arXiv subjects

Yongjun Xu

Publications and source records attributed to Yongjun Xu.

At least 37 records · Page 2Linked to original sources

arXiv2026

Quantized Visual Geometry Grounded Transformer

Learning-based 3D reconstruction models, represented by Visual Geometry Grounded Transformers (VGGTs), have made remarkable progress with the use of large-scale transformers. Their prohibitive computational and memory costs severely hinder real-world deployment. Post-Training Quantization (PTQ) has become a common practice for compressing and accelerating models. However, we empirically observe that PTQ faces unique obstacles when compressing billion-scale VGGTs: the data-independent special tokens induce heavy-tailed activation distributions, while the multi-view nature of 3D data makes calibration sample selection highly unstable. This paper proposes the first Quantization framework for VGGTs, namely QuantVGGT. This mainly relies on two technical contributions: First, we introduce Dual-Smoothed Fine-Grained Quantization, which integrates pre-global Hadamard rotation and post-local channel smoothing to mitigate heavy-tailed distributions and inter-channel variance robustly. Second, we design Noise-Filtered Diverse Sampling, which filters outliers via deep-layer statistics and constructs frame-aware diverse calibration clusters to ensure stable quantization ranges. Comprehensive experiments demonstrate that QuantVGGT achieves the state-of-the-art results across different benchmarks and bit-width, surpassing the previous state-of-the-art generic quantization method with a great margin. We highlight that our 4-bit QuantVGGT can deliver a 3.7$\times$ memory reduction and 2.5$\times$ acceleration in real-hardware inference, while maintaining reconstruction accuracy above 98\% of its full-precision counterpart. This demonstrates the vast advantages and practicality of QuantVGGT in resource-constrained scenarios. Our code is released in https://github.com/wlfeng0509/QuantVGGT.

cs.CV↗

arXiv2026

QuantSparse: Comprehensively Compressing Video Diffusion Transformer with Model Quantization and Attention Sparsification

Diffusion transformers exhibit remarkable video generation capability, yet their prohibitive computational and memory costs hinder practical deployment. Model quantization and attention sparsification are two promising directions for compression, but each alone suffers severe performance degradation under aggressive compression. Combining them promises compounded efficiency gains, but naive integration is ineffective. The sparsity-induced information loss exacerbates quantization noise, leading to amplified attention shifts. To address this, we propose \textbf{QuantSparse}, a unified framework that integrates model quantization with attention sparsification. Specifically, we introduce \textit{Multi-Scale Salient Attention Distillation}, which leverages both global structural guidance and local salient supervision to mitigate quantization-induced bias. In addition, we develop \textit{Second-Order Sparse Attention Reparameterization}, which exploits the temporal stability of second-order residuals to efficiently recover information lost under sparsity. Experiments on HunyuanVideo-13B demonstrate that QuantSparse achieves 20.88 PSNR, substantially outperforming the state-of-the-art quantization baseline Q-VDiT (16.85 PSNR), while simultaneously delivering a \textbf{3.68$\times$} reduction in storage and \textbf{1.88$\times$} acceleration in end-to-end inference. Our code will be released in https://github.com/wlfeng0509/QuantSparse.

cs.CV↗

arXiv2026

From Isolation to Integration: Building an Adaptive Expert Forest for Pre-Trained Model-based Class-Incremental Learning

Class-Incremental Learning (CIL) requires models to learn new classes without forgetting old ones. A common method is to freeze a pre-trained model and train a new, lightweight adapter for each task. While this prevents forgetting, it treats the learned knowledge as a simple, unstructured collection and fails to use the relationships between tasks. To this end, we propose the Semantic-guided Adaptive Expert Forest (SAEF), a new method that organizes adapters into a structured hierarchy for better knowledge sharing. SAEF first groups tasks into conceptual clusters based on their semantic relationships. Then, within each cluster, it builds a balanced expert tree by creating new adapters from merging the adapters of similar tasks. At inference time, SAEF finds and activates a set of relevant experts from the forest for any given input. The final prediction is made by combining the outputs of these activated experts, weighted by how confident each expert is. Experiments on several benchmark datasets show that SAEF achieves SOTA performance.

cs.LG↗

arXiv2026

Semantic-Guided Dynamic Sparsification for Pre-Trained Model-based Class-Incremental Learning

Class-Incremental Learning (CIL) requires a model to continually learn new classes without forgetting old ones. A common and efficient solution freezes a pre-trained model and employs lightweight adapters, whose parameters are often forced to be orthogonal to prevent inter-task interference. However, we argue that this parameter-constraining method is detrimental to plasticity. To this end, we propose Semantic-Guided Dynamic Sparsification (SGDS), a novel method that proactively guides the activation space by governing the orientation and rank of its subspaces through targeted sparsification. Specifically, SGDS promotes knowledge transfer by encouraging similar classes to share a compact activation subspace, while simultaneously preventing interference by assigning non-overlapping activation subspaces to dissimilar classes. By sculpting class-specific sparse subspaces in the activation space, SGDS effectively mitigates interference without imposing rigid constraints on the parameter space. Extensive experiments on various benchmark datasets demonstrate the state-of-the-art performance of SGDS.

cs.CV↗

arXiv2026

Teacher-Guided Student Self-Knowledge Distillation Using Diffusion Model

Existing Knowledge Distillation (KD) methods often align feature information between teacher and student by exploring meaningful feature processing and loss functions. However, due to the difference in feature distributions between the teacher and student, the student model may learn incompatible information from the teacher. To address this problem, we propose teacher-guided student Diffusion Self-KD, dubbed as DSKD. Instead of the direct teacher-student alignment, we leverage the teacher classifier to guide the sampling process of denoising student features through a light-weight diffusion model. We then propose a novel locality-sensitive hashing (LSH)-guided feature distillation method between the original and denoised student features. The denoised student features encapsulate teacher knowledge and could be regarded as a teacher role. In this way, our DSKD method could eliminate discrepancies in mapping manners and feature distributions between the teacher and student, while learning meaningful knowledge from the teacher. Experiments on visual recognition tasks demonstrate that DSKD significantly outperforms existing KD methods across various models and datasets. Our code is attached in supplementary material.

cs.CV↗

arXiv2026

Low-redundancy Distillation for Continual Learning

Continual learning (CL) aims to learn new tasks without erasing previous knowledge. However, current CL methods primarily emphasize improving accuracy while often neglecting training efficiency, which consequently restricts their practical application. Drawing inspiration from the brain's contextual gating mechanism, which selectively filters neural information and continuously updates past memories, we propose Low-redundancy Distillation (LoRD), a novel CL method that enhances model performance while maintaining training efficiency. This is achieved by eliminating redundancy in three aspects of CL: student model redundancy, teacher model redundancy, and rehearsal sample redundancy. By compressing the learnable parameters of the student model and pruning the teacher model, LoRD facilitates the retention and optimization of prior knowledge, effectively decoupling task-specific knowledge without manually assigning isolated parameters for each task. Furthermore, we optimize the selection of rehearsal samples and refine rehearsal frequency to improve training efficiency. Through a meticulous design of distillation and rehearsal strategies, LoRD effectively balances training efficiency and model precision. Extensive experimentation across various benchmark datasets and environments demonstrates LoRD's superiority, achieving the highest accuracy with the lowest training FLOPs.

cs.LG↗

arXiv2026

Dynamical Adapter Fusion: Constructing A Global Adapter for Pre-Trained Model-based Class-Incremental Learning

Class-Incremental Learning (CIL) requires models to continuously acquire new classes without forgetting previously learned ones. A dominant paradigm involves freezing a pre-trained model and training lightweight, task-specific adapters. However, maintaining task-specific parameters hinders knowledge transfer and incurs high retrieval costs, while naive parameter fusion often leads to destructive interference and catastrophic forgetting. To address these challenges, we propose Dynamical Adapter Fusion (DAF) to construct a single robust global adapter. Grounded in the PAC-Bayes theorem, we derive a fusion mechanism that explicitly integrates three components: the optimized task-specific adapter parameters, the previous global adapter parameters, and the initialization parameters. We utilize the Taylor expansion of the loss function to derive the optimal fusion coefficients, dynamically achieving the best balance between stability and plasticity. Furthermore, we propose a Robust Initialization strategy to effectively capture global knowledge patterns. Experiments on multiple CIL benchmarks demonstrate that DAF achieves state-of-the-art (SOTA) performance.

cs.CV↗

arXiv2026

Movable-Antenna Empowered Backscatter ISAC: Toward Geometry-Adaptive, Low-Power Networks

Backscatter-based integrated sensing and communication (B-ISAC) elevates passive tags into information-bearing scatterers, offering an ultra-low-power path toward dual-function wireless systems. However, this promise is fundamentally undermined by a cascaded backscattering link that suffers from severe double fading and is exquisitely sensitive to geometric misalignment. This article tackles this geometric bottleneck by integrating movable antenna systems (MAS) at the transceiver side. MAS provides real-time, controllable spatial degrees of freedom through sub-wavelength antenna repositioning, enabling active reconfiguration of the cascaded channel without modifying passive tags or consuming additional spectrum. We position this solution within a unified ISAC-backscatter communication-B-ISAC evolution, describe the resulting MAS-assisted B-ISAC architecture and operating principles, and demonstrate its system-level gains through comparative analysis and numerical results. Finally, we showcase the potential of this geometry-adaptive paradigm across key IoT application scenarios, pointing toward future motion-aware wireless networks.

cs.IT↗

arXiv2025

HUTFormer: Hierarchical U-Net Transformer for Long-Term Traffic Forecasting

Traffic forecasting, which aims to predict traffic conditions based on historical observations, has been an enduring research topic and is widely recognized as an essential component of intelligent transportation. Recent proposals on Spatial-Temporal Graph Neural Networks~(STGNNs) have made significant progress by combining sequential models with graph convolution networks. However, due to high complexity issues, STGNNs only focus on short-term traffic forecasting (e.g., 1-h ahead), while ignoring more practical long-term forecasting. In this paper, we make the first attempt to explore long-term traffic forecasting (e.g., 1-day ahead). To this end, we first reveal its unique challenges in exploiting multi-scale representations. Then, we propose a novel Hierarchical U-Net TransFormer~(HUTFormer) to address the issues of long-term traffic forecasting. HUTFormer consists of a hierarchical encoder and decoder to jointly generate and utilize multi-scale representations of traffic data. Specifically, for the encoder, we {\color{black}propose} window self-attention and segment merging to extract multi-scale representations from long-term traffic data. For the decoder, we design a cross-scale attention mechanism to effectively incorporate multi-scale representations. In addition, HUTFormer employs an efficient input embedding strategy to address the complexity issues. Extensive experiments on four traffic datasets show that the proposed HUTFormer significantly outperforms state-of-the-art traffic forecasting and long time series forecasting baselines.

cs.LG↗

arXiv2025

On $4n$-dimensional neither pointed nor semisimple Hopf algebras and the associated weak Hopf algebras

For a class of neither pointed nor semisimple Hopf algebras $H_{4n}$ of dimension $4n$, it is shown that they are quasi-triangular, which universal $R$-matrices are described. The corresponding weak Hopf algebras $\mathfrak{w}H_{4n}$ and their representations are constructed. Finally, their duality and their Green rings are established by generators and relations explicitly. It turns out that the Green rings of the associated weak Hopf algebras are not commutative even if the Green rings of $H_{4n}$ are commutative.

math.RA↗

arXiv2025

IOR: Inversed Objects Replay for Incremental Object Detection

Existing Incremental Object Detection (IOD) methods partially alleviate catastrophic forgetting when incrementally detecting new objects in real-world scenarios. However, many of these methods rely on the assumption that unlabeled old-class objects may co-occur with labeled new-class objects in the incremental data. When unlabeled old-class objects are absent, the performance of existing methods tends to degrade. The absence can be mitigated by generating old-class samples, but it incurs high costs. This paper argues that previous generation-based IOD suffers from redundancy, both in the use of generative models, which require additional training and storage, and in the overproduction of generated samples, many of which do not contribute significantly to performance improvements. To eliminate the redundancy, we propose Inversed Objects Replay (IOR). Specifically, we generate old-class samples by inversing the original detectors, thus eliminating the necessity of training and storing additional generative models. We propose augmented replay to reuse the objects in generated samples, reducing redundant generations. Moreover, we propose high-value knowledge distillation focusing on the positions of old-class objects overwhelmed by the background, which transfers the knowledge to the incremental detector. Extensive experiments conducted on MS COCO 2017 demonstrate that our method can efficiently improve detection performance in IOD scenarios with the absence of old-class objects. The code is available at https://github.com/JiaJia075/IOR.

cs.CV↗

arXiv2025

KnowCoder-A1: Incentivizing Agentic Reasoning Capability with Outcome Supervision for KBQA

Knowledge Base Question Answering (KBQA) aims to answer natural-language questions over a structured Knowledge Base (KB). Recent work improves KBQA by adopting an agentic reasoning paradigm, in which Large Language Models (LLMs) iteratively decompose a question, generate its corresponding logical queries, and interact with the KB to derive the answer. However, these methods typically fine-tune LLMs on reasoning trajectories synthesized via process supervision, which offers weak incentives for exploration and thus fails to strengthen the agentic reasoning ability. In this paper, we propose KnowCoder-A1, an LLM that can autonomously perform agentic reasoning on KBs to obtain answers. To incentivize autonomous exploration, KnowCoder-A1 trains the LLM under outcome-only supervision via a multi-stage curriculum reinforcement learning with an easy-to-hard curriculum. To establish foundational agentic capabilities, KnowCoder-A1 first fine-tunes the LLM on a small set of high-quality trajectories obtained through outcome-based rejection sampling. Then, to alleviate the reward sparsity inherent in outcome-only supervision, it applies multi-stage curriculum RL with reward schedules that progress from easy to hard. Trained with outcome-only supervision, KnowCoder-A1 exhibits powerful reasoning behaviors and consistently outperforms prior approaches across three mainstream datasets. Notably, on the zero-shot subset of GrailQA, KnowCoder-A1 achieves up to an 11.1% relative improvement while using only one-twelfth of the training data, demonstrating strong agentic reasoning capabilities.

cs.AI↗

arXiv2025

APT: Affine Prototype-Timestamp For Time Series Forecasting Under Distribution Shift

Time series forecasting under distribution shift remains challenging, as existing deep learning models often rely on local statistical normalization (e.g., mean and variance) that fails to capture global distribution shift. Methods like RevIN and its variants attempt to decouple distribution and pattern but still struggle with missing values, noisy observations, and invalid channel-wise affine transformation. To address these limitations, we propose Affine Prototype Timestamp (APT), a lightweight and flexible plug-in module that injects global distribution features into the normalization-forecasting pipeline. By leveraging timestamp conditioned prototype learning, APT dynamically generates affine parameters that modulate both input and output series, enabling the backbone to learn from self-supervised, distribution-aware clustered instances. APT is compatible with arbitrary forecasting backbones and normalization strategies while introducing minimal computational overhead. Extensive experiments across six benchmark datasets and multiple backbone-normalization combinations demonstrate that APT significantly improves forecasting performance under distribution shift.

cs.LG↗

arXiv2025

STA-GANN: A Valid and Generalizable Spatio-Temporal Kriging Approach

Spatio-temporal tasks often encounter incomplete data arising from missing or inaccessible sensors, making spatio-temporal kriging crucial for inferring the completely missing temporal information. However, current models struggle with ensuring the validity and generalizability of inferred spatio-temporal patterns, especially in capturing dynamic spatial dependencies and temporal shifts, and optimizing the generalizability of unknown sensors. To overcome these limitations, we propose Spatio-Temporal Aware Graph Adversarial Neural Network (STA-GANN), a novel GNN-based kriging framework that improves spatio-temporal pattern validity and generalization. STA-GANN integrates (i) Decoupled Phase Module that senses and adjusts for timestamp shifts. (ii) Dynamic Data-Driven Metadata Graph Modeling to update spatial relationships using temporal data and metadata; (iii) An adversarial transfer learning strategy to ensure generalizability. Extensive validation across nine datasets from four fields and theoretical evidence both demonstrate the superior performance of STA-GANN.

cs.LG↗

arXiv2025

Revisiting Wireless-Powered MEC: A Cooperative Energy Recycling Framework for Task-Energy Co-Design

Cooperative energy recycling (CER) offers a new way to boost energy utilization in wireless-powered multi-access edge computing (MEC) networks, yet its integration with computation-communication co-design remains underexplored. This paper proposes a CER-enabled MEC framework that maximizes the minimum computable data among users under energy causality, latency, and power constraints. The intractable problem is reformulated into a convex form through relaxation, maximum ratio combining, and variable substitution, and closed-form solutions are derived via Lagrangian duality and alternating optimization, offering analytical insights. Simulation results verify that the proposed CER mechanism markedly increases total computable data while maintaining equitable performance across heterogeneous users.

cs.IT↗

arXiv2025

Selective Learning for Deep Time Series Forecasting

Benefiting from high capacity for capturing complex temporal patterns, deep learning (DL) has significantly advanced time series forecasting (TSF). However, deep models tend to suffer from severe overfitting due to the inherent vulnerability of time series to noise and anomalies. The prevailing DL paradigm uniformly optimizes all timesteps through the MSE loss and learns those uncertain and anomalous timesteps without difference, ultimately resulting in overfitting. To address this, we propose a novel selective learning strategy for deep TSF. Specifically, selective learning screens a subset of the whole timesteps to calculate the MSE loss in optimization, guiding the model to focus on generalizable timesteps while disregarding non-generalizable ones. Our framework introduces a dual-mask mechanism to target timesteps: (1) an uncertainty mask leveraging residual entropy to filter uncertain timesteps, and (2) an anomaly mask employing residual lower bound estimation to exclude anomalous timesteps. Extensive experiments across eight real-world datasets demonstrate that selective learning can significantly improve the predictive performance for typical state-of-the-art deep models, including 37.4% MSE reduction for Informer, 8.4% for TimesNet, and 6.5% for iTransformer.

cs.LG↗

arXiv2025

On the Integration of Spatial-Temporal Knowledge: A Lightweight Approach to Atmospheric Time Series Forecasting

Transformers have gained attention in atmospheric time series forecasting (ATSF) for their ability to capture global spatial-temporal correlations. However, their complex architectures lead to excessive parameter counts and extended training times, limiting their scalability to large-scale forecasting. In this paper, we revisit ATSF from a theoretical perspective of atmospheric dynamics and uncover a key insight: spatial-temporal position embedding (STPE) can inherently model spatial-temporal correlations even without attention mechanisms. Its effectiveness arises from the integration of geographical coordinates and temporal features, which are intrinsically linked to atmospheric dynamics. Based on this, we propose STELLA, a Spatial-Temporal knowledge Embedded Lightweight modeL for ASTF, utilizing only STPE and an MLP architecture in place of Transformer layers. With 10k parameters and one hour of training, STELLA achieves superior performance on five datasets compared to other advanced methods. The paper emphasizes the effectiveness of spatial-temporal knowledge integration over complex architectures, providing novel insights for ATSF. The code is available at https://github.com/GestaltCogTeam/STELLA.

cs.LG↗

arXiv2025

CBPNet: A Continual Backpropagation Prompt Network for Alleviating Plasticity Loss on Edge Devices

To meet the demands of applications like robotics and autonomous driving that require real-time responses to dynamic environments, efficient continual learning methods suitable for edge devices have attracted increasing attention. In this transition, using frozen pretrained models with prompts has become a mainstream strategy to combat catastrophic forgetting. However, this approach introduces a new critical bottleneck: plasticity loss, where the model's ability to learn new knowledge diminishes due to the frozen backbone and the limited capacity of prompt parameters. We argue that the reduction in plasticity stems from a lack of update vitality in underutilized parameters during the training process. To this end, we propose the Continual Backpropagation Prompt Network (CBPNet), an effective and parameter efficient framework designed to restore the model's learning vitality. We innovatively integrate an Efficient CBP Block that counteracts plasticity decay by adaptively reinitializing these underutilized parameters. Experimental results on edge devices demonstrate CBPNet's effectiveness across multiple benchmarks. On Split CIFAR-100, it improves average accuracy by over 1% against a strong baseline, and on the more challenging Split ImageNet-R, it achieves a state of the art accuracy of 69.41%. This is accomplished by training additional parameters that constitute less than 0.2% of the backbone's size, validating our approach.

cs.CV↗