论文检索

Qi Xu, Shuai Gong, Xuming Ran, Haihua Luo, Yangfan Hu

While deep learning models have shown strong performance in simulating neural responses, they often fail to clearly separate stable visual encoding from condition-specific adaptation, which limits their ability to generalize across stimuli and individuals. We introduce the Adaptive Visual Model (AVM), a structure-preserving framework that enables condition-aware adaptation through modular subnetworks, without modifying the core representation. AVM keeps a Vision Transformer-based encoder frozen to capture consistent visual features, while independently trained modulation paths account for neural response variations driven by stimulus content and subject identity. We evaluate AVM in three experimental settings, including stimulus-level variation, cross-subject generalization, and cross-dataset adaptation, all of which involve structured changes in inputs and individuals. Across two large-scale mouse V1 datasets, AVM outperforms the state-of-the-art V1T model by approximately 2% in predictive correlation, demonstrating robust generalization, interpretable condition-wise modulation, and high architectural efficiency. Specifically, AVM achieves a 9.1% improvement in explained variance (FEVE) under the cross-dataset adaptation setting. These results suggest that AVM provides a unified framework for adaptive neural modeling across biological and experimental conditions, offering a scalable solution under structural constraints. Its design may inform future approaches to cortical modeling in both neuroscience and biologically inspired AI systems.

Meng Xu, Xinhong Chen, Zhongying Chen, Guanyi Zhao, Yang Jin, Jianping Wang

Federated Deep Reinforcement Learning (FDRL) aims to enable distributed collaborative training of multiple DRL models while preserving privacy. Existing FDRL methods function in static client environments, but real-world scenarios often involve dynamic state transitions, such as noise, which render static model topologies inadequate and result in biased policy loss. This degrades client performance and leads to suboptimal global policies. To address this challenge, we develop a generic solution, referred to as the self-regulating training framework, which can be seamlessly integrated into existing FDRL approaches to address dynamic state transitions. Specifically, we propose a Sparse Training (ST) method that dynamically sparsifies and adjusts the topology of each model during training to maximize model performance and reduce model complexity. Additionally, we introduce an auxiliary model to adaptively regulate the policy loss of client models, mitigating loss bias and facilitating updates that yield improved returns. Experimental results demonstrate that our method enhances six state-of-the-art (SOTA) FDRL approaches across nine tasks in terms of return.

Ke Xu, Yixin Wang, Zhongcheng Li, Hao Cui, Jinshui Hu, Xingyi Zhang

Elastic precision quantization enables multi-bit deployment via a single optimization pass, fitting diverse quantization scenarios. Yet, the high storage and optimization costs associated with the Transformer architecture, research on elastic quantization remains limited, particularly for large language models. This paper proposes QuEPT, an efficient post-training scheme that reconstructs block-wise multi-bit errors with one-shot calibration on a small data slice. It can dynamically adapt to various predefined bit-widths by cascading different low-rank adapters, and supports real-time switching between uniform quantization and mixed precision quantization without repeated optimization. To enhance accuracy and robustness, we introduce Multi-Bit Token Merging (MB-ToMe) to dynamically fuse token features across different bit-widths, improving robustness during bit-width switching. Additionally, we propose Multi-Bit Cascaded Low-Rank adapters (MB-CLoRA) to strengthen correlations between bit-width groups, further improve the overall performance of QuEPT. Extensive experiments demonstrate that QuEPT achieves comparable or better performance to existing state-of-the-art post-training quantization methods.

Junlin Xu, Zhuang Zhang, Zhenghang Gong, Jincan Li, Pan Zeng, Zilong Zhang, Xiong Li, Shuting Jin, Haowen Chen, Yajie Meng

Accurate prediction of compound protein interactions (CPIs) is crucial for drug discovery. However, existing deep learning-based methods suffer from hidden biases and poor cross-domain generalization, leading to spurious correlations and inadequate representation of unseen compound-protein pairs. To address these limitations, we propose FuseMine, a multimodal deep learning framework that jointly leverages molecular structures and biological sequences for reliable CPI prediction. Specifically, FuseMine adopts a dual-representation strategy for each molecule. It employs a convolutional encoder to capture structural features, combined with pretrained large language models for extracting semantic information from sequences. We propose a novel Multi-modal Feature Orchestration Aggregation (MFOA) module that enables deep and synergistic fusion between the structural features and the sequential semantics of molecules, effectively capturing the complementary patterns across modalities. Additionally, we design a Reduction Differential Feature Mining (RDFM) module to further enhance the representation of discriminative features, thereby improving the model’s generalization capability. Extensive experiments on multiple benchmark datasets demonstrate that our framework consistently outperforms state-of-the-art methods in both intra-domain and cross-domain scenarios. These results highlight the synergistic value of combining structural and sequential data for CPIs.

Jinfeng Xu, Zheyu Chen, Shuo Yang, Jinze Li, Ziyue Peng, Zewei Liu, Hewei Wang, Jiayi Zhang, Edith C. H. Ngai

Multiple clustering aims to discover diverse latent structures from different perspectives, yet existing methods generate exhaustive clusterings without discerning user interest, necessitating laborious manual screening. Current multi-modal solutions suffer from static semantic rigidity: predefined candidate words fail to adapt to dataset-specific concepts, and fixed fusion strategies ignore evolving feature interactions. To overcome these limitations, we propose Multi-DProxy, a novel multi-modal dynamic proxy learning framework that leverages cross-modal alignment through learnable textual proxies. Multi-DProxy introduces 1) gated cross-modal fusion that synthesizes discriminative joint representations by adaptively modeling feature interactions. 2) dual-constraint proxy optimization where user interest constraints enforce semantic consistency with domain concepts while concept constraints employ hard example mining to enhance cluster discrimination. 3) dynamic candidate management that refines textual proxies through iterative clustering feedback. Therefore, Multi-DProxy not only effectively captures a user's interest through proxies but also enables the identification of relevant clusterings with greater precision. Extensive experiments demonstrate state-of-the-art performance with significant improvements over existing methods across a broad set of multi-clustering benchmarks.

Jiangshuai Xu, Peiyun Xue, Jiacheng Song, Xuhui Huang, Qingshan Hou

Catastrophic forgetting remains a fundamental barrier to artificial continual learning (CL) - a capability innate to humans. Existing CL methods often incur prohibitive computational costs in resource-constrained scenarios. Spiking neural networks (SNNs), with their biological plausibility and energy efficiency, offer distinct advantages for CL. Inspired by cortico-hippocampal memory mechanisms, we propose a spiking neural network framework integrating Hebbian plasticity with meta-learning, named HLML-SNN. This architecture emulates a dual-phase CL process: (1) In the short-term phase, sample-level Hebbian learning rapidly adapts to new inputs through local synaptic updates; (2) In the long-term phase, task-level meta-learning optimizes cross-task parameters using consolidated synaptic weights, mimicking cortical memory integration to refine shared representations and initialize subsequent Hebbian learning. HLML-SNN incrementally transforms short-term adaptations into stable long-term knowledge, where the synergy of rapid synaptic updates and meta-driven global optimization enables efficient continual learning while balancing stability and plasticity. Empirical results establish HLML-SNN's state-of-the-art performance across split-MNIST/CIFAR10/CIFAR100/TinyImageNet while markedly reducing training time compared to existing methods, demonstrating substantial practical potential for rapid deployment scenarios. The code and appendix are available on https://github.com/JiangshuaiXu/HLML SNN.

Hengyuan Xu, Wenjun Ke, Yao He, Jiajun Liu, Dong Nie, Peng Wang, Ziyu Shang, Zijie Xu

Mixture of experts (MoE) dynamically routes inputs to specialized expert networks to scale model capacity with low inference overhead. However, the excessive parameter growth in MoE models poses challenges in low-resource settings. To address these issues, MoE with parameter-efficient fine-tuning (PEFT) methods have emerged as a lightweight adaptation paradigm that distributes knowledge among experts via multiple LoRA blocks. Existing MoE-PEFT methods can be broadly categorized into External and Internal PEFT methods. External PEFT methods incorporate lightweight models into existing MoE architectures without modifying their routing, which limits the model’s parameter efficiency. To overcome these issues, Internal PEFT methods integrate MoE architectures into PEFT, enabling minimal parameter overhead. However, they still face two major challenges: (1) lack of expert functional differentiation, resulting in overlapping specialization across modules, and (2) absence of a structured attribution mechanism to guide expert selection based on semantic relevance. To alleviate these challenges, we propose TopicLoRA, a novel three-stage framework that leverages topic knowledge as semantic anchors to guide expert allocation. Specifically, (1) to address expert redundancy, we construct a topic-level prior graph using Graph Neural Network-enhanced representation learning over Big-Bench categories, enforcing structural separation among expert embeddings, and (2) to introduce semantic attribution, we design a dual-loss training mechanism that softly aligns input-query relevance with topic-guided routing distributions via KL divergence. Extensive experiments on representative datasets (e.g., MMLU, GSM8K, Flanv2) demonstrate that TopicLoRA outperforms state-of-the-art PEFT baselines by 2.40% on average in accuracy. Notably, the maximum improvement is 4.21%. Furthermore, ablation studies demonstrate that our framework's robustness to intricate topics and input sequence variations, which stems from the dual-loss training mechanism.

Haonan Xu, Yang Yang

Out-of-distribution (OOD) detection is a well-known challenge due to deep models often producing overconfident. In this paper, we reveal a key insight that trained classifiers tend to rely on sparse parameter contribution patterns, meaning that only a few dominant parameters drive predictions. This brittleness can be exploited by OOD inputs that anomalously trigger these parameters, resulting in overconfident predictions. To address this issue, we propose a simple yet effective method called Shaping Parameter Contribution Patterns (SPCP), which enhances OOD detection robustness by encouraging the classifier to learn boundary-oriented dense contribution patterns. Specifically, SPCP operates during training by rectifying excessively high parameter contributions based on a dynamically estimated threshold. This mechanism promotes the classifier to rely on a broader set of parameters for decision-making, thereby reducing the risk of overconfident predictions caused by anomalously triggered parameters, while preserving in-distribution (ID) performance. Extensive experiments under various OOD detection setups verify the effectiveness of SPCP.

Ganxi Xu, Jinyi Long, Jia Zhang

Brain decoding currently faces significant challenges in individual differences, modality alignment, and high-dimensional embeddings. To address individual differences, researchers often use source subject data, which leads to issues such as privacy leakage and heavy data storage burdens. In modality alignment, current works focus on aligning the softmax probability distribution but neglect the alignment of marginal probability distributions, resulting in modality misalignment. Additionally, images and text are aligned separately with fMRI without considering the complex interplay between images and text, leading to poor image reconstruction. Finally, the enormous dimensionality of CLIP embeddings causes significant computational costs. Although the dimensionality of CLIP embeddings can be reduced by ignoring the number of patches obtained from images and the number of tokens acquired from text, this comes at the cost of a significant drop in model performance, creating a dilemma. To overcome these limitations, we propose a source-free domain adaptation-based brain decoding framework. Firstly, we apply source-free domain adaptation, which only acquires the source model without accessing source data during target model adaptation, to brain decoding to address cross-subject variations, privacy concerns, and the heavy burden of data storage. Secondly, we employ maximum mean discrepancy (MMD) to align the marginal probability distributions between embeddings of different modalities. Moreover, to accommodate the complex interplay between image and text, we concatenate the embeddings of image and text and then use singular value decomposition (SVD) to obtain a new embedding. What’s more, to achieve better image generation quality, we employ the Wasserstein distance (WD) to align the probability distributions of new embeddings. Finally, in the target model adaptation phase of source-free domain adaptation, we employ low-rank adaptation (LoRA) to reduce the high expense of tuning the target model. Sufficient experiments demonstrate our work outperforms state-of-the-art methods for brain decoding tasks.

Furui Xu, Shaobo Wang, Jiajun Zhang, Chenghao Sun, Haixiang Tang, Linfeng Zhang

The growing scale of datasets in deep learning has introduced significant computational challenges. Dataset pruning addresses this challenge by constructing a compact but informative coreset from the full dataset with comparable performance. Previous approaches typically establish scoring metrics based on specific criteria to identify representative samples. However, these methods predominantly rely on sample scores obtained from the model's performance during the training (i.e., fitting) phase. As scoring models achieve near-optimal performance on training data, such fitting-centric approaches induce a dense distribution of sample scores within a narrow numerical range. This concentration reduces the distinction between samples and hinders effective selection. To address this challenge, we conduct dataset pruning from the perspective of generalization, i.e., scoring samples based on models not exposed to them during training. We propose a plug-and-play framework, UNSEEN, which can be integrated into existing dataset pruning methods. Additionally, conventional score-based methods are single-step and rely on models trained solely on the complete dataset, providing limited perspective on the importance of samples. To address this limitation, we scale UNSEEN to multi-step scenarios and propose an incremental selection technique through scoring models trained on varying coresets, and optimize the quality of the coreset dynamically. Extensive experiments demonstrate that our method significantly outperforms existing state-of-the-art (SOTA) methods on CIFAR-10, CIFAR-100, and ImageNet-1K. Notably, on ImageNet-1K, UNSEEN achieves lossless performance while reducing training data by 30%.

Changjian Xu, Yong Wang, Ruizheng Huang, Zhicheng Zhang, Wen Yin, Kexin Li

Most state-of-the-art time series imputation methods can leverage textual information to improve imputation quality, but they often struggle because they fail to effectively filter noisy information from large language model (LLM) derived textual information. Some existing solutions only filter over the entire token set, which can introduce erroneous conditional constraints, extreme token frequency effects and increased computational complexity. To address this, we propose CaT-Diff, a novel cascaded text-enhanced diffusion model for probabilistic imputation of multivariate time series under Missing Not At Random (MNAR) scenarios. To suppress irrelevant semantics and focus on context most predictive of missing values, CaT-Diff introduces an innovative Hierarchical Semantic Filter (HSF) that collaborates with a Mixture-of-Experts (MoE) Network. The MoE projects heterogeneous text embeddings into the time series latent space, and the HSF cascade-filters text embeddings from the segment level to the token level, thereby avoiding the pitfalls of direct token-level filtering and reducing overhead. We also incorporate a lightweight Missing Mechanism Estimator, jointly optimized with the denoising network to explicitly capture MNAR missingness patterns. Extensive tests on nine domains show that CaT-Diff outperforms state-of-the-art baselines. Our work presents a new approach for selectively fusing LLM-derived textual information.

Boxun Xu, Yu Wang, Zihu Wang, Peng Li

Visual autoregressive modeling (VAR) via next-scale prediction has emerged as a scalable image generation paradigm. While Key and Value (KV) caching in large language models (LLMs) has been extensively studied, next-scale prediction presents unique challenges, and KV caching design for next-scale based VAR transformers remains largely unexplored. A major bottleneck is the excessive KV memory growth with the increasing number of scales—severely limiting scalability. Our systematic investigation reveals that: (1) Attending to tokens from local scales significantly contributes to generation quality (2) Allocating a small amount of memory for the coarsest scales, termed as condensed scales, stabilizes multi-scale image generation (3) Strong KV similarity across finer scales is predominantly observed in cache-efficient layers, whereas cache-demanding layers exhibit weaker inter-scale similarity. Based on the observations, we introduce AMS-KV, a scale-adaptive KV caching policy for next-scale prediction in VAR models. AMS-KV prioritizes storing KVs from condensed and local scales, preserving the most relevant tokens to maintain generation quality. It further optimizes KV cache utilization and computational efficiency identifying cache-demanding layers through inter-scale similarity analysis. Compared to the vanilla next-scale prediction-based VAR models, AMS-KV reduces KV cache usage by up to 84.83% and self-attention latency by 60.48%. Moreover, when the baseline VAR-d30 model encounters out-of-memory failures at a batch size of 128, AMS-KV enables stable scaling to a batch size of 256 with improved throughput.

Binyan Xu, Fan Yang, Di Tang, Xilin Dai, Kehuan Zhang

Clean-image backdoor attacks, which use only label manipulation in training datasets to compromise deep neural networks, pose a significant threat to security-critical applications. A critical flaw in existing methods is that the poison rate required for a successful attack induces a proportional, and thus noticeable, drop in Clean Accuracy (CA), undermining their stealthiness. This paper presents a new paradigm for clean-image attacks that minimizes this accuracy degradation by optimizing the trigger itself. We introduce Generative Clean-Image Backdoors (GCB), a framework that uses a conditional InfoGAN to identify naturally occurring image features that can serve as potent and stealthy triggers. By ensuring these triggers are easily separable from benign task-related features, GCB enables a victim model to learn the backdoor from an extremely small set of poisoned examples, resulting in a CA drop of less than 1%. Our experiments demonstrate GCB's remarkable versatility, successfully adapting to six datasets, five architectures, and four tasks, including the first demonstration of clean-image backdoors in regression and segmentation. GCB also exhibits resilience against most of the existing backdoor defenses.

Beicheng Xu, Wei Liu, Keyao Ding, Yupeng Lu, Bin Cui

The Combined Algorithm Selection and Hyperparameter Optimization (CASH) problem is fundamental in Automated Machine Learning (AutoML). Inspired by the success of ensemble learning, recent AutoML systems construct post-hoc ensembles for final predictions rather than relying on the best single model. However, while most CASH methods conduct extensive searches for the optimal single model, they typically employ fixed strategies during the ensemble phase that fail to adapt to specific task characteristics. To tackle this issue, we propose PSEO, a framework for post-hoc stacking ensemble optimization. First, we conduct base model selection through binary quadratic programming, with a trade-off between diversity and performance. Furthermore, we introduce two mechanisms to fully realize the potential of multi-layer stacking. Finally, PSEO builds a hyperparameter space and searches for the optimal post-hoc ensemble strategy within it. Empirical results on 80 public datasets show that PSEO achieves the best average test rank (2.96) among 16 methods, including post-hoc designs in recent AutoML systems and state-of-the-art ensemble learning methods.

Xuantang Xiong, Ni Mu, Runpeng Xie, Senhao Yang, Yaqing Wang, Lexiang Wang, Yao Luan, Siyuan Li, Shuang Xu, Yiqin Yang 等

Model-based reinforcement learning (MBRL) is a crucial approach to enhance the generalization capabilities and improve the sample efficiency of RL algorithms. However, current MBRL methods focus primarily on building world models for single tasks and rarely address generalization across different scenarios. Building on the insight that dynamics within the same simulation engine share inherent properties, we attempt to construct a unified world model capable of generalizing across different scenarios, named Meta-Regularized Contextual World-Model (MrCoM). This method first decomposes the latent state space into various components based on the dynamic characteristics, thereby enhancing the accuracy of world-model prediction. Further, MrCoM adopts meta-state regularization to extract unified representation of scenario-relevant information, and meta-value regularization to align world-model optimization with policy learning across diverse scenario objectives. We theoretically analyze the generalization error upper bound of MrCoM in multi-scenario settings. We systematically evaluate our algorithm's generalization ability across diverse scenarios, demonstrating significantly better performance than previous state-of-the-art methods.

Lang Xiong, Ning Liu, Ao Ren, Yuheng Bai, Haining Fang, Binyan Zhang, Zhe Jiang, Yujuan Tan, Duo Liu

Large language models (LLMs) face significant deployment challenges due to their massive computational demands. While pruning offers a promising compression solution, existing methods suffer from two critical limitations: (1) They neglect activation distribution shifts between calibration data and test data, resulting in inaccurate error estimations; (2) Overlooking the long-tail distribution characteristics of activations in the attention module. To address these limitations, this paper proposes a novel pruning method, D²Prune. First, we propose a dual Taylor expansion-based method that jointly models weight and activation perturbations for precise error estimation, leading to precise pruning mask selection and weight updating and facilitating error minimization during pruning. Second, we propose an attention-aware dynamic update strategy that preserves the long-tail attention pattern by jointly minimizing the KL divergence of attention distributions and the reconstruction error. Extensive experiments show that D²Prune consistently outperforms SOTA methods across various LLMs (e.g., OPT-125M, LLaMA2/3, Qwen3). Moreover, the dynamic attention update mechanism also generalizes well to ViT-based vision models like DeiT, achieving superior accuracy on ImageNet-1K.

Guojun Xiong, Milind Tambe

In social impact optimization, AI decision systems often rely on solvers that optimize well-calibrated mathematical objectives. However, these solvers cannot directly accommodate evolving human preferences, typically expressed in natural language rather than formal constraints. Recent approaches address this by using large language models (LLMs) to generate new reward functions from preference descriptions. While flexible, they risk sacrificing the system's core utility guarantees. In this paper, we propose VORTEX, a language-guided reward shaping framework that preserves established optimization goals while adaptively incorporating human feedback. By formalizing the problem as multi-objective optimization, we use LLMs to iteratively generate shaping rewards based on verbal reinforcement and text-gradient prompt updates. This allows stakeholders to steer decision behavior via natural language without modifying solvers or specifying trade-off weights. We provide theoretical guarantees that VORTEX converges to Pareto-optimal trade-offs between utility and preference satisfaction. Empirical results in real-world allocation tasks demonstrate that VORTEX outperforms baselines in satisfying human-aligned coverage goals while maintaining high task performance. This work introduces a practical and theoretically grounded paradigm for human-AI collaborative optimization guided by natural language.

Feng Xiong, Zongxia Xie, Yanru Sun, Haoyu Wang, Jianhong Lin

Effective multivariate time series forecasting often benefits from accurately modeling complex inter-variable dependencies. However, existing attention- or graph-based methods face three key issues: (a) strong temporal self-dependencies are often disrupted by irrelevant variables; (b) softmax normalization ignores and reverses negative correlations; (c) variables struggle to perceive their temporal positions. To address these, we propose **SEED**, a Spectral Entropy-guided evaluation framework for spatial-temporal dependency modeling. SEED introduces a Dependency Evaluator, a key innovation that leverages spectral entropy to dynamically provide a preliminary evaluation of the spatial and temporal dependencies of each variable, enabling the model to adaptively balance Channel Independence (CI) and Channel Dependence (CD) strategies. To account for temporal regularities originating from the influence of other variables rather than intrinsic dynamics, we propose Spectral Entropy-based Fuser to further refine the evaluated dependency weights, effectively separating this part. Moreover, to preserve negative correlations, we introduce a Signed Graph Constructor that enables signed edge weights, overcoming the limitations of softmax. Finally, to help variables perceive their temporal positions and thereby construct more comprehensive spatial features, we introduce the Context Spatial Extractor, which leverages local contextual windows to extract spatial features. Extensive experiments on 12 real-world datasets from various application domains demonstrate that SEED achieves state-of-the-art performance, validating its effectiveness and generality.

Yu Xing, Qizhuo Xie, Yunhui Liu, Qing Gu, Tao Zheng, Bin Chong, Tieke He

Multi-modal entity alignment aims to identify equivalent entities across different multi-modal knowledge graphs (MMKGs). While prior work has achieved notable progress through improved multi-modal encoding and cross-modal fusion techniques, two critical challenges remain unresolved. First, due to the heterogeneous and often inconsistent sources from which MMKGs are constructed, the quality and informativeness of modalities vary significantly across entities, leading to the modality weighting problem. Second, existing cross-modal fusion mechanisms predominantly emphasize modality-shared information, often at the expense of modality-specific signals that are also essential for precise alignment. To address these issues, we propose HUMEA, a novel framework that integrates hierarchical Mixture-of-Experts (MoE) with unimodal distillation. HUMEA consists of: (1) A hierarchical MoE module comprising intra-modal and inter-modal experts, which adaptively modulates modality contributions by capturing entity representations at fine-to-coarse semantic granularities. In addition, we introduce a contrastive mutual information loss to enhance expert diversity and reduce redundancy. (2) A unimodal distillation strategy that preserves modality-specific information in the fused representations through single-modality alignment and distillation, achieving a balanced integration of shared and unique modality features. Extensive experiments on two benchmark datasets, FB15K-DB15K and FB15K-YAGO15K, demonstrate state-of-the-art performance, validating the effectiveness of our approach.

Yinghui Xing, Xiaoting Su, Shizhou Zhang, Donghao Chu, Di Xu

Infrared imaging plays a critical role in low-light and adverse weather conditions. However, due to the distinct characteristics of infrared images, existing foundation models such as Masked Autoencoder (MAE) trained on visible data perform suboptimal in infrared image interpretation tasks. To bridge this gap, an infrared foundation model known as InfMAE was developed and pre-trained on large-scale infrared datasets. Despite its effectiveness, InfMAE still faces several limitations, including the omission of informative tokens, insufficient modeling of global associations, and neglect of non-uniform noise. In this paper, we propose a Dual-domain Guided Infrared foundation model based on MAE (DuGI-MAE). First, we design a deterministic masking strategy based on token entropy, preserving only high-entropy tokens for reconstruction to enhance informativeness. Next, we introduce a Dual-Domain Guidance (DDG) module, which simultaneously captures global token relationships and adaptively filters non-uniform background noise commonly present in infrared imagery. To facilitate large-scale pretraining, we construct Inf-590K, a comprehensive infrared image dataset encompassing diverse scenes, various target types, and multiple spatial resolutions. Pretrained on Inf-590K, DuGI-MAE demonstrates strong generalization capabilities across various downstream tasks, including infrared object detection, semantic segmentation, and small target detection. Experimental results validate the superiority of the proposed method over both supervised and self-supervised comparison methods.