Explore the vast range of titles available.

Dense correspondence across semantically related images has been extensively studied, but still faces two challenges: 1) large variations in appearance, scale and pose exist even for objects from the same category, and 2) labeling pixel-level dense correspondences is labor intensive and infeasible to scale. Most existing methods focus on designing various matching modules using fully-supervised ImageNet pretrained networks. On the other hand, while a variety of self-supervised approaches are proposed to explicitly measure image-level similarities, correspondence matching the pixel level remains under-explored. In this work, we propose a multi-level contrastive learning approach for semantic matching, which does not rely on any ImageNet pretrained model. We show that image-level contrastive learning is a key component to encourage the convolutional features to find correspondence between similar objects, while the performance can be further enhanced by regularizing cross-instance cycle-consistency at intermediate feature levels. Experimental results on the PF-PASCAL, PF-WILLOW, and SPair-71k benchmark datasets demonstrate that our method performs favorably against the state-of-the-art approaches.

Human Activity Recognition (HAR) using wearable sensors has shown great potential for personalized health management and ubiquitous computing. However, existing deep learning-based HAR models often suffer from poor user-level generalization, which limits their deployment in real-world scenarios. In this work, we propose a novel multi-task contrastive learning framework that jointly optimizes activity classification and supervised contrastive objectives to enhance generalization across unseen users. By leveraging both activity and user labels to construct semantically meaningful contrastive pairs, our method improves representation learning while maintaining user-agnostic inference at test time. We evaluate the proposed framework on three public HAR datasets using cross-user splits, achieving comparable results to both supervised and self-supervised baselines. Extensive ablation studies further confirm the effectiveness of our design choices, including multi-task training and the integration of user-aware contrastive supervision. These results highlight the potential of our approach for building more generalizable and scalable HAR systems.

In this letter, we developed novel Structure Enhanced Channel Attention Network (SECANet) for scene text image super‐resolution (STISR). The newly proposed SECANet integrates a group of Structure‐Enhanced Attention Modules to focus more on both local and global structural features in the character regions of text images. Moreover, we elaborately formulate a Dual‐Domain Contrastive Learning framework that integrates one pixel‐level contrastive loss and the other semantic‐level contrastive loss to jointly optimize the SECANet for generating more visually pleasing yet better recognizable high‐quality SR images without introducing any additional prior generators in both the training and testing stages, showing promising computational efficiency. Experimental results on the Textzoom dataset indicate that our method can achieve both decent performance in super‐resolving more impressive scene text images from low‐resolution ones and better recognition accuracy than other competitors. The proposed structure enhanced channel attention network assembles a group of structure‐enhanced attention blocks to learn both global and local structure features for the detailed recovery of scene text images. Moreover, a joint dual‐domain contrastive loss function is formulated to optimize the model parameters, benefiting to synthesizing more recognizable text images.

We aimed to develop a new artificial intelligence (AI)-based method for evaluating endoscopic ultrasound-guided fine-needle biopsy (EUS-FNB) specimens in pancreatic diseases using deep learning and contrastive learning. We analysed a total of 173 specimens from 96 patients who underwent EUS-FNB with a 22 G Franseen needle for pancreatic diseases. In the initial study, the deep learning method based on stereomicroscopic images of 98 EUS-FNB specimens from 63 patients showed an accuracy of 71.8% for predicting the histological diagnosis, which was lower than that of macroscopic on-site evaluation (MOSE) performed by EUS experts (81.6%). Then, we used image analysis software to mark the core tissues in the photomicrographs of EUS-FNB specimens after haematoxylin and eosin staining and verified whether the diagnostic performance could be improved by applying contrastive learning for the features of the stereomicroscopic images and stained images. The sensitivity, specificity, and accuracy of MOSE were 88.97%, 53.5%, and 83.24%, respectively, while those of the AI-based diagnostic method using contrastive learning were 90.34%, 53.5%, and 84.39%, respectively. The AI-based evaluation method using contrastive learning was comparable to MOSE performed by EUS experts and can be a novel objective evaluation method for EUS-FNB.

Collaborative filtering (CF) aims to make recommendations for users by detecting user’s preference from the historical user–item interactions. Existing graph neural networks (GNN) based methods achieve satisfactory performance by exploiting the high-order connectivity between users and items, however they suffer from the poor training efficiency problem and easily introduce bias for information propagation. Moreover, the widely applied Bayesian personalized ranking (BPR) loss is insufficient to provide supervision signals for training due to the extremely sparse observed interactions. To deal with the above issues, we propose the Efficient Graph Collaborative Filtering (EGCF) method. Specifically, EGCF adopts merely one-layer graph convolution to model the collaborative signal for users and items from the first-order neighbors in the user–item interactions. Moreover, we introduce contrastive learning to enhance the representation learning of users and items by deriving the self-supervisions, which is jointly trained with the supervised learning. Extensive experiments are conducted on two benchmark datasets, i.e., Yelp2018 and Amazon-book, and the experimental results demonstrate that EGCF can achieve the state-of-the-art performance in terms of Recall and normalized discounted cumulative gain (NDCG), especially on ranking the target items at right positions. In addition, EGCF shows obvious advantages in the training efficiency compared with the competitive baselines, making it practicable for potential applications.

Deep neural networks have achieved significant success in the artificial intelligence community and various downstream tasks. They encode images or texts into dense feature representations and are supervised by a large amount of labeled data. Due to the expensiveness of high-quality labeled data, a huge number of easy-to-access instances are collected to conduct supervised learning. However, they have not been annotated by experts and thus can contain numerous noisy instances, which will degrade the performance. To learn robust feature representations despite misleading noisy labels, we employ supervised contrastive learning to directly perform supervision in the hidden space, rather than in the prediction space like the prevalent cross-entropy loss function. However, cutting-edge noisy label learning methods with supervised contrastive learning always discard the data considered to be noisy, and thus cannot tolerate high-ratio noisy datasets. Therefore, we propose a novel training strategy named Supervised Contrastive Learning with Corrected Labels (Scl2) to defend against the attack of noisy labels. Scl2 corrects the noisy labels with an empirical small-loss assumption and conducts supervised contrastive learning using these corrected data. Specifically, we employ the generated soft labels as supervisory information to facilitate our implementation of supervised contrastive learning. This expansion of contrastive learning ensures the integrity of the supervisory information while effectively enhancing the learning process. In addition, samples sharing the same soft labels are treated as positive sample pairs, while those with different soft labels are considered to be negative sample pairs. With this strategy, the representations from neural networks keep the local discrimination in one mini-batch. Besides, we also employ a prototype contrastive learning technique to ensure global discrimination. Our Scl2 has demonstrated excellent performance on numerous benchmark datasets, showcasing its effectiveness in various standardized evaluation scenarios. Additionally, our model has proven to be highly valuable when applied to real-world noisy datasets.

Recent advances in spatial transcriptomics have enabled simultaneous preservation of high‐throughput gene expression profiles and the spatial context, enabling high‐resolution exploration of distinct regional characterization in tissue. To effectively understand the underlying biological mechanisms within tissue microenvironments, there is a requisite for methods that can accurately capture external spatial heterogeneity and interpret internal gene regulation from spatial transcriptomics data. However, current methods for region identification often lack the simultaneous characterizing of spatial structure and gene regulation, thereby limiting the ability of spatial dissection and gene interpretation. Here, stDCL is developed, a dual graph contrastive learning method to identify spatial domains and interpret gene regulation in spatial transcriptomics data. stDCL adaptively incorporates gene expression data and spatial information via a graph embedding autoencoder, thereby preserving critical information within the latent embedding representations. In addition, dual graph contrastive learning is proposed to train the model, ensuring that the latent embedding representation closely resembles the actual spatial distribution and exhibits cluster similarity. Benchmarking stDCL against other state‐of‐the‐art clustering methods using complex cortex datasets demonstrates its superior accuracy and effectiveness in identifying spatial domains. Our analysis of the imputation matrices generated by stDCL reveals its capability to reconstruct spatial hierarchical structures and refine differential expression assessment. Furthermore, it is demonstrated that the versatility of stDCL in interpretability of gene regulation, spatial heterogeneity at high resolution, and embryonic developmental patterns. In addition, it is also showed that stDCL can successfully annotate disease‐associated astrocyte subtypes in Alzheimer's disease and unravel multiple relevant pathways and regulatory mechanisms. stDCL, a dual graph contrastive learning method, captures spatial heterogeneity and interprets gene regulation in spatial transcriptomics data. Integrating spatial and gene expression data through graph embeddings, stDCL provides robust spatial characterization and accurate region identification, reconstructs spatial hierarchies, and identifies disease‐associated cell subtypes, unveiling new insights into tissue microenvironments and disease mechanisms.

Existing contrastive deep graph clustering methods typically employ fixed-threshold strategies when constructing positive and negative sample pairs, and fail to integrate both graph structure information and clustering structure information effectively. However, this fixed-threshold and binary partitioning approach is overly rigid, limiting the model’s utilization of potentially learnable samples. To address this problem, this paper proposes a contrastive deep graph clustering model with a progressive relaxation weighting strategy (DCPRES). By introducing the progressive relaxation weighting strategy (PRES), DCPRES dynamically allocates sample weights, constructing a progressive training strategy from easy to difficult samples. This effectively mitigates the impact of pseudo-label noise and enhances the quality of positive and negative sample pair construction. Building upon this, DCPRES designs two contrastive learning losses: an instance-level loss and a cluster-level loss. These respectively focus on local node information and global cluster distribution characteristics, promoting more robust representation learning and clustering performance. Extensive experiments demonstrated that DCPRES significantly outperforms existing methods on multiple public graph datasets, exhibiting a superior robustness and stability. For instance, on the CORA dataset, our model achieved a significant improvement over the static approach of CCGC, with the NMI increasing by 4.73%, the ACC by 4.77%, the ARI value by 7.03%, and the F1-score by 5.89%. It provides an efficient and stable solution for unsupervised graph clustering tasks.

Background: Deep learning models have been used in the past for non-invasive liver fibrosis classification based on liver ultrasound scans. After numerous improvements in the network architectures, optimizers, and development of hybrid methods, the performance of these models has barely improved. This creates a need for a sophisticated method that helps improve this slow-improving performance. Methods: We propose LivSCP, a method to train liver fibrosis classification models for better accuracy than the traditional supervised learning (SL). Our method needs no changes in the network architecture, optimizer, etc. Results: The proposed method achieves state-of-the-art performance, with an accuracy, precision, recall, and F1-score of 98.10% each, and an AUROC of 0.9972. A major advantage of LivSCP is that it does not require any modification to the network architecture. Our method is particularly well-suited for scenarios with limited labeled data and computational resources. Conclusions: In this work, we successfully propose a training method for liver fibrosis classification models in low-data and computation settings. By comparing the proposed method with our baseline (Vision Transformer with SL) and multiple models, we demonstrate the state-of-the-art performance of our method.

Predicting accurate drug associations, including drug‐target interactions, drug side effects, and drug‐disease relationships, is crucial in biomedical research and precision medicine. Recently, the research community is increasingly adopted graph representation learning methods to investigate drug associations. However, translating advancements in graph pre‐training to the domain of drug development faces significant challenges, particularly in multi‐task learning and few‐shot scenarios. A unified Multi‐task Graph PrompT (MGPT) learning model is proposed providing generalizable and robust graph representations for few‐shot drug association prediction. MGPT constructs a heterogeneous graph network using different entity pairs as nodes and utilizes self‐supervised contrastive learning of sub‐graphs in pre‐training. For downstream tasks, MGPT employs learnable functional prompts, embedded with task‐specific knowledge, to enable robust performance across a range of tasks. MGPT demonstrates the ability of seamless task switching and outperforms competitive approaches in few‐shot scenarios. MGPT emerges as a robust solution to the complexities of multi‐task learning and the challenges associated with limited data in drug development. MGPT is a unified multi‐task graph prompt learning model providing generalizable and robust graph representations for few‐shot drug association prediction. MGPT demonstrates the ability of seamless task switching and outperforms competitive approaches in few‐shot scenarios. MGPT emerges as a robust solution to the complexities of multi‐task learning and the challenges associated with limited data in drug development.