Abstract Genomic selection (GS) leverages genome-wide markers and phenotypes to predict breeding values, with its effectiveness largely dependent on the accuracy of genomic prediction (GP) models. However, GP methods often struggle to capture inter-individual variability and are limited by the curse of dimensionality, where the number of SNPs far exceeds the sample size. To address these challenges, we present CLCNet (Contrastive Learning and Chromosome-aware Network), a novel deep learning framework that integrates contrastive learning and chromosome-aware feature modeling. CLCNet comprises two key components: (i) a contrastive learning module that enhances the model’s ability to capture fine-grained, genotype-dependent phenotypic differences among individuals, and (ii) a chromosome-aware module that captures structured feature selection at both chromosome and genome levels, thereby distilling the most informative SNPs. We evaluated CLCNet across four crop species, covering ten agronomically important traits, and compared it with a diverse set of classical linear, machine learning, and deep learning models. CLCNet achieved superior prediction performance, with statistically significant improvements in Pearson correlation coefficient (PCC), ranging from 0.34% to 12.19% over baseline, together with reduced mean squared error (MSE). Performance gains were more pronounced for traits with moderate linkage disequilibrium (LD; r 2 = 0.21-0.36) and high heritability ( h 2 > 0.66), such as those in maize, rapeseed, and soybean. For cotton traits characterized by high LD ( r 2 = 0.74) and lower heritability ( h 2 < 0.50), CLCNet maintained robust performance without degradation. Overall, these results demonstrate that CLCNet is an effective framework for improving genomic prediction accuracy and holds strong potential for practical applications in plant breeding. Short abstract CLCNet is a novel deep learning framework for genomic prediction that integrates contrastive learning with chromosome-aware feature selection. By jointly modeling inter-individual genotype–phenotype variation and chromosomal genomic structure, CLCNet improves prediction accuracy under high-dimensional, low-sample-size conditions. Across four crop species and ten agronomic traits, CLCNet consistently outperformed classical statistical, machine learning, and existing deep learning models. The framework also identified biologically relevant SNPs and candidate genes, demonstrating its potential for practical applications in genomic selection and computational plant breeding. Key points We propose CLCNet, a multi-task deep learning framework that integrates contrastive learning with chromosome-aware feature selection for genomic prediction, under high-dimensional, low-sample-size conditions. The chromosome-aware module explicitly exploits genomic structural information to select representative and informative SNPs across chromosomes. Contrastive learning improves model robustness by stabilizing representation learning and reducing the influence of random effects across samples. By complementing GWAS analyses, CLCNet provides additional insights into genotype–phenotype relationships with potential relevance for gene discovery. Biographical Note Jiangwei Huang is a PhD candidate at the Institute of Genetics and Developmental Biology, Chinese Academy of Sciences. His research focuses on genomic prediction, deep learning, and computational plant breeding. Zhihan Yang is a PhD candidate at the Institute of Genetics and Developmental Biology, Chinese Academy of Sciences. Her research interests include genomic prediction and bioinformatics. Rongcheng Han is an associate professor at the Institute of Genetics and Developmental Biology, Chinese Academy of Sciences. His research focuses on bioinformatics and plant phenomics. Yuqiang Jiang is a professor at the Institute of Genetics and Developmental Biology, Chinese Academy of Sciences. His research interests include plant genomics, plant phenomics and genetic improvement. Organization description The Institute of Genetics and Developmental Biology, Chinese Academy of Sciences, is a leading research institute focusing on genetics, genomics, molecular breeding, bioinformatics, and systems biology in plants and animals.

Diffusion models represent the state-of-the-art in generative modeling. Due to their high training costs, many works leverage pre-trained diffusion models' powerful representations for downstream tasks, such as face super-resolution (FSR), through fine-tuning or prior-based methods. However, relying solely on priors without supervised training makes it challenging to meet the pixel-level accuracy requirements of discrimination task. Although prior-based methods can achieve high fidelity and high-quality results, ensuring consistency remains a significant challenge. In this paper, we propose a masking strategy with strong and weak constraints and iterative refinement for real-world FSR, termed Diffusion Prior Interpolation (DPI). We introduce conditions and constraints on consistency by masking different sampling stages based on the structural characteristics of the face. Furthermore, we propose a condition Corrector (CRT) to establish a reciprocal posterior sampling process. DPI can balance consistency and diversity and can be seamlessly integrated into pre-trained models. In extensive experiments conducted on synthetic and real datasets, along with consistency validation in face recognition, DPI demonstrates superiority over SOTA FSR methods.

Recently, video-based large language models (video-based LLMs) have achieved impressive performance across various video comprehension tasks. However, this rapid advancement raises significant privacy and security concerns, particularly regarding the unauthorized use of personal video data in automated annotation by video-based LLMs. These unauthorized annotated video-text pairs can then be used to improve the performance of downstream tasks, such as text-to-video generation. To safeguard personal videos from unauthorized use, we propose two series of protective video watermarks with imperceptible adversarial perturbations, named Ramblings and Mutes. Concretely, Ramblings aim to mislead video-based LLMs into generating inaccurate captions for the videos, thereby degrading the quality of video annotations through inconsistencies between video content and captions. Mutes, on the other hand, are designed to prompt video-based LLMs to produce exceptionally brief captions, lacking descriptive detail. Extensive experiments demonstrate that our video watermarking methods effectively protect video data by significantly reducing video annotation performance across various video-based LLMs, showcasing both stealthiness and robustness in protecting personal video content. Our code is available at https://github.com/ttthhl/Protecting_Your_Video_Content.

Human-machine interaction through augmented reality (AR) and virtual reality (VR) is increasingly prevalent, requiring accurate and efficient gaze estimation which hinges on the accuracy of eye segmentation to enable smooth user experiences. We introduce EyeSeg, a novel eye segmentation framework designed to overcome key challenges that existing approaches struggle with: motion blur, eyelid occlusion, and train-test domain gaps. In these situations, existing models struggle to extract robust features, leading to suboptimal performance. Noting that these challenges can be generally quantified by uncertainty, we design EyeSeg as an uncertainty-aware eye segmentation framework for AR/VR wherein we explicitly model the uncertainties by performing Bayesian uncertainty learning of a posterior under the closed set prior. Theoretically, we prove that a statistic of the learned posterior indicates segmentation uncertainty levels and empirically outperforms existing methods in downstream tasks, such as gaze estimation. EyeSeg outputs an uncertainty score and the segmentation result, weighting and fusing multiple gaze estimates for robustness, which proves to be effective especially under motion blur, eyelid occlusion and cross-domain challenges. Moreover, empirical results suggest that EyeSeg achieves segmentation improvements of MIoU, E1, F1, and ACC surpassing previous approaches.

Learning 3D representation plays a critical role in masked autoencoder (MAE) based pre-training methods for point cloud, including single-modal and cross-modal based MAE. Specifically, although cross-modal MAE methods learn strong 3D representations via the auxiliary of other modal knowledge, they often suffer from heavy computational burdens and heavily rely on massive cross-modal data pairs that are often unavailable, which hinders their applications in practice. Instead, single-modal methods with solely point clouds as input are preferred in real applications due to their simplicity and efficiency. However, such methods easily suffer from limited 3D representations with global random mask input. To learn compact 3D representations, we propose a simple yet effective Point Feature Enhancement Masked Autoencoders (Point-FEMAE), which mainly consists of a global branch and a local branch to capture latent semantic features. Specifically, to learn more compact features, a share-parameter Transformer encoder is introduced to extract point features from the global and local unmasked patches obtained by global random and local block mask strategies, followed by a specific decoder to reconstruct. Meanwhile, to further enhance features in the local branch, we propose a Local Enhancement Module with local patch convolution to perceive fine-grained local context at larger scales. Our method significantly improves the pre-training efficiency compared to cross-modal alternatives, and extensive downstream experiments underscore the state-of-the-art effectiveness, particularly outperforming our baseline (Point-MAE) by 5.16%, 5.00%, and 5.04% in three variants of ScanObjectNN, respectively. Code is available at https://github.com/zyh16143998882/AAAI24-PointFEMAE.

Knowledge distillation is a widely adopted model compression technique that has been successfully applied to object detection. In feature distillation, it is common practice for the student model to imitate the feature responses of the teacher model, with the underlying objective of improving its own abilities by reducing the disparity with the teacher. However, it is crucial to recognize that the disparities between the student and teacher are inconsistent, highlighting their varying abilities. In this paper, we explore the inconsistency in the disparity between teacher and student feature maps and analyze their impact on the efficiency of the distillation. We find that regions with varying degrees of difference should be treated separately, with different distillation constraints applied accordingly. We introduce our distillation method called Disparity Feature Distillation(DFD). The core idea behind DFD is to apply different treatments to regions with varying learning difficulties, simultaneously incorporating leniency and strictness. It enables the student to better assimilate the teacher’s knowledge. Through extensive experiments, we demonstrate the effectiveness of our proposed DFD in achieving significant improvements. For instance, when applied to detectors based on ResNet50 such as RetinaNet, FasterRCNN, and RepPoints, our method enhances their mAP from 37.4%, 38.4%, 38.6% to 41.7%, 42.4%, 42.7%, respectively. Our approach also demonstrates substantial improvements on YOLO and ViT-based models. The code is available at https://github.com/luckin99/DFD.

Deep neural networks (DNNs) have witnessed remarkable achievement in image super-resolution (SR), and plenty of DNN-based SR models with elaborated network designs have recently been proposed. However, existing methods usually require substantial computations by operating in spatial domain. To address this issue, we propose a general frequency-oriented framework (FSR) to accelerate SR networks by considering data characteristics in frequency domain. Our FSR mainly contains dual feature aggregation module (DFAM) to extract informative features in both spatial and transform domains, followed by a four-path SR-Module with different capacities to super-resolve in the frequency domain. Specifically, DFAM further consists of a transform attention block (TABlock) and a spatial context block (SCBlock) to extract global spectral information and local spatial information, respectively, while SR-Module is a parallel network container that contains four to-be-accelerated branches. Furthermore, we propose an adaptive weight strategy for a trade-off between image details recovery and visual quality. Extensive experiments show that our FSR can save FLOPs by almost 40% while reducing inference time by 50% for other SR methods (e.g., FSRCNN, CARN, SRResNet and RCAN). Code is available at https://github.com/THU-Kingmin/FSR.