The rapid advancement of artificial intelligence (AI) in healthcare has significantly enhanced diagnostic accuracy and clinical decision-making processes. This review examines four pivotal studies that highlight the integration of large language models (LLMs) and multimodal systems in medical diagnostics. BioBERT demonstrates the efficacy of domain-specific pretraining on biomedical texts, improving performance in tasks such as named entity recognition, relation extraction, and question answering. Med-PaLM, a large-scale language model tailored for clinical question answering, leverages instruction prompt tuning to enhance accuracy and reduce harmful outputs, validated through the MultiMedQA benchmark. DR.KNOWS integrates medical knowledge graphs with LLMs, enhancing diagnostic reasoning and interpretability by grounding model predictions in structured medical knowledge. Medical Multimodal Foundation Models (MMFMs) combine textual and imaging data to improve tasks like segmentation, lesion detection, and automated report generation. These studies demonstrate the importance of domain adaptation, structured knowledge integration, and multimodal data fusion in developing robust and interpretable AI-driven diagnostic tools.

Remote sensing image classification is pivotal in applications ranging from environmental monitoring to urban planning. However, the scarcity of labeled data in target domains often impedes the generalization of models trained on source domains. To address this challenge, we propose Semantic-Aware Contrastive Adaptation (SACA), a novel framework that leverages contrastive learning to construct a discriminative and class-balanced embedding space. SACA emphasizes semantic details by contrasting samples with category centroids and other samples in a distribution-aware manner. It first identifies distinctive features using category centroids as semantic anchors, then expands the sample scope by incorporating statistical information from the labeled source domain. This approach enables the embedding space to better reflect the true distribution of each semantic class. Our rigorous theoretical analysis reveals that SACA implicitly considers an infinite number of similar and dissimilar sample pairs, leading to a tighter upper bound on the contrastive loss compared to existing methods. This not only improves computational efficiency but also enhances the robustness of learned representations. To demonstrate SACA’s effectiveness, we conduct extensive experiments on three challenging remote sensing datasets: Hyperion, National Center for Airborne Laser Mapping, andWorldview-2. The results show that SACA significantly boosts the performance of self-training methods, outperforming state-of-the-art approaches by a large margin. In-depth ablation studies provide insights into the contribution of each SACA component. SACA represents a significant step towards bridging the domain gap in remote sensing image classification. By leveraging semantic-aware contrastive learning, it enables models to capture consistent semantic concepts across domains, enhancing transferability and robustness. This work opens new avenues for research in domain adaptation and has the potential to benefit a wide range of real-world applications, from precision agriculture to disaster response.

Industrial multimedia recommendation systems extensively utilize cascade architectures to deliver personalized content for users, generally consisting of multiple stages like retrieval and ranking. However, retrieval models have long suffered from Sample Selection Bias (SSB) due to the distribution discrepancy between the exposed items used for model training and the candidates (almost unexposed) during inference, affecting recommendation performance. Traditional methods utilize retrieval candidates as augmented training data, indiscriminately treating unexposed data as negative samples, which leads to inaccuracies and noise. Some efforts rely on unbiased datasets, while they are costly to collect and insufficient for industrial models. In this paper, we propose a debiasing framework named DAMCAR, which introduces Domain Adaptation to mitigate SSB in Multimedia CAscade Recommendation systems. Firstly, we sample hard-to-distinguish samples from unexposed data to serve as the target domain, optimizing data quality and resource utilization. Secondly, adversarial domain adaptation is employed to generate pseudo-labels for each sample. To enhance robustness, we utilize Exponential Moving Average (EMA) to create a teacher model that supervises the generation of pseudo-labels via self-distillation. Finally, we obtain a retrieval model that maintains stable performance during inference through a hybrid training mechanism. We conduct offline experiments on two real-world datasets and deploy our approach in the retrieval model of a multimedia video recommendation system for online A/B testing. Comprehensive experimental results demonstrate the effectiveness of DAMCAR in practical applications.

Accurate semantic interpretation of 3D point clouds is essential for the digital transformation of architecture, engineering, and construction (AEC). Building Information Modeling (BIM) depends on both geometric precision and semantic consistency, yet raw scans are typically noisy, redundant, and computationally expensive to process. This work presents an Information Bottleneck (IB) formulation that regularizes latent features to preserve only task-relevant information, yielding compact and interpretable representations within point-based neural networks. Our method, named IB-PC (Information Bottleneck for Point Clouds), introduces an Information Bottleneck (IB) layer as an auxiliary regularization task alongside supervised prediction, encouraging information compression and improving model robustness. Evaluations are conducted on two representative benchmarks, Semantic3D (outdoor) and TUM RGB-D (indoor) across, five criteria: (i) segmentation accuracy, (ii) calibration, (iii) robustness to noise and occlusion, (iv) computational efficiency, and (v) structural fidelity of architectural elements. The IB-regularized model consistently improves mean Intersection over Union (mIoU), overall accuracy, and macro F1, while reducing Expected Calibration Error (ECE) and Negative Log-Likelihood (NLL). The model remains stable under noise, occlusion, and varying point densities, and yields more consistent segmentation of architectural components such as walls, floors, and columns. These improvements are achieved with roughly 30% fewer FLOPs and reduced memory consumption, demonstrating the method’s efficiency and suitability for large-scale scan-to-BIM and digital twin applications.