Pattern Recognition
Издатель: Elsevier · ISSN 0031-3203 · 60 статей
- Towards robust models: Mitigating adversarial shifts in feature statistics with uncertainty-aware reconstruction
- Cross-Domain Few-Shot Object Detection needs no RPN
- Fast low rank tensor completion approach with nonconvex bilinear representation
- Efficient and effective deep batch active learning via gradient sparsification
- DR: Optimizing spectral graph neural networks via directional graph rewiring
- HFBFusion: Leveraged high-frequency and boundary guiding for multimodal image fusion
- Self-supervised real-time portrait relighting via facial-symmetry-guided model of illumination decoupling
- Reactivating pre-trained backbone uncertainty for post-hoc evidential deep learning
- A multi-feature shapelet learning framework for interpretable time series classification
- Inverse dynamics-aware offline actor–critic for cross-domain reinforcement learning
- Unsupervised multimodal remote sensing change detection via change-guided selective alignment and commonality-structure collaborative learning
- A novel active sample acquisition strategy with Sinkhorn distance-based sensitivity
- Visual CoT makes VLMs smarter but more fragile
- Boosting transferability with Feature-Statistic Perturbation and Patch-Aligned Shifts
- Bridging theory and algorithm for Generalized Label Shift: An ambiguity perspective
- Rethinking fair graph representation learning via structural rebalancing
- Deep boosting projection completion network for sparse-view helical CT reconstruction
- AeroVerse-SkyNav: Neuroscience-inspired unified vision-language-action model for UAV embodied navigation in open urban environments
- EsophMoE: A knowledge-guided graph mixture-of-experts framework for multi-task clinical prediction in esophageal cancer
- Collaborative object- and Patch–Text Alignment for efficient Vision–Language Foundation Models
- Multimodal Ollivier–Ricci curvature probing for backdoor detection in text-to-image diffusion models
- HSRI: Hybrid Swin-ResNet cross-view feature interaction for biomedical lesion segmentation
- Multi-source heterogeneous data fusion for autonomous driving lane-change with delay-compensated tracking control
- ProDream: Proactive threat dreaming with a latent world model for robust autonomous driving
- S2U-Surv: From seen to unseen sample-specific generation for incomplete multimodal survival prediction
- A deep reinforcement learning framework for slice alignment of spatial transcriptomics with 3D tensor representations
- Distribution-sharing-based modality alignment for federated foundation models in medical image segmentation
- COCO: Conflict-aware Gradient Consistency Learning for cross-modal retrieval with noisy labels
- M3: An integrated multimodal framework for constructing cardiovascular knowledge graphs
- SFPrompt: Spatial-frequency dual prompting for lightweight all-in-one medical image restoration