Positive Sample Pairs Research Articles

GC-STCL: A Granger Causality-Based Spatial-Temporal Contrastive Learning Framework for EEG Emotion Recognition.

EEG signals capture information through multi-channel electrodes and hold promising prospects for human emotion recognition. However, the presence of high levels of noise and the diverse nature of EEG signals pose significant challenges, leading to potential overfitting issues that further complicate the extraction of meaningful information. To address this issue, we propose a Granger causal-based spatial-temporal contrastive learning framework, which significantly enhances the ability to capture EEG signal information by modeling rich spatial-temporal relationships. Specifically, in the spatial dimension, we employ a sampling strategy to select positive sample pairs from individuals watching the same video. Subsequently, a Granger causality test is utilized to enhance graph data and construct potential causality for each channel. Finally, a residual graph convolutional neural network is employed to extract features from EEG signals and compute spatial contrast loss. In the temporal dimension, we first apply a frequency domain noise reduction module for data enhancement on each time series. Then, we introduce the Granger-Former model to capture time domain representation and calculate the time contrast loss. We conduct extensive experiments on two publicly available sentiment recognition datasets (DEAP and SEED), achieving 1.65% improvement of the DEAP dataset and 1.55% improvement of the SEED dataset compared to state-of-the-art unsupervised models. Our method outperforms benchmark methods in terms of prediction accuracy as well as interpretability.

Interpolation-Based Contrastive Learning for Few-Label Semi-Supervised Learning.

Semi-supervised learning (SSL) has long been proved to be an effective technique to construct powerful models with limited labels. In the existing literature, consistency regularization-based methods, which force the perturbed samples to have similar predictions with the original ones have attracted much attention for their promising accuracy. However, we observe that the performance of such methods decreases drastically when the labels get extremely limited, e.g., 2 or 3 labels for each category. Our empirical study finds that the main problem lies with the drift of semantic information in the procedure of data augmentation. The problem can be alleviated when enough supervision is provided. However, when little guidance is available, the incorrect regularization would mislead the network and undermine the performance of the algorithm. To tackle the problem, we: 1) propose an interpolation-based method to construct more reliable positive sample pairs and 2) design a novel contrastive loss to guide the embedding of the learned network to change linearly between samples so as to improve the discriminative capability of the network by enlarging the margin decision boundaries. Since no destructive regularization is introduced, the performance of our proposed algorithm is largely improved. Specifically, the proposed algorithm outperforms the second best algorithm (Comatch) with 5.3% by achieving 88.73% classification accuracy when only two labels are available for each class on the CIFAR-10 dataset. Moreover, we further prove the generality of the proposed method by improving the performance of the existing state-of-the-art algorithms considerably with our proposed strategy. The corresponding code is available at https://github.com/xihongyang1999/ICL_SSL.

Towards better laparoscopic video segmentation: A class-wise contrastive learning approach with multi-scale feature extraction.

The task of segmentation is integral to computer-aided surgery systems. Given the privacy concerns associated with medical data, collecting a large amount of annotated data for training is challenging. Unsupervised learning techniques, such as contrastive learning, have shown powerful capabilities in learning image-level representations from unlabelled data. This study leverages classification labels to enhance the accuracy of the segmentation model trained on limited annotated data. The method uses a multi-scale projection head to extract image features at various scales. The partitioning method for positive sample pairs is then improved to perform contrastive learning on the extracted features at each scale to effectively represent the differences between positive and negative samples in contrastive learning. Furthermore, the model is trained simultaneously with both segmentation labels and classification labels. This enables the model to extract features more effectively from each segmentation target class and further accelerates the convergence speed. The method was validated using the publicly available CholecSeg8k dataset for comprehensive abdominal cavity surgical segmentation. Compared to select existing methods, the proposed approach significantly enhances segmentation performance, even with a small labelled subset (1-10%) of the dataset, showcasing a superior intersection over union (IoU)score.

Patch-Mixing Contrastive Regularization for Few-Label Semi-Supervised Learning

Recently, consistency regularization has become a fundamental component in semi-supervised learning, which tries to make the network's predictions on unlabeled data to be invariant to perturbations. However, its performance decreases drastically when there are scarce labels, e.g., 2 labels per category. In this paper, we analyze the semantic bias problem in consistency regularization for semi-supervised learning and find that this problem stems from imposing consistency regularization on some semantically biased positive sample pairs derived from indispensable data augmentation. Based on the above analysis, we propose a patch-mixing contrastive regularization approach called <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"><tex-math notation="LaTeX">$p$</tex-math></inline-formula> -Mix, for semi-supervised learning with scarce labels. In <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"><tex-math notation="LaTeX">$p$</tex-math></inline-formula> -Mix, the magnitude of semantic bias is estimated by weighting augmented samples in the embedding space. Specifically, the samples are mixed in both sample space and embedding space respectively, to construct more reliable and task-relevant positive sample pairs. Then, a patch-mixing contrastive objective is designed to indicate the magnitude of semantic bias by utilizing a mixed embedding weighted by virtual soft labels. Extensive experiments were conducted, demonstrating that <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"><tex-math notation="LaTeX">$p$</tex-math></inline-formula> -Mix significantly outperforms current state-of-the-art approaches. Especially, <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"><tex-math notation="LaTeX">$p$</tex-math></inline-formula> -Mix achieves an accuracy of 91.95% on the CIFAR-10 benchmark with only 2 labels available for each category, which exceeds the second-best method ICL-SSL by 3.22%. Source code is available at <uri xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">https://github.com/XiaochangHu/P-Mix</uri>

Deep Contrastive Clustering via Hard positive sample Debiased

Deep graph clustering aims to reveal the underlying information of the graph and provide accurate embedding for the node clustering task, in which contrastive learning plays an important role. However, the commonly used contrastive loss function incorrectly classifies elements outside the diagonal of the cross view as negative samples, which contain a large number of positive sample pairs. In order to overcome the above problems, we propose a new deep graph contrastive clustering method, which combines hard positive sample debiasing and sample pair weighting, and improves the recognition ability of the network by removing the potential positive sample pairs and hard sample pair weighting in the negative sample pair of the loss function. More specifically, we developed a symmetric graph neural network to encode node representations. Using two sets of node representations, the correctness of negative cases is increased by clustering to generate high-confidence pseudo-label pairs for labels and confidence. The similarity distribution differences are weighted by adapting to different dataset samples to improve the sample recognition ability. To verify the efficacy of our DCHD, we compare it to existing state-of-the-art methods for node clustering tasks on six real-world datasets. Overall, the experimental results show that our proposed method outperforms current state-of-the-art graph clustering methods.

A self-supervised learning of semantic feature consistency for image clustering

Contrastive learning is a powerful technique for learning feature representations without manual annotation. The K-nearest neighbor (KNN) method is commonly used to construct positive sample pairs to calculate the contrastive loss. However, it is challenging to distinguish positive sample pairs, reducing clustering performance. We propose a novel Deep Contrastive Clustering method based on a GrapH convolutional network called GHDCC. It uses an instance-level contrastive loss with mean square error (MSE) regularization and a cluster-level contrastive loss to incorporate semantic features and perform cluster assignments. The method utilizes a graph convolutional network (GCN) to improve the semantic consistency of features and linear interpolation data augmentation to improve the representation ability of the model. To minimize the occurrence of false positive sample pairs, we select only samples whose similarity exceeds a predefined threshold to construct the adjacency matrix. The experimental results on six public datasets demonstrate that the GHDCC significantly outperforms contrastive clustering (CC, 500) by a large margin except on CIFAR-10. The GHDCC performs well compared to other deep contrastive clustering methods and achieves the highest clustering accuracy of 0.913 on ImageNet-10.

Asymmetric Graph Contrastive Learning

Learning effective graph representations in an unsupervised manner is a popular research topic in graph data analysis. Recently, contrastive learning has shown its success in unsupervised graph representation learning. However, how to avoid collapsing solutions for contrastive learning methods remains a critical challenge. In this paper, a simple method is proposed to solve this problem for graph representation learning, which is different from existing commonly used techniques (such as negative samples or predictor network). The proposed model mainly relies on an asymmetric design that consists of two graph neural networks (GNNs) with unequal depth layers to learn node representations from two augmented views and defines contrastive loss only based on positive sample pairs. The simple method has lower computational and memory complexity than existing methods. Furthermore, a theoretical analysis proves that the asymmetric design avoids collapsing solutions when training together with a stop-gradient operation. Our method is compared to nine state-of-the-art methods on six real-world datasets to demonstrate its validity and superiority. The ablation experiments further validated the essential role of the asymmetric architecture.

Self-supervised learning representation for abnormal acoustic event detection based on attentional contrastive learning

Most abnormal acoustic event detection (AAED) is completed by supervised training of deep learning methods, but manually labeled samples are costly and scarce. This work proposes a self-supervised learning representation for AAED based on contrastive learning to overcome the abovementioned problem. Auditory and visual data augmentations are applied simultaneously to create positive sample pairs. An attention mechanism is introduced into the encoder during self-supervised pre-training. A comparison between fused features by discriminant correlation analysis and a single feature is made to verify the ability of feature grasping for the self-supervised pre-trained model. The pre-training is completed on an abnormal acoustic dataset with noise. Research results show that the self-supervised pre-trained model can achieve an accuracy of 87.72% in linear evaluation and 88.70% in the downstream task with a pure small AAED dataset, which directly exceeds the results of supervised learning. This work releases the stress of the demand for abnormal acoustic event labels.

Direct comparison of clinical diagnostic sensitivity of saliva from buccal swabs versus combined oro-/nasopharyngeal swabs in the detection of SARS-CoV-2 B.1.1.529 Omicron

While current guidelines recommend the use of respiratory tract specimens for the direct detection of SARS-CoV-2 infection, saliva has recently been suggested as preferred sample type for the sensitive detection of SARS-CoV-2 B.1.1.529 (Omicron). By comparing saliva collected using buccal swabs and oro-/nasopharyngeal swabs from patients hospitalized due to COVID-19, we aimed at identifying potential differences in virus detection sensitivity between these sample types. We compare the clinical diagnostic sensitivity of paired buccal swabs and combined oro-/nasopharyngeal swabs from hospitalized, symptomatic COVID-19 patients collected at median six days after symptom onset by real-time polymerase chain reaction (PCR) and antigen test. Of the tested SARS-CoV-2 positive sample pairs, 55.8% were identified as SARS-CoV-2 Omicron BA.1 and 44.2% as Omicron BA.2. Real-time PCR from buccal swabs generated significantly higher quantification cycle (Cq) values compared to those from matched combined oro-/nasopharyngeal swabs and resulted in an increased number of false-negative PCR results. Reduced diagnostic sensitivity of buccal swabs by real-time PCR was observed already at day one after symptom onset. Similarly, antigen test detection rates were reduced in buccal swabs compared to combined oro-/nasopharyngeal swabs. Our results suggest reduced clinical diagnostic sensitivity of saliva collected using buccal swabs when compared to combined oro-/nasopharyngeal swabs in the detection of SARS-CoV-2 Omicron in symptomatic individuals.

Effective sample pairs based contrastive learning for clustering

As an indispensable branch of unsupervised learning, deep clustering is rapidly emerging along with the growth of deep neural networks. Recently, contrastive learning paradigm has been combined with deep clustering to achieve more competitive performance. However, previous works mostly employ random augmentations to construct sample pairs for contrastive clustering. Different augmentations of a sample are treated as positive sample pairs, which may result in false positives and ignore the semantic variations of different samples. To address these limitations, we present a novel end-to-end contrastive clustering framework termed Contrastive Clustering with Effective Sample pairs construction (CCES), which obtains more semantic information by jointly leveraging an effective data augmentation method ContrastiveCrop and constructing positive sample pairs based on nearest-neighbor mining. Specifically, we augment original samples by adopting ContrastiveCrop, which explicitly reduces false positives and enlarges the variance of samples. Further, with the extracted feature representations, we provide a strategy to construct positive sample pairs via a sample and its nearest neighbor for instance-wise and cluster-wise contrastive learning. Experimental results on four challenging datasets demonstrate the effectiveness of CCES for clustering, which surpasses the state-of-the-art deep clustering methods.

Hard Sample Aware Network for Contrastive Deep Graph Clustering

Contrastive deep graph clustering, which aims to divide nodes into disjoint groups via contrastive mechanisms, is a challenging research spot. Among the recent works, hard sample mining-based algorithms have achieved great attention for their promising performance. However, we find that the existing hard sample mining methods have two problems as follows. 1) In the hardness measurement, the important structural information is overlooked for similarity calculation, degrading the representativeness of the selected hard negative samples. 2) Previous works merely focus on the hard negative sample pairs while neglecting the hard positive sample pairs. Nevertheless, samples within the same cluster but with low similarity should also be carefully learned. To solve the problems, we propose a novel contrastive deep graph clustering method dubbed Hard Sample Aware Network (HSAN) by introducing a comprehensive similarity measure criterion and a general dynamic sample weighing strategy. Concretely, in our algorithm, the similarities between samples are calculated by considering both the attribute embeddings and the structure embeddings, better revealing sample relationships and assisting hardness measurement. Moreover, under the guidance of the carefully collected high-confidence clustering information, our proposed weight modulating function will first recognize the positive and negative samples and then dynamically up-weight the hard sample pairs while down-weighting the easy ones. In this way, our method can mine not only the hard negative samples but also the hard positive sample, thus improving the discriminative capability of the samples further. Extensive experiments and analyses demonstrate the superiority and effectiveness of our proposed method. The source code of HSAN is shared at https://github.com/yueliu1999/HSAN and a collection (papers, codes and, datasets) of deep graph clustering is shared at https://github.com/yueliu1999/Awesome-Deep-Graph-Clustering on Github.

PSNSleep: a self-supervised learning method for sleep staging based on Siamese networks with only positive sample pairs.

Traditional supervised learning methods require large quantities of labeled data. However, labeling sleep data according to polysomnography by well-trained sleep experts is a very tedious job. In the present day, the development of self-supervised learning methods is making significant progress in many fields. It is also possible to apply some of these methods to sleep staging. This is to remove the dependency on labeled data at the stage of representation extraction. Nevertheless, they often rely too much on negative samples for sample selection and construction. Therefore, we propose PSNSleep, a novel self-supervised learning method for sleep staging based on Siamese networks. The crucial step to the success of our method is to select appropriate data augmentations (the time shift block) to construct the positive sample pair. PSNSleep achieves satisfactory results without relying on any negative samples. We evaluate PSNSleep on Sleep-EDF and ISRUC-Sleep and achieve accuracy of 80.0% and 74.4%. The source code is publicly available at https://github.com/arthurxl/PSNSleep.

Semi-supervised LDA pedestrian re-identification algorithm based on K-nearest neighbor resampling

Person re-identification identify a specific person in surveillance network by similarity measurement between images of different camera views. However, existing metric learning based methods suffer from over-fitting problem. To solve this problem, a resampled linear discriminant analysis (LDA) method was proposed based on the statistical and topological characteristics of pedestrian images. This method utilized the k-nearest neighbours to form potential positive sample pairs. The potential positive pairs are used to improve the metric model and generalize the metric model to the test data. By minimizing the inter-class divergence of potential positive sample pairs, a semi-supervised re-sampling LDA person re-identification algorithm was established. It was then tested on the VIPeR, CUHK01 and Market 1501datasets. The results show that the proposed method achieves the best performance compared to some available methods. Especially, the proposed method outplays the best comparison method by 0.6% and 5.76% at rank-1 identification rate on the VIPeR and CUHK01 datasets respectively. At the same time, the improved LDA algorithm has improved the rank-1 identification accuracy of traditional LDA method by 9.36% and 32.11% on these two datasets respectively. Besides, the proposed method is limited to Market-1501 dataset when the test data is of large size.

Spectrum Sensing Algorithm Based on Self-Supervised Contrast Learning

The traditional spectrum sensing algorithm based on deep learning requires a large number of labeled samples for model training, but it is difficult to obtain them in the actual sensing scene. This paper applies self-supervised contrast learning in order to solve this problem, and a spectrum sensing algorithm based on self-supervised contrast learning (SSCL) is proposed. The algorithm mainly includes two stages: pre-training and fine-tuning. In the pre-training stage, according to the characteristics of communication signals, data augmentation methods are designed to obtain the pre-trained positive sample pairs, and the features of the positive sample pairs of unlabeled samples are extracted by self-supervised contrast learning to obtain the feature extractor. In the fine-tuning stage, the parameters of the feature extraction layer are frozen, and a small number of labeled samples are used to update the parameters of the classification layer, and the features and labels are connected to get the spectrum sensing classifier. The simulation results demonstrate that the SSCL algorithm has better detection performance over the semi-supervised algorithm and the traditional energy detection algorithm. When the number of labeled samples used is only 10% of the supervised algorithm and the SNR is higher than −12 dB, the detection probability of the SSCL algorithm is higher than 97%, which is slightly lower than the supervised algorithm.

VLCDoC: Vision-Language contrastive pre-training model for cross-Modal document classification

Multimodal learning from document data has achieved great success lately as it allows to pre-train semantically meaningful features as a prior into a learnable downstream task. In this paper, we approach the document classification problem by learning cross-modal representations through language and vision cues, considering intra- and inter-modality relationships. Instead of merging features from different modalities into a joint representation space, the proposed method exploits high-level interactions and learns relevant semantic information from effective attention flows within and across modalities. The proposed learning objective is devised between intra- and inter-modality alignment tasks, where the similarity distribution per task is computed by contracting positive sample pairs while simultaneously contrasting negative ones in the joint representation space. Extensive experiments on public benchmark datasets demonstrate the effectiveness and the generality of our model both on low-scale and large-scale datasets.

Pig Face Recognition Based on Metric Learning by Combining a Residual Network and Attention Mechanism

As machine vision technology has advanced, pig face recognition has gained wide attention as an individual pig identification method. This study establishes an improved ResNAM network as a backbone network for pig face image feature extraction by combining an NAM (normalization-based attention module) attention mechanism and a ResNet model to probe non-contact open-set pig face recognition. Then, an open-set pig face recognition framework is designed by integrating three loss functions and two metrics to finish the task with no crossover of individuals in the training and test sets. The SphereFace loss function with the cosine distance as a metric and ResNAM are combined in the framework to obtain the optimal open-set pig face recognition model. To train our model, 37 pigs with a total of 12,993 images were randomly selected from the collected pig face images, and 9 pigs with a total of 3431 images were set as a test set. 900 pairs of positive sample pairs and 900 pairs of negative pairs were obtained from the images in the test set. A series of experimental results show that our accuracy reached 95.28%, which was 2.61% higher than that of a human face recognition model. NAM was more effective in improving the performance of the pig face recognition model than the mainstream BAM (bottleneck attention module) and CBAM (convolutional block attention module). The research results can provide technological support for non-contact open-set individual recognition for intelligent farming processes.

Supervised Contrastive Learning Based on Fusion of Global and Local Features for Remote Sensing Image Retrieval

With the rapid development of remote sensing sensor technology, the number of remote sensing images (RSIs) has exploded. How to effectively retrieve and manage these massive data has become an urgent problem. At present, content-based image retrieval (CBIR) methods have become a mainstream method due to their excellent performance. However, most of the existing retrieval methods only consider the global features of images, which lacks the ability to discriminate images with the same semantic information but different visual representations. To alleviate this issue, supervised contrastive learning based on the fusion of global and local features method is proposed in this paper, named SCFR. Firstly, a fusion module is designed to combine global and local features to enhance the ability of image expression. Secondly, supervised contrastive learning is introduced into the retrieval task to effectively improve the feature distribution, so that the positive sample pairs are close to each other, and the negative sample pairs are far away from each other in the feature space. Furthermore, to make the distribution of features of the same class more compact, the center contrastive loss is added to the constraints, and combines the class centers that change iteratively with the network. Experimental results on three RSI datasets show that our proposed method has more effective retrieval performance than the state-of-the-art methods. The code and models are available at https://github.com/xdplay17/SCFR.

A Self-Supervised Model Advance OCTA Image Disease Diagnosis

Due to the lack of medical image datasets, transfer learning/fine-tuning is generally used to realize disease detection (mainly the ImageNet transfer model). Significant differences of dominance between natural and medical images seriously restrict the performance of the model. In this paper, a contrastive learning method (BY-OCTA) combined with patient metadata is proposed to detect the pathology in fundus OCTA images. This method uses the patient’s metadata to construct positive sample pairs. By introducing super-parameters into the loss function, we can reasonably adjust the approximate proportion of the same patient metadata sample pair, so as to produce a better representation and initialization model. This paper evaluates the performance of downstream tasks by fine-tuning the multi-layer perceptron of the model. Experiments show that the linear model pretrained by BY-OCTA is better than that pretrained by ImageNet and BYOL on multiple datasets. Furthermore, in the case of limited labeled training data, BY-OCTA provides the most significant benefit. This shows that the BY-OCTA pretraining model has better characterization extraction ability and transferability. This method allows a flexible combination of medical opinions and uses metadata to construct positive sample pairs, which can be widely used in medical image interpretation.

SCAN: Scattering Characteristics Analysis Network for Few-Shot Aircraft Classification in High-Resolution SAR Images

Recently, deep learning in synthetic aperture radar (SAR) automatic target recognition (ATR) has made significant progress, but the sample limitation problem in the SAR field is still obvious. Compared with the optical remote sensing images, the SAR images are insufficient, especially those containing the geospatial targets with certain target attitude angles (TAAs). To solve these problems, a novel few-shot learning framework named scattering characteristics analysis network (SCAN) is proposed in this article. First, a scattering extraction module (SEM) is designed to combine the target imaging mechanism with the network, which learns the number and distribution of the scattering points for each target type via explicit supervision. Besides, considering the imaging variability of SAR targets, a TAA-guided metalearning network consisting of an angle self-adaption classifier (ASC) and a frequency embedded module (FEM) is designed. ASC guides the network to focus on the positive sample pairs with different TAAs. FEM combines pulse cosine transform (PCT) with the network training process effectively to enrich frequency-domain information. In addition, a new dataset named SAR aircraft category dataset is constructed for the experiments. Compared with other few-shot SAR target classification approaches, our model efficiently integrates the scattering characteristics with the learning process, and the test accuracy for 5-way 1-shot has been improved by 4.74%. Finally, the experimental results are provided to demonstrate the validity of the proposed method.

Comparison of Oral Fluid and Urine for Detection of Fentanyl Use Using Liquid Chromatography with Tandem Mass Spectrometry.

We compared oral fluid (OF) and urine (UR) for detection of fentanyl (FEN) use in addiction medicine-psychiatry (AMP) clinics. We measured FEN and norfentanyl (NRFEN) in UR with a limit of detection (LOD) of 2.0 µg/L and FEN in OF with an LOD of 0.5 µg/L by LC-MS/MS in 311 paired samples and compared the 2 matrices when higher OF and UR LODs were used. Urine (UR) detected more FEN use than OF using a LOD of 2.0 µg/L and 0.5 µg/L, respectively. FEN and/or NRFEN were detected in 44 and 59 UR specimens, respectively, and FEN in 46 OF specimens (43 OF+UR+, 3 OF+UR-, 16 OF-UR+, and 249 OF-UR-). In UR there were no instances with FEN positive and NORFEN negative. UR creatinine was <20 mg/dL in the 3 OF+UR- specimen pairs. The median OF/UR analyte concentration ratios in positive sample pairs were 0.23 for OF FEN/UR FEN and 0.02 for OF FEN/UR NRFEN. We demonstrate that UR detects more FEN use than OF in an AMP setting when UR FEN and UR NORFEN LODs of 2.0 µg/L are used. OF is less sensitive than UR in detecting FEN use, but is still valuable for cases with low UR creatinine and/or suspected adulteration or substitution of UR. The UR vs OF comparison statistics are greatly impacted by even minimal adjustments of the LOD.