished in social media. To tackle this issue, a dual-stream residual network (ReMReNet) using ResNet as a backbone is proposed for JPEG-resistant image operator chain detection. ReMReNet consists of noise residual extraction (NRE) and compression feature extraction (CFE) streams for better results against JPEG compression. The CFE stream is trained on discrete cosine transform (DCT) residual coefficients by producing clues left by JPEG-agent images with generic JPEG compression artefacts. In addition, the NRE stream is utilized to extract noise residuals to suppress the image content and enhance the manipulation clues. Hence, the domain transformation between uncompressed and post-JPEG compressed images is reduced by training the backbone network with a domain adaption approach, which helps forensic investigators pinpoint the exact location of the tampering. The proposed framework outperforms other state-of-the-art algorithms in image multiple manipulations and operator chain detection, notably for lossy JPEG compression, as shown by extensive experimental findings on numerous heterogeneous datasets.

A novel method for resampling detection in highly compressed JPEG images through BAR using a deep learning technique

As JPEG has become a prevalent compression standard, identifying manipulations in JPEG images has become increasingly necessary. However, many of the inspected images are highly compressed, making it challenging for forensics tools and methodologies to detect clues of resampling. To overcome this issue, a new method has been proposed that uses a deep convolutional neural network (CNN) to detect image resampling by removing blocking artifacts (BAR) from recompressed images without altering resampling clues. The proposed framework has three stages: noise residual extraction, feature extraction, and classification. Initially, the image content and BAR are suppressed, and noise residuals from recompressed images are extracted using ten convolutional layers without pooling layers. The feature extraction phase then learns resampling features from the noise residuals in an adaptive manner. Deep resampling characteristics are processed through the fully connected layer to identify whether the input image is double compressed and resized (DCR) or double compressed and unresized (DCUR). The proposed deep CNN has been tested on heterogeneous datasets such as RAISE and DRESDEN, identifying resampling with a mean accuracy of 91.85% and 90.21% in cross-dataset settings. In addition, detecting downscaling with more than an accuracy of 87.98% in highly recompressed images indicates that it outperforms state-of-the-art approaches. Furthermore, the proposed method yields notable results in estimating the resampling factor in recompressed images.

Differential privacy preserved federated transfer learning for multi-institutional 68Ga-PET image artefact detection and disentanglement

PurposeImage artefacts continue to pose challenges in clinical molecular imaging, resulting in misdiagnoses, additional radiation doses to patients and financial costs. Mismatch and halo artefacts occur frequently in gallium-68 (68Ga)-labelled compounds whole-body PET/CT imaging. Correcting for these artefacts is not straightforward and requires algorithmic developments, given that conventional techniques have failed to address them adequately. In the current study, we employed differential privacy-preserving federated transfer learning (FTL) to manage clinical data sharing and tackle privacy issues for building centre-specific models that detect and correct artefacts present in PET images.MethodsAltogether, 1413 patients with 68Ga prostate-specific membrane antigen (PSMA)/DOTA-TATE (TOC) PET/CT scans from 3 countries, including 8 different centres, were enrolled in this study. CT-based attenuation and scatter correction (CT-ASC) was used in all centres for quantitative PET reconstruction. Prior to model training, an experienced nuclear medicine physician reviewed all images to ensure the use of high-quality, artefact-free PET images (421 patients’ images). A deep neural network (modified U2Net) was trained on 80% of the artefact-free PET images to utilize centre-based (CeBa), centralized (CeZe) and the proposed differential privacy FTL frameworks. Quantitative analysis was performed in 20% of the clean data (with no artefacts) in each centre. A panel of two nuclear medicine physicians conducted qualitative assessment of image quality, diagnostic confidence and image artefacts in 128 patients with artefacts (256 images for CT-ASC and FTL-ASC).ResultsThe three approaches investigated in this study for 68Ga-PET imaging (CeBa, CeZe and FTL) resulted in a mean absolute error (MAE) of 0.42 ± 0.21 (CI 95%: 0.38 to 0.47), 0.32 ± 0.23 (CI 95%: 0.27 to 0.37) and 0.28 ± 0.15 (CI 95%: 0.25 to 0.31), respectively. Statistical analysis using the Wilcoxon test revealed significant differences between the three approaches, with FTL outperforming CeBa and CeZe (p-value < 0.05) in the clean test set. The qualitative assessment demonstrated that FTL-ASC significantly improved image quality and diagnostic confidence and decreased image artefacts, compared to CT-ASC in 68Ga-PET imaging. In addition, mismatch and halo artefacts were successfully detected and disentangled in the chest, abdomen and pelvic regions in 68Ga-PET imaging.ConclusionThe proposed approach benefits from using large datasets from multiple centres while preserving patient privacy. Qualitative assessment by nuclear medicine physicians showed that the proposed model correctly addressed two main challenging artefacts in 68Ga-PET imaging. This technique could be integrated in the clinic for 68Ga-PET imaging artefact detection and disentanglement using multicentric heterogeneous datasets.

Signature Verification System: Using Big Transfer (BiT-M-R50x1) for Accurate Authentication

In the realm of document security, signature verification stands as a vital pillar for establishing authenticity. This study delves into the utilization of the potent Big Transfer (BiT) BiT-M-R50x1 model for the intricate task of signature validation. This dataset encompasses 2149 signature images sourced from diverse individuals, exhibiting notable fluctuations in writing styles, pen pressures, and signature dimensions. By harnessing the prowess of the pre-trained BiT-M-R50x1 model, renowned for its domaingeneralization capability, we fine-tune it to excel in signature verification. The results of our approach unveil remarkable accomplishments on the dataset, yielding a validation accuracy of 98.60%. The meticulously calibrated BiT-M-R50x1 model adeptly distinguishes between authentic and counterfeit signatures, even when confronted with substantial variation. Through the mechanism of transfer learning, the model captures intrinsic attributes that extrapolate effectively to previously unseen signature specimens. Furthermore, we meticulously assess the model's performance concerning the dataset's distinctive signature idiosyncrasies, scrutinizing its adaptability to diverse styles and dimensions. This experiment underscores the potential of harnessing robust pre-trained models like BiT-M-R50x1 for signature verification undertakings, particularly when grappling with intricate and heterogeneous datasets.