arXiv subjects

Qi Zhang

Publications and source records attributed to Qi Zhang.

At least 397 records · Page 22Linked to original sources

arXiv2025

Stabilizing and Optimizing Inter-Shell Routing in LEO Networks with Integrated Routing Cost

The low Earth orbit (LEO) mega-constellation network (LMCN), which uses thousands of satellites across multi-shell architectures to deliver different services, is facing challenges in inter-shell routing stability due to dynamic network topologies and frequent inter-satellite link (ISL) switching. Existing strategies, such as the Minimum Hop Path set, prioritize minimizing hop counts to reduce latency, but ignore ISL switching costs, which leads to high instability. To overcome this, the Adaptive Path Routing Scheme introduces path similarity thresholds to reduce the ISL switching frequency between shells. However, the greedy approach of Adaptive Path Routing Scheme is often trapped in local optima, sacrificing inter-shell path distance efficiency. To address these limitations, we propose the Dynamic Programming-based Integrated Routing Cost (DP-IRC) algorithm, which is designed explicitly for inter-shell routing optimization. By formulating multi-shell paths as a multistage decision problem, DP-IRC balances hop counts and ISL stability through an Integrated Routing Cost (IRC) metric, combining inter-/intra-shell hops and switching costs. Experiments over 60 time slots with real-world Starlink and OneWeb configurations show that DP-IRC reduces inter-shell ISL switching rates by 39.1% and 22.0% compared to the Minimum Hop Path set strategy and Adaptive Path Routing Scheme, respectively, while still maintaining near-optimal end-to-end distances.

cs.NI↗

arXiv2025

Signal Prediction for Loss Mitigation in Tactile Internet: A Leader-Follower Game-Theoretic Approach

Tactile Internet (TI) requires achieving ultra-low latency and highly reliable packet delivery for haptic signals. In the presence of packet loss and delay, the signal prediction method provides a viable solution for recovering the missing signals. To this end, we introduce the Leader-Follower (LeFo) approach based on a cooperative Stackelberg game, which enables both users and robots to learn and predict actions. With accurate prediction, the teleoperation system can safely relax its strict delay requirements. Our method achieves high prediction accuracy, ranging from 80.62% to 95.03% for remote robot signals at the Human ($H$) side and from 70.44% to 89.77% for human operation signals at the remote Robot ($R$) side. We also establish an upper bound for maximum signal loss using Taylor Expansion, ensuring robustness.

eess.SP↗

arXiv2025

BMMR: A Large-Scale Bilingual Multimodal Multi-Discipline Reasoning Dataset

In this paper, we introduce BMMR, a large-scale bilingual, multimodal, multi-disciplinary reasoning dataset for the community to develop and evaluate large multimodal models (LMMs). BMMR comprises 110k college-level questions spanning 300 UNESCO-defined subjects, spanning diverse formats-multiple-choice, fill-in-the-blank, and open-ended QA-and sourced from both print and digital media such as books, exams, and quizzes. All data are curated and filtered via a human-in-the-loop and scalable framework, and each instance is paired with a high-quality reasoning path. The dataset is organized into two parts: BMMR-Eval that comprises 20,458 high-quality instances to comprehensively assess LMMs' knowledge and reasoning across multiple disciplines in both Chinese and English; and BMMR-Train that contains 88,991 instances to support further research and development, extending the current focus on mathematical reasoning to diverse disciplines and domains. In addition, we propose the process-based multi-discipline verifier (i.e., BMMR-Verifier) for accurate and fine-grained evaluation of reasoning paths. Extensive experiments on 24 models reveal that (i) even SOTA models (e.g., o3 and Gemini-2.5-Pro) leave substantial headroom on BMMR-Eval; (ii) reasoning models exhibit discipline bias and outperform LMMs only on specific subjects; (iii) open-source models still trail their proprietary counterparts; and (iv) fine-tuning on BMMR-Train narrows this gap. Additionally, we conduct reasoning-chain analyses using BMMR-Verifier and other in-depth studies, uncovering the challenges LMMs currently face in multidisciplinary reasoning. We will release the data, and we hope our work can offer insights and contributions to the community.

cs.CL↗

arXiv2025

Improving RL Exploration for LLM Reasoning through Retrospective Replay

Reinforcement learning (RL) has increasingly become a pivotal technique in the post-training of large language models (LLMs). The effective exploration of the output space is essential for the success of RL. We observe that for complex problems, during the early stages of training, the model exhibits strong exploratory capabilities and can identify promising solution ideas. However, its limited capability at this stage prevents it from successfully solving these problems. The early suppression of these potentially valuable solution ideas by the policy gradient hinders the model's ability to revisit and re-explore these ideas later. Consequently, although the LLM's capabilities improve in the later stages of training, it still struggles to effectively address these complex problems. To address this exploration issue, we propose a novel algorithm named Retrospective Replay-based Reinforcement Learning (RRL), which introduces a dynamic replay mechanism throughout the training process. RRL enables the model to revisit promising states identified in the early stages, thereby improving its efficiency and effectiveness in exploration. To evaluate the effectiveness of RRL, we conduct extensive experiments on complex reasoning tasks, including mathematical reasoning and code generation, and general dialogue tasks. The results indicate that RRL maintains high exploration efficiency throughout the training period, significantly enhancing the effectiveness of RL in optimizing LLMs for complicated reasoning tasks. Moreover, it also improves the performance of RLHF, making the model both safer and more helpful.

cs.LG↗

arXiv2025

Perception Activator: An intuitive and portable framework for brain cognitive exploration

Recent advances in brain-vision decoding have driven significant progress, reconstructing with high fidelity perceived visual stimuli from neural activity, e.g., functional magnetic resonance imaging (fMRI), in the human visual cortex. Most existing methods decode the brain signal using a two-level strategy, i.e., pixel-level and semantic-level. However, these methods rely heavily on low-level pixel alignment yet lack sufficient and fine-grained semantic alignment, resulting in obvious reconstruction distortions of multiple semantic objects. To better understand the brain's visual perception patterns and how current decoding models process semantic objects, we have developed an experimental framework that uses fMRI representations as intervention conditions. By injecting these representations into multi-scale image features via cross-attention, we compare both downstream performance and intermediate feature changes on object detection and instance segmentation tasks with and without fMRI information. Our results demonstrate that incorporating fMRI signals enhances the accuracy of downstream detection and segmentation, confirming that fMRI contains rich multi-object semantic cues and coarse spatial localization information-elements that current models have yet to fully exploit or integrate.

cs.CV↗

arXiv2025

Quantum Anharmonic Effects in Hydrogen-Bond Symmetrization of High-Pressure Ice

The nuclear quantum effects of hydrogen play a significant role in determining the phase stability of water ice. Hydrogen-bond symmetrization occurs as hydrogen atoms tunnel in a double-well potential, ultimately occupying the midpoint between oxygen atoms and transforming ice VIII into ice X under high pressure. Quantum fluctuations lower this transition from classical predictions of over 100 GPa to 60 GPa. We reveal that the Perdew-Burke-Ernzerhof functional underestimates the hydrogen double-well barrier, thus resulting in a transition pressure over 10 GPa lower than the strongly constrained and appropriately normed functional, which is validated against quantum Monte Carlo calculations. Nuclear quantum anharmonicity, treated via neural canonical transformation (NCT), reveals that this transition pressure is temperature-independent and observes a 2 GPa reduction when comparing the non-Gaussian flow models based wavefunction compared to the self-consistent harmonic approximation. Despite increasing pressure typically shortens chemical bonds and hardens phonon modes, NCT calculations reveal that hydrogen bond softens hydrogen-oxygen stretching in ice VIII upon pressure.

cond-mat.mtrl-sci↗

arXiv2025

Transformer-Based Person Search with High-Frequency Augmentation and Multi-Wave Mixing

The person search task aims to locate a target person within a set of scene images. In recent years, transformer-based models in this field have made some progress. However, they still face three primary challenges: 1) the self-attention mechanism tends to suppress high-frequency components in the features, which severely impacts model performance; 2) the computational cost of transformers is relatively high. To address these issues, we propose a novel High-frequency Augmentation and Multi-Wave mixing (HAMW) method for person search. HAMW is designed to enhance the discriminative feature extraction capabilities of transformers while reducing computational overhead and improving efficiency. Specifically, we develop a three-stage framework that progressively optimizes both detection and re-identification performance. Our model enhances the perception of high-frequency features by learning from augmented inputs containing additional high-frequency components. Furthermore, we replace the self-attention layers in the transformer with a strategy based on multi-level Haar wavelet fusion to capture multi-scale features. This not only lowers the computational complexity but also alleviates the suppression of high-frequency features and enhances the ability to exploit multi-scale information. Extensive experiments demonstrate that HAMW achieves state-of-the-art performance on both the CUHK-SYSU and PRW datasets.

cs.CV↗

arXiv2025

Seedream 3.0 Technical Report

We present Seedream 3.0, a high-performance Chinese-English bilingual image generation foundation model. We develop several technical improvements to address existing challenges in Seedream 2.0, including alignment with complicated prompts, fine-grained typography generation, suboptimal visual aesthetics and fidelity, and limited image resolutions. Specifically, the advancements of Seedream 3.0 stem from improvements across the entire pipeline, from data construction to model deployment. At the data stratum, we double the dataset using a defect-aware training paradigm and a dual-axis collaborative data-sampling framework. Furthermore, we adopt several effective techniques such as mixed-resolution training, cross-modality RoPE, representation alignment loss, and resolution-aware timestep sampling in the pre-training phase. During the post-training stage, we utilize diversified aesthetic captions in SFT, and a VLM-based reward model with scaling, thereby achieving outputs that well align with human preferences. Furthermore, Seedream 3.0 pioneers a novel acceleration paradigm. By employing consistent noise expectation and importance-aware timestep sampling, we achieve a 4 to 8 times speedup while maintaining image quality. Seedream 3.0 demonstrates significant improvements over Seedream 2.0: it enhances overall capabilities, in particular for text-rendering in complicated Chinese characters which is important to professional typography generation. In addition, it provides native high-resolution output (up to 2K), allowing it to generate images with high visual quality.

cs.CV↗

arXiv2025

Quantum secure direct communication based on fully passive source

In practical quantum communication, imperfect devices may introduce side channels, creating opportunities for eavesdroppers. Especially on the source side, the side channels created by active modulation may compromise the security of the protocol. We proposes a passively-sourced quantum secure direct communication (QSDC) protocol based on fully passive source. By passively modulating both the quantum state and the intensity of the decoy state, we can avoid active modulation operations at the source, thereby enhancing the robustness of QSDC against side-channel attacks. We developed a system model and conducted parameter optimization to obtain the maximum secrecy message transmission rate achievable by the protocol for each channel attenuation. At a channel attenuation of 2, 4, 6 dB (corresponding to a communication distance of 5, 10, 15 km), the secrecy message transmission rates are 5.76 * 10^-5, 9.92 * 10^-6, and 4.99 * 10^-7 bit/sec. And its maximum communication distance is about 16.875 km, which is about 94.4% of that of actively modulated QSDC.

quant-ph↗

arXiv2025

End-to-End RGB-IR Joint Image Compression With Channel-wise Cross-modality Entropy Model

RGB-IR(RGB-Infrared) image pairs are frequently applied simultaneously in various applications like intelligent surveillance. However, as the number of modalities increases, the required data storage and transmission costs also double. Therefore, efficient RGB-IR data compression is essential. This work proposes a joint compression framework for RGB-IR image pair. Specifically, to fully utilize cross-modality prior information for accurate context probability modeling within and between modalities, we propose a Channel-wise Cross-modality Entropy Model (CCEM). Among CCEM, a Low-frequency Context Extraction Block (LCEB) and a Low-frequency Context Fusion Block (LCFB) are designed for extracting and aggregating the global low-frequency information from both modalities, which assist the model in predicting entropy parameters more accurately. Experimental results demonstrate that our approach outperforms existing RGB-IR image pair and single-modality compression methods on LLVIP and KAIST datasets. For instance, the proposed framework achieves a 23.1% bit rate saving on LLVIP dataset compared to the state-of-the-art RGB-IR image codec presented at CVPR 2022.

cs.CV↗

arXiv2025

Recalling The Forgotten Class Memberships: Unlearned Models Can Be Noisy Labelers to Leak Privacy

Machine Unlearning (MU) technology facilitates the removal of the influence of specific data instances from trained models on request. Despite rapid advancements in MU technology, its vulnerabilities are still underexplored, posing potential risks of privacy breaches through leaks of ostensibly unlearned information. Current limited research on MU attacks requires access to original models containing privacy data, which violates the critical privacy-preserving objective of MU. To address this gap, we initiate an innovative study on recalling the forgotten class memberships from unlearned models (ULMs) without requiring access to the original one. Specifically, we implement a Membership Recall Attack (MRA) framework with a teacher-student knowledge distillation architecture, where ULMs serve as noisy labelers to transfer knowledge to student models. Then, it is translated into a Learning with Noisy Labels (LNL) problem for inferring the correct labels of the forgetting instances. Extensive experiments on state-of-the-art MU methods with multiple real datasets demonstrate that the proposed MRA strategy exhibits high efficacy in recovering class memberships of unlearned instances. As a result, our study and evaluation have established a benchmark for future research on MU vulnerabilities.

cs.LG↗

arXiv2025

COLUR: Confidence-Oriented Learning, Unlearning and Relearning with Noisy-Label Data for Model Restoration and Refinement

Large deep learning models have achieved significant success in various tasks. However, the performance of a model can significantly degrade if it is needed to train on datasets with noisy labels with misleading or ambiguous information. To date, there are limited investigations on how to restore performance when model degradation has been incurred by noisy label data. Inspired by the ``forgetting mechanism'' in neuroscience, which enables accelerating the relearning of correct knowledge by unlearning the wrong knowledge, we propose a robust model restoration and refinement (MRR) framework COLUR, namely Confidence-Oriented Learning, Unlearning and Relearning. Specifically, we implement COLUR with an efficient co-training architecture to unlearn the influence of label noise, and then refine model confidence on each label for relearning. Extensive experiments are conducted on four real datasets and all evaluation results show that COLUR consistently outperforms other SOTA methods after MRR.

cs.LG↗

arXiv2025

LettinGo: Explore User Profile Generation for Recommendation System

User profiling is pivotal for recommendation systems, as it transforms raw user interaction data into concise and structured representations that drive personalized recommendations. While traditional embedding-based profiles lack interpretability and adaptability, recent advances with large language models (LLMs) enable text-based profiles that are semantically richer and more transparent. However, existing methods often adhere to fixed formats that limit their ability to capture the full diversity of user behaviors. In this paper, we introduce LettinGo, a novel framework for generating diverse and adaptive user profiles. By leveraging the expressive power of LLMs and incorporating direct feedback from downstream recommendation tasks, our approach avoids the rigid constraints imposed by supervised fine-tuning (SFT). Instead, we employ Direct Preference Optimization (DPO) to align the profile generator with task-specific performance, ensuring that the profiles remain adaptive and effective. LettinGo operates in three stages: (1) exploring diverse user profiles via multiple LLMs, (2) evaluating profile quality based on their impact in recommendation systems, and (3) aligning the profile generation through pairwise preference data derived from task performance. Experimental results demonstrate that our framework significantly enhances recommendation accuracy, flexibility, and contextual awareness. This work enhances profile generation as a key innovation for next-generation recommendation systems.

cs.IR↗

arXiv2025

ALPS: Attention Localization and Pruning Strategy for Efficient Alignment of Large Language Models

Aligning general-purpose large language models (LLMs) to downstream tasks often incurs significant training adjustment costs. Prior research has explored various avenues to enhance alignment efficiency, primarily through minimal-data training or data-driven activations to identify key attention heads. However, these approaches inherently introduce data dependency, which hinders generalization and reusability. To address this issue and enhance model alignment efficiency, we propose the Attention Localization and Pruning Strategy (ALPS), an efficient algorithm that localizes the most task-sensitive attention heads and prunes by restricting attention training updates to these heads, thereby reducing alignment costs. Experimental results demonstrate that our method activates only 10% of attention parameters during fine-tuning while achieving a 2% performance improvement over baselines on three tasks. Moreover, the identified task-specific heads are transferable across datasets and mitigate knowledge forgetting. Our work and findings provide a novel perspective on efficient LLM alignment. The code is available at https://github.com/VoiceBeer/ALPS.

cs.CL↗

arXiv2025

Capability Salience Vector: Fine-grained Alignment of Loss and Capabilities for Downstream Task Scaling Law

Scaling law builds the relationship between training computation and validation loss, enabling researchers to effectively predict the loss trending of models across different levels of computation. However, a gap still remains between validation loss and the model's downstream capabilities, making it untrivial to apply scaling law to direct performance prediction for downstream tasks. The loss typically represents a cumulative penalty for predicted tokens, which are implicitly considered to have equal importance. Nevertheless, our studies have shown evidence that when considering different training data distributions, we cannot directly model the relationship between downstream capability and computation or token loss. To bridge the gap between validation loss and downstream task capabilities, in this work, we introduce Capability Salience Vector, which decomposes the overall loss and assigns different importance weights to tokens to assess a specific meta-capability, aligning the validation loss with downstream task performance in terms of the model's capabilities. Experiments on various popular benchmarks demonstrate that our proposed Capability Salience Vector could significantly improve the predictability of language model performance on downstream tasks.

cs.CL↗

arXiv2025

Efficient Sequential Decision Making with Large Language Models

This paper focuses on extending the success of large language models (LLMs) to sequential decision making. Existing efforts either (i) re-train or finetune LLMs for decision making, or (ii) design prompts for pretrained LLMs. The former approach suffers from the computational burden of gradient updates, and the latter approach does not show promising results. In this paper, we propose a new approach that leverages online model selection algorithms to efficiently incorporate LLMs agents into sequential decision making. Statistically, our approach significantly outperforms both traditional decision making algorithms and vanilla LLM agents. Computationally, our approach avoids the need for expensive gradient updates of LLMs, and throughout the decision making process, it requires only a small number of LLM calls. We conduct extensive experiments to verify the effectiveness of our proposed approach. As an example, on a large-scale Amazon dataset, our approach achieves more than a 6x performance gain over baselines while calling LLMs in only 1.5% of the time steps.

cs.LG↗

arXiv2025

Conditional Diffusion Model-Driven Generative Channels for Double RIS-Aided Wireless Systems

With the development of the upcoming sixth-generation networks (6G), reconfigurable intelligent surfaces (RISs) have gained significant attention due to its ability of reconfiguring wireless channels via smart reflections. However, traditional channel state information (CSI) acquisition techniques for double-RIS systems face challenges (e.g., high pilot overhead or multipath interference). This paper proposes a new channel generation method in double-RIS communication systems based on the tool of conditional diffusion model (CDM). The CDM is trained on synthetic channel data to capture channel characteristics. It addresses the limitations of traditional CSI generation methods, such as insufficient model understanding capability and poor environmental adaptability. We provide a detailed analysis of the diffusion process for channel generation, and it is validated through simulations. The simulation results demonstrate that the proposed CDM based method outperforms traditional channel acquisition methods in terms of normalized mean squared error (NMSE). This method offers a new paradigm for channel acquisition in double-RIS systems, which is expected to improve the quality of channel acquisition with low pilot overhead.

eess.SP↗

arXiv2025

A Survey on Deep Learning based Time Series Analysis with Frequency Transformation

Recently, frequency transformation (FT) has been increasingly incorporated into deep learning models to significantly enhance state-of-the-art accuracy and efficiency in time series analysis. The advantages of FT, such as high efficiency and a global view, have been rapidly explored and exploited in various time series tasks and applications, demonstrating the promising potential of FT as a new deep learning paradigm for time series analysis. Despite the growing attention and the proliferation of research in this emerging field, there is currently a lack of a systematic review and in-depth analysis of deep learning-based time series models with FT. It is also unclear why FT can enhance time series analysis and what its limitations are in the field. To address these gaps, we present a comprehensive review that systematically investigates and summarizes the recent research advancements in deep learning-based time series analysis with FT. Specifically, we explore the primary approaches used in current models that incorporate FT, the types of neural networks that leverage FT, and the representative FT-equipped models in deep time series analysis. We propose a novel taxonomy to categorize the existing methods in this field, providing a structured overview of the diverse approaches employed in incorporating FT into deep learning models for time series analysis. Finally, we highlight the advantages and limitations of FT for time series modeling and identify potential future research directions that can further contribute to the community of time series analysis.

cs.LG↗