Siamese Network Research Articles

Shape-Aware Adversarial Learning for Scribble-Supervised Medical Image Segmentation with a MaskMix Siamese Network: A Case Study of Cardiac MRI Segmentation

The transition in medical image segmentation from fine-grained to coarse-grained annotation methods, notably scribble annotation, offers a practical and efficient preparation for deep learning applications. However, these methods often compromise segmentation precision and result in irregular contours. This study targets the enhancement of scribble-supervised segmentation to match the accuracy of fine-grained annotation. Capitalizing on the consistency of target shapes across unpaired datasets, this study introduces a shape-aware scribble-supervised learning framework (MaskMixAdv) addressing two critical tasks: (1) Pseudo label generation, where a mixup-based masking strategy enables image-level and feature-level data augmentation to enrich coarse-grained scribbles annotations. A dual-branch siamese network is proposed to generate fine-grained pseudo labels. (2) Pseudo label optimization, where a CNN-based discriminator is proposed to refine pseudo label contours by distinguishing them from external unpaired masks during model fine-tuning. MaskMixAdv works under constrained annotation conditions as a label-efficient learning approach for medical image segmentation. A case study on public cardiac MRI datasets demonstrated that the proposed MaskMixAdv outperformed the state-of-the-art methods and narrowed the performance gap between scribble-supervised and mask-supervised segmentation. This innovation cuts annotation time by at least 95%, with only a minor impact on Dice performance, specifically a 2.6% reduction. The experimental outcomes indicate that employing efficient and cost-effective scribble annotation can achieve high segmentation accuracy, significantly reducing the typical requirement for fine-grained annotations.

Intrusion event recognition method for distributed fibre optic sensing systems based on Siamese networks

ABSTRACT Due to its high sensitivity and wide measurement range, the distributed optical fibre sensing system has been widely used in long-distance infrastructure monitoring, where it detects vibration signals caused by external events and provides effective early warnings. Given the complexity and diversity of external intrusion events, traditional closed-set classification methods cannot effectively exclude unknown signals. Open-set recognition methods, on the other hand, involve complex model designs, requiring boundary Algorithms and often the inclusion of information related to unknown classes. In this paper, we combine a Siamese network with a residual network to recognise events in the distributed optical fibre sensing system. Through experiments, by comparing the similarity with data in the database, we not only ensure very high accuracy in closed-set classification but also effectively exclude unknown class signals. Experimental results show that our method successfully rejects unknown class data with a 92% success rate, while maintaining closed-set classification accuracy at an average of 99%. Our approach demonstrates excellent performance.

WKNN-Based Wi-Fi Fingerprinting with Deep Distance Metric Learning via Siamese Triplet Network for Indoor Positioning

Weighted k-nearest neighbor (WKNN)-based Wi-Fi fingerprinting is popular in indoor location-based services due to its ease of implementation and low computational cost. KNN-based methods rely on distance metrics to select the nearest neighbors. However, traditional metrics often fail to capture the complexity of indoor environments and have limitations in identifying non-linear relationships. To address these issues, we propose a novel WKNN-based Wi-Fi fingerprinting method that incorporates distance metric learning. In the offline phase, our method utilizes a Siamese network with a triplet loss function to learn a meaningful distance metric from training fingerprints (FPs). This process employs a unique triplet mining strategy to handle the inherent noise in FPs. Subsequently, in the online phase, the learned metric is used to calculate the embedding distance, followed by a signal-space distance filtering step to optimally select neighbors and estimate the user’s location. The filtering step mitigates issues from an overfitted distance metric influenced by hard triplets, which could lead to incorrect neighbor selection. We evaluate the proposed method on three benchmark datasets, UJIIndoorLoc, Tampere, and UTSIndoorLoc, and compare it with four WKNN models. The results show a mean positioning error reduction of 3.55% on UJIIndoorLoc, 16.21% on Tampere, and 16.49% on UTSIndoorLoc, demonstrating enhanced positioning accuracy.

STGRL: SNN based two-stage geomagnetic road localization method

Abstract Geomagnetic navigation is a widely used positioning method capable of correcting the cumulative errors of odometers and inertial navigation systems, thereby ensuring long-distance positioning for vehicles in GPS-denied environments. However, common geomagnetic road navigation algorithms are susceptible to measurement noise, which hinder improvements in positioning efficiency and accuracy. To address this issue, this paper proposes a Siamese Neural Network (SNN) based two-stage geomagnetic road localization method. First, attitude angle information is combined with geomagnetic scalar and vector value to establish geomagnetic reference database to increase the feature dimensions of geomagnetic matching. Then, we use the Random Forest algorithm to perform a coarse matching of the data sequence to determine the current road, balancing the increased computational load resulting from the addition of feature dimensions. Finally, to further reduce the impact of random noise, this paper employs the SNN algorithm based on Transformer Encoder for fine matching of the data sequence. Experiments show that compared to existing methods, the average absolute positioning error of our algorithm has been reduced from 32.36 m to 4.07 m, and the increase in computational load is kept within an acceptable range.

Advancements in deep learning for visual object tracking

Abstract. Since contemporary information-retrieval systems rely heavily on the content of titles and abstracts to identify relevant articles in literature searches, great care should be taken in constructing both. This comprehensive review delves into the transformative impact of deep learning on the domain of visual object tracking. Since the inception of AlexNet in 2012, deep learning has revolutionized feature extraction, leading to significant advancements in tracking accuracy and robustness. The article explores the integration of deep learning with various tracking algorithms, including deep correlation filters, classification-based approaches, Siamese networks, gradient-based methods, and the innovative application of Transformer architectures. Moreover, the role of tracking datasets in fostering algorithm development and innovation is highlighted, with an emphasis on the expansion in scale, diversity, and annotation quality. Furthermore, the article also examines the multifaceted evaluation metrics for tracking algorithms, encompassing precision, robustness, efficiency, generalization, and real-time capabilities. Looking ahead, the review outlines future research directions, such as algorithm optimization for lightweight and accelerated performance, enhancing generalizability, leveraging multimodal data fusion, and refining Transformer models for improved temporal information processing. The challenges of long-term tracking and the growing importance of algorithm interpretability and transparency are also discussed. In summary, the article underscores the promising trajectory of deep learning in visual object tracking, with ongoing research poised to make tracking technologies smarter, more efficient, and robust, catering to a wide array of practical applications and environments.

Automated mapping of electronic data capture fields to SDTM.

The goal of this work is to reduce the amount of manual work required to go from data capture to regulatory submission. It will be shown that the use of Siamese networks will allow for the generation of embeddings that can be used by traditional machine learning classifiers to perform the classification at much higher levels of accuracy than standard approaches. Siamese networks are a method for training data embeddings such that data within the same class are closer with respect to a given distance metric than they are to data points in another class. Because they are designed to learn similarity within pairs of data points, they work well in situations where the number of classes is relatively large compared to the number of training samples. In this work, we will show that embeddings generated via a Siamese network from metadata associated with electronic data capture forms can be used to predict the associated SDTM field. With a relatively simple network coupled with a basic classification algorithm, the proposed method can achieve accuracies greater than 90%, which is significantly higher than what has been achieved with traditional methods, with many of the inaccurate mappings due to a lack of training data. In many cases, there is a 15% increase in accuracy vs. more traditional methods. Leveraging Siamese networks, it is possible to generate embeddings that efficiently represent data fields in a lower dimensional space. This allows the creation of a system that can automatically map between data schemas at high levels of accuracy. Such systems represent the first step in automating one of the many labor-intensive data management tasks associated with clinical trials.

Attack-Defending Contrastive Learning for Volumetric Medical Image Zero-Watermarking

Zero-watermarking is an emerging distortion-free copyright protection method for volumetric medical images. However, achieving both robustness against various malicious attacks and distinguishability between individual images remains challenging. In this paper, we propose a novel attack-defending contrastive learning zero-watermarking scheme (ADCL-ZW) to tackle the above challenge using deep learning-based representations. In our approach, we design an attack-defending data enrichment mechanism to enhance the watermarking robustness by generating a large number of image samples under various watermarking attacks. Subsequently, features for both watermarking distinguishability and robustness are enhanced through application of a contrastive loss. In particular, we implement a dual-stream Siamese network architecture to effectively handle both signal attacks and geometric attacks in order to enhance the wartermarking performance. Experimental results demonstrate that ADCL-ZW achieves stronger watermarking robustness and a better tradeoff between watermarking robustness and distinguishability compared with state-of-the art zero-watermarking methods. One of the highlighted metrics the false negative rate of ADCL-ZW achieves 0.01 when a fixed false positive rate is set to 1%, which is more than 13.3 times better than the benchmark methods.

OPTIMIZING AUTHENTICATION SECURITY IN INTELLIGENT SYSTEMS THROUGH VISUAL BIOMETRICS FOR ENHANCED EFFICIENCY

Context. The primary objective of this article is to explore aspects related to ensuring security and enhancing the efficiency of authentication processes in intelligent systems through the application of visual biometrics. The focus is on advancing and refining authentication systems by employing sophisticated biometric identification methods. Objective. A specialized intelligent system has been developed, utilizing a Siamese neural network to establish secure user authentication within the existing system. Beyond incorporating fundamental security measures such as hashing and secure storage of user credentials, the contemporary significance of implementing two-factor authentication is underscored. This approach significantly fortifies user data protection, thwarting most contemporary hacking methods and safeguarding against data breaches. The study acknowledges certain limitations in its approach, possibly affecting the generalizability of the findings. These limitations provide avenues for future research and exploration, contributing to the ongoing evolution of authentication methodologies in intelligent systems. Method. The two-factor authentication system integrates facial recognition technology, employing visual biometrics for heightened security compared to alternative two-factor authentication methods. Various implementations of the Siamese neural network, utilizing Contrastive loss function and Triplet loss function, were evaluated. Subsequently, a neural network employing the Triplet loss function was implemented and trained. Results. The article emphasizes the practical implications of the developed intelligent system, showcasing its effectiveness in minimizing the risk of unauthorized access to user accounts. The integration of contemporary authentication methodologies ensures a secure and robust user authentication process. Conclusions. The implementation of facial recognition technology in authentication processes has broader social implications. It contributes to a more secure digital environment by preventing unauthorized access, ultimately safeguarding user privacy and data. The study’s originality lies in its innovative approach to authentication, utilizing visual biometrics within a Siamese neural network framework. The developed intelligent system represents a valuable contribution to the field, offering an effective and contemporary solution to user authentication challenges.

Two-level integrated verification evaluation method based on comprehensive weighted assessment in multiple scenarios

The complexity and diversity brought by the distributed architecture of the new generation electric information collection system have deepened the difficulty of constructing evaluation verification index systems and quality evaluation models. Moreover, the presence of differentiated components has made fair and scientific verification challenging. Therefore, leveraging graph neural networks and siamese networks, a integrated construction quality evaluation system based on comprehensive weighted assessment in multiple scenarios was developed. Firstly, a graph neural network was constructed based on terminal data of the branches electricity usage information collection system and the link topology structure. Subsequently, this network was deployed on the headquarters side to directly acquire terminal data and generate mirror network input from the branches data, enabling real-time acquisition of various system operational indicators. Finally, the similarity between the headquarters and branches data was calculated using siamese networks to compute accuracy compensation weights for checking and evaluating various indicators, thereby obtaining comprehensive weighted quality evaluation indicators of the branches new generation electric information collection system. We use three types of services including electricity data collection, load forecasting, and task scheduling as experimental scenarios. The results showed that the multidimensional comprehensive weighted quality assessment combined with accuracy compensation obtained from the siamese network resulted in business construction quality assessment values of 97.92%, 95.95%, and 99.96% in branch. This value is approximately equal to the quality evaluation value of manual work, so the method can effectively verify the construction quality of new systems in the branch.

Siamese neural networks can identify subjects from anonymised ECGs

Abstract Background/Introduction Many research databases contain anonymised electrocardiograms (ECGs) linked to other sensitive information. ECGs hold features unique to individuals, potentially enabling subject identification from anonymised ECGs. Purpose We assessed if artificial intelligence approaches to output ECG pair similarity can re-identify individuals from anonymised ECGs. Additionally, we aimed to explore clinical risk prediction using ECG similarity over time. Methods We used a convolutional Siamese neural network (SNN), with a triplet loss function, to train a deep learning model determining if two ECGs belong to the same individual (Figure 1). The model aims to encode ECG inputs from same subjects closer than ECG inputs from different subjects. An ECG similarity score is output, corresponding to the probability two ECGs belong to the same subject. This continuous metric can be used in a binary manner. Below a threshold, two ECGs are classed as from the same individual. Our dataset comprised 72,455 secondary care subjects from a USA cohort with 4 to 75 ECGs each, totalling 864,283 ECGs. These were split 50:10:40% at the subject level for training, validation and testing. Results In 2,689,124 same-subject pairs and 2,689,124 different-subject pairs from the test set, the model achieves an accuracy of 91.68% against the binary threshold. This improves to 93.61% and 95.97% in outpatient and normal ECG subsets, respectively. We tested the model using a subject’s ECG to identify them from a group where only one ECG also belongs to that subject. 89.4% success rates occur in groups of 100 normal ECGs (if random, success would be 1%). 91.4%, 95.1%, and 96.3% success occurs when supplemented with subject gender, age (decade-bracket), and both, respectively. We calculated a model certainty score for this task. 99% success occurs when >95% certain. This is in 65% of group size 100 trials. In groups of 1000 normal ECGs, the success rate is 72.5% with no additional information, and 75.9%, 84.2% and 87.5% when supplemented as before (if random, success would be 0.1%). >95% certainty occurs in 33% of group size 1000 trials. For a given subject, ECG pair similarity proved clinically informative. The model accounts for slight variations within a subject’s ECGs, even when far apart in time (Figure 2A). However, it incorrectly predicts two ECGs as being from different individuals for substantial ECG morphology changes (Figure 2B), which could be used to identify clinical deterioration. The temporal evolution of ECG pair similarity reflects clinical trajectory, with greater dissimilarity associating with all-cause mortality (Hazard ratio, 1.22 per 1 standard deviation change, p < 0.0001). Conclusion(s) Anonymised ECGs retain information useful for subject re-identification. Whilst this implies possible ethical and data protection issues with large datasets, such approaches offer potential for continuous monitoring and identifying clinical deterioration.Figure 1Figure 2

IoU-guided Siamese network with high-confidence template fusion for visual tracking

Existing IoU-guided trackers use IoU score to weight the classification score only in testing phase, this model mismatch between training and testing phases leads to poor tracking performance especially when facing background distractors. In this paper, we propose an IoU-guided Siamese network with High-confidence template fusion (SiamIH) for visual tracking. An IoU-guided distractor suppression network is proposed, which uses IoU information to guide classification in training phase and testing phase, and makes the tracking model to suppress background distractors. To cope with appearance variations, we design a high-confidence template fusion network that fuses APCE-based high-confidence template and the initial template to generate more reliable template. Experimental results on OTB2013, OTB2015, UAV123, LaSOT, and GOT10k demonstrate that the proposed SiamIH achieves state-of-the-art tracking performance.

Cleft Lip and Palate Classification Through Vision Transformers and Siamese Neural Networks

This study introduces a novel approach for the diagnosis of Cleft Lip and/or Palate (CL/P) by integrating Vision Transformers (ViTs) and Siamese Neural Networks. Our study is the first to employ this integration specifically for CL/P classification, leveraging the strengths of both models to handle complex, multimodal data and few-shot learning scenarios. Unlike previous studies that rely on single-modality data or traditional machine learning models, we uniquely fuse anatomical data from ultrasound images with functional data from speech spectrograms. This multimodal approach captures both structural and acoustic features critical for accurate CL/P classification. Employing Siamese Neural Networks enables effective learning from a small number of labeled examples, enhancing the model’s generalization capabilities in medical imaging contexts where data scarcity is a significant challenge. The models were tested on the UltraSuite CLEFT dataset, which includes ultrasound video sequences and synchronized speech data, across three cleft types: Bilateral, Unilateral, and Palate-only clefts. The two-stage model demonstrated superior performance in classification accuracy (82.76%), F1-score (80.00–86.00%), precision, and recall, particularly distinguishing Bilateral and Unilateral Cleft Lip and Palate with high efficacy. This research underscores the significant potential of advanced AI techniques in medical diagnostics, offering valuable insights into their application for improving clinical outcomes in patients with CL/P.

Terahertz Non-Destructive Testing of Porosity in Multi-Layer Thermal Barrier Coatings Based on Small-Sample Data

Accurately characterizing the internal porosity rate of thermal barrier coatings (TBCs) was essential for prolonging their service life. This work concentrated on atmospheric plasma spray (APS)-prepared TBCs and proposed the utilization of terahertz non-destructive detection technology to evaluate their internal porosity rate. The internal porosity rates were ascertained through a metallographic analysis and scanning electron microscopy (SEM), followed by the reconstruction of the TBC model using a four-parameter method. Terahertz time-domain simulation data corresponding to various porosity rates were generated employing the time-domain finite difference method. In simulating actual test signals, white noise with a signal-to-noise ratio of 10 dB was introduced, and various wavelet transforms were utilized for denoising purposes. The effectiveness of different signal processing techniques in mitigating noise was compared to extract key features associated with porosity. To address dimensionality challenges and further enhance model performance, kernel principal component analysis (kPCA) was employed for data processing. To tackle issues related to limited sample sizes, this work proposed to use the Siamese neural network (SNN) and generative adversarial network (GAN) algorithms to solve this challenge in order to improve the generalization ability and detection accuracy of the model. The efficacy of the constructed model was assessed using multiple evaluation metrics; the results indicate that the novel hybrid WT-kPCA-GAN model achieves a prediction accuracy exceeding 0.9 while demonstrating lower error rates and superior predictive performance overall. Ultimately, this work presented an innovative, convenient, non-destructive online approach that was safe and highly precise for measuring the porosity rate of TBCs, particularly in scenarios involving small sample sizes facilitating assessments regarding their service life.

Detecting phase transitions based on siamese neural network

Machine learning has been widely applied in physics research. Although unsupervised learning can extract the critical points of phase transitions, the percolation model remains a challenge. Unsupervised learning using the raw configurations of the percolation model fails to capture the critical points. To capture the configuration characteristics of the percolation model, this paper proposes using the maximum cluster as input to the neural network. It is well understood that the order parameter of the percolation model is not simply the particle density, but rather the probability that a given site or bond belongs to the percolating cluster. Additionally, we introduce the use of a Siamese Neural Network (SNN) to detect percolation phase transitions. Unlike unsupervised dimensionality reduction methods or supervised binary classification outputs, the SNN produces a scalar output referred to as similarity. By combining the maximum cluster and the SNN, we not only successfully extract the critical value of the percolation model, but also calculate the correlation exponent via data collapse. We believe that the SNN has great potential in handling phase transition classification problems and can serve as a reference for studying other phase transition systems.

An optimized siamese neural network with deep linear graph attention model for gynaecological abdominal pelvic masses classification.

An adnexal mass, also known as a pelvic mass, is a growth that develops in or near the uterus, ovaries, fallopian tubes, and supporting tissues. For women suspected of having ovarian cancer, timely and accurate detection of a malignant pelvic mass is crucial for effective triage, referral, and follow-up therapy. While various deep learning techniques have been proposed for identifying pelvic masses, current methods are often not accurate enough and can be computationally intensive. To address these issues, this manuscript introduces an optimized Siamese circle-inspired neural network with deep linear graph attention (SCINN-DLGN) model designed for pelvic mass classification. The SCINN-DLGN model is intended to classify pelvic masses into three categories: benign, malignant, and healthy. Initially, real-time MRI pelvic mass images undergo pre-processing using semantic-aware structure-preserving median morpho-filtering to enhance image quality. Following this, the region of interest (ROI) within the pelvic mass images is segmented using an EfficientNet-based U-Net framework, which reduces noise and improves the accuracy of segmentation. The segmented images are then analysed using the SCINN-DLGN model, which extracts geometric features from the ROI. These features are classified into benign, malignant, or healthy categories using a deep clustering algorithm integrated into the linear graph attention model. The proposed system is implemented on a Python platform, and its performance is evaluated using real-time MRI pelvic mass datasets. The SCINN-DLGN model achieves an impressive 99.9% accuracy and 99.8% recall, demonstrating superior efficiency compared to existing methods and highlighting its potential for further advancement in the field.

SiamVIT: A patchwise network for gamma-ray point source detection

Conventional point source detection methods generally work in a pixelwise manner and can hardly exploit the overall semantic information of sources; consequently, these methods usually suffer from low precision. In this work we achieve point source detection in fully patchwise mode by proposing a siamese network called SiamVIT that includes a visual transformer (VIT). SiamVIT can effectively and accurately locate point sources from gamma -ray maps with high purity not only in higher flux regions, but also in lower flux regions, which is extremely challenging to achieve with state-of-the-art methods. SiamVIT consists of two VIT branches and a matching block. In the feature extraction stage, gamma -ray maps are fed into one VIT branch to obtain patch representations with adequate semantic and contextual information, whereas detection templates with location information are fed into the other branch to produce template representations. In the location stage, a patch representation and all template representations are fed into the matching block to determine whether the associated gamma -ray map patch contains a point source and where that point source is located, if applicable. We compare our proposed SiamVIT with the current advanced methods and find that SiamVIT has significantly better purity and completeness and a superior Dice coefficient on the test set. In addition, when point sources overlap, SiamVIT can better distinguish different point sources.

An ensemble deep learning model for medical image fusion with Siamese neural networks and VGG-19.

Multimodal medical image fusion methods, which combine complementary information from many multi-modality medical images, are among the most important and practical approaches in numerous clinical applications. Various conventional image fusion techniques have been developed for multimodality image fusion. Complex procedures for weight map computing, fixed fusion strategy and lack of contextual understanding remain difficult in conventional and machine learning approaches, usually resulting in artefacts that degrade the image quality. This work proposes an efficient hybrid learning model for medical image fusion using pre-trained and non-pre-trained networks i.e. VGG-19 and SNN with stacking ensemble method. The model leveraging the unique capabilities of each architecture, can effectively preserve the detailed information with high visual quality, for numerous combinations of image modalities in image fusion challenges, notably improved contrast, increased resolution, and lower artefacts. Additionally, this ensemble model can be more robust in the fusion of various combinations of source images that are publicly available from Havard-Medical-Image-Fusion Datasets, GitHub. and Kaggle. Our proposed model performance is superior in terms of visual quality and performance metrics to that of the existing fusion methods in literature like PCA+DTCWT, NSCT, DWT, DTCWT+NSCT, GADCT, CNN and VGG-19.

SiamRCSC: Robust siamese network with channel and spatial constraints for visual object tracking

SiamRCSC: Robust siamese network with channel and spatial constraints for visual object tracking

Feasibility of Transformer Model for User Authentication Using Electromyogram Signals

Transformer models are widely used in natural language processing (NLP) and time-series data analysis. Applications of these models include prediction systems and hand gesture recognition using electromyogram (EMG) signals. However, in the case of time-series analysis, the models perform similarly to traditional networks, contrary to expectations. This study aimed to compare the performance of the transformer model and its various modified versions in terms of accuracy through a user authentication system using EMG signals, which exhibit significant variability and pose challenges in feature extraction. A Siamese network was employed to distinguish subtle differences in the EMG signals between users, using Euclidean distance. Data from 100 individuals were used to create a challenging scenario while ensuring accuracy. Three scenarios were considered: data preprocessing, integration with existing models, and the modification of the internal structure of the transformer model. The method that achieved the highest accuracy was the bidirectional long short-term memory (BiLSTM)–transformer approach. Based on this, a network was further constructed and optimized, resulting in a user authentication accuracy of 99.7% using EMG data from 100 individuals.

The unsupervised bearing fault diagnosis method based on the dual-framework Siamese network

Abstract Aiming at the problem of insufficient bearing fault samples in practical engineering, an unsupervised bearing fault diagnosis method with double frame twin network is proposed in this paper. This method only uses the source domain data to diagnose the target domain data, and can minimize the problem of domain difference. The concrete method is firstly to carry on continuous wavelet transform (CWT) to the rolling bearing signal, and the processed signal is used as the input feature of convolutional neural network (CNN). The twin network structure is optimized and Manhattan distance is used as the loss function in the training process. The proposed CM network is applied to the diagnosis of inner ring fault, outer ring fault and normal state of rolling bearing. The results show that CM network can achieve over 90% accuracy on SQ variable speed vibration signal data set (SQV), Shijiazhuang Railway University data set (STD) and Case Western Reserve University data set (CWRU), which is suitable for bearing fault diagnosis under different working conditions and provides a new direction for fault diagnosis. The PyTorch code is available at https://github.com/xiaotianqu/The-unsupervised-bearing-fault-diagnosis-method-based-on-the-dual-framework-Siamese-network