arXiv subjects

Xu Zhang

Publications and source records attributed to Xu Zhang.

At least 253 records · Page 14Linked to original sources

arXiv2024

Personalized Prompt for Sequential Recommendation

Pre-training models have shown their power in sequential recommendation. Recently, prompt has been widely explored and verified for tuning in NLP pre-training, which could help to more effectively and efficiently extract useful knowledge from pre-training models for downstream tasks, especially in cold-start scenarios. However, it is challenging to bring prompt-tuning from NLP to recommendation, since the tokens in recommendation (i.e., items) do not have explicit explainable semantics, and the sequence modeling should be personalized. In this work, we first introduces prompt to recommendation and propose a novel Personalized prompt-based recommendation (PPR) framework for cold-start recommendation. Specifically, we build the personalized soft prefix prompt via a prompt generator based on user profiles and enable a sufficient training of prompts via a prompt-oriented contrastive learning with both prompt- and behavior-based augmentations. We conduct extensive evaluations on various tasks. In both few-shot and zero-shot recommendation, PPR models achieve significant improvements over baselines on various metrics in three large-scale open datasets. We also conduct ablation tests and sparsity analysis for a better understanding of PPR. Moreover, We further verify PPR's universality on different pre-training models, and conduct explorations on PPR's other promising downstream tasks including cross-domain recommendation and user profile prediction.

cs.IR↗

arXiv2024

All-in-One Image Coding for Joint Human-Machine Vision with Multi-Path Aggregation

Image coding for multi-task applications, catering to both human perception and machine vision, has been extensively investigated. Existing methods often rely on multiple task-specific encoder-decoder pairs, leading to high overhead of parameter and bitrate usage, or face challenges in multi-objective optimization under a unified representation, failing to achieve both performance and efficiency. To this end, we propose Multi-Path Aggregation (MPA) integrated into existing coding models for joint human-machine vision, unifying the feature representation with an all-in-one architecture. MPA employs a predictor to allocate latent features among task-specific paths based on feature importance varied across tasks, maximizing the utility of shared features while preserving task-specific features for subsequent refinement. Leveraging feature correlations, we develop a two-stage optimization strategy to alleviate multi-task performance degradation. Upon the reuse of shared features, as low as 1.89% parameters are further augmented and fine-tuned for a specific task, which completely avoids extensive optimization of the entire model. Experimental results show that MPA achieves performance comparable to state-of-the-art methods in both task-specific and multi-objective optimization across human viewing and machine analysis tasks. Moreover, our all-in-one design supports seamless transitions between human- and machine-oriented reconstruction, enabling task-controllable interpretation without altering the unified model. Code is available at https://github.com/NJUVISION/MPA.

cs.CV↗

arXiv2024

Simpler Gradient Methods for Blind Super-Resolution with Lower Iteration Complexity

We study the problem of blind super-resolution, which can be formulated as a low-rank matrix recovery problem via vectorized Hankel lift (VHL). The previous gradient descent method based on VHL named PGD-VHL relies on additional regularization such as the projection and balancing penalty, exhibiting a suboptimal iteration complexity. In this paper, we propose a simpler unconstrained optimization problem without the above two types of regularization and develop two new and provable gradient methods named VGD-VHL and ScalGD-VHL. A novel and sharp analysis is provided for the theoretical guarantees of our algorithms, which demonstrates that our methods offer lower iteration complexity than PGD-VHL. In addition, ScalGD-VHL has the lowest iteration complexity while being independent of the condition number. Furthermore, our novel analysis reveals that the blind super-resolution problem is less incoherence-demanding, thereby eliminating the necessity for incoherent projections to achieve linear convergence. Empirical results illustrate that our methods exhibit superior computational efficiency while achieving comparable recovery performance to prior arts.

cs.IT↗

arXiv2024

Projected Gradient Descent for Spectral Compressed Sensing via Symmetric Hankel Factorization

Current spectral compressed sensing methods via Hankel matrix completion employ symmetric factorization to demonstrate the low-rank property of the Hankel matrix. However, previous non-convex gradient methods only utilize asymmetric factorization to achieve spectral compressed sensing. In this paper, we propose a novel nonconvex projected gradient descent method for spectral compressed sensing via symmetric factorization named Symmetric Hankel Projected Gradient Descent (SHGD), which updates only one matrix and avoids a balancing regularization term. SHGD reduces about half of the computation and storage costs compared to the prior gradient method based on asymmetric factorization. {Besides, the symmetric factorization employed in our work is completely novel to the prior low-rank factorization model, introducing a new factorization ambiguity under complex orthogonal transformation}. Novel distance metrics are designed for our factorization method and a linear convergence guarantee to the desired signal is established with $O(r^2\log(n))$ observations. Numerical simulations demonstrate the superior performance of the proposed SHGD method in phase transitions and computation efficiency compared to state-of-the-art methods.

cs.IR↗

arXiv2024

Enhancing Jailbreak Attacks with Diversity Guidance

As large language models(LLMs) become commonplace in practical applications, the security issues of LLMs have attracted societal concerns. Although extensive efforts have been made to safety alignment, LLMs remain vulnerable to jailbreak attacks. We find that redundant computations limit the performance of existing jailbreak attack methods. Therefore, we propose DPP-based Stochastic Trigger Searching (DSTS), a new optimization algorithm for jailbreak attacks. DSTS incorporates diversity guidance through techniques including stochastic gradient search and DPP selection during optimization. Detailed experiments and ablation studies demonstrate the effectiveness of the algorithm. Moreover, we use the proposed algorithm to compute the risk boundaries for different LLMs, providing a new perspective on LLM safety evaluation.

cs.CL↗

arXiv2024

Collaborative Group: Composed Image Retrieval via Consensus Learning from Noisy Annotations

Composed image retrieval extends content-based image retrieval systems by enabling users to search using reference images and captions that describe their intention. Despite great progress in developing image-text compositors to extract discriminative visual-linguistic features, we identify a hitherto overlooked issue, triplet ambiguity, which impedes robust feature extraction. Triplet ambiguity refers to a type of semantic ambiguity that arises between the reference image, the relative caption, and the target image. It is mainly due to the limited representation of the annotated text, resulting in many noisy triplets where multiple visually dissimilar candidate images can be matched to an identical reference pair (i.e., a reference image + a relative caption). To address this challenge, we propose the Consensus Network (Css-Net), inspired by the psychological concept that groups outperform individuals. Css-Net comprises two core components: (1) a consensus module with four diverse compositors, each generating distinct image-text embeddings, fostering complementary feature extraction and mitigating dependence on any single, potentially biased compositor; (2) a Kullback-Leibler divergence loss that encourages learning of inter-compositor interactions to promote consensual outputs. During evaluation, the decisions of the four compositors are combined through a weighting scheme, enhancing overall agreement. On benchmark datasets, particularly FashionIQ, Css-Net demonstrates marked improvements. Notably, it achieves significant recall gains, with a 2.77% increase in R@10 and 6.67% boost in R@50, underscoring its competitiveness in addressing the fundamental limitations of existing methods.

cs.CV↗

arXiv2024

Enhancing Adaptive Deep Networks for Image Classification via Uncertainty-aware Decision Fusion

Handling varying computational resources is a critical issue in modern AI applications. Adaptive deep networks, featuring the dynamic employment of multiple classifier heads among different layers, have been proposed to address classification tasks under varying computing resources. Existing approaches typically utilize the last classifier supported by the available resources for inference, as they believe that the last classifier always performs better across all classes. However, our findings indicate that earlier classifier heads can outperform the last head for certain classes. Based on this observation, we introduce the Collaborative Decision Making (CDM) module, which fuses the multiple classifier heads to enhance the inference performance of adaptive deep networks. CDM incorporates an uncertainty-aware fusion method based on evidential deep learning (EDL), that utilizes the reliability (uncertainty values) from the first c-1 classifiers to improve the c-th classifier' accuracy. We also design a balance term that reduces fusion saturation and unfairness issues caused by EDL constraints to improve the fusion quality of CDM. Finally, a regularized training strategy that uses the last classifier to guide the learning process of early classifiers is proposed to further enhance the CDM module's effect, called the Guided Collaborative Decision Making (GCDM) framework. The experimental evaluation demonstrates the effectiveness of our approaches. Results on ImageNet datasets show CDM and GCDM obtain 0.4% to 2.8% accuracy improvement (under varying computing resources) on popular adaptive networks. The code is available at the link https://github.com/Meteor-Stars/GCDM_AdaptiveNet.

cs.CV↗

arXiv2024

Quantum highway: Observation of minimal and maximal speed limits for few and many-body states

Tracking the time evolution of a quantum state allows one to verify the thermalization rate or the propagation speed of correlations in generic quantum systems. Inspired by the energy-time uncertainty principle, bounds have been demonstrated on the maximal speed at which a quantum state can change, resulting in immediate and practical tasks. Based on a programmable superconducting quantum processor, we test the dynamics of various emulated quantum mechanical systems encompassing single- and many-body states. We show that one can test the known quantum speed limits and that modifying a single Hamiltonian parameter allows the observation of the crossover of the different bounds on the dynamics. We also unveil the observation of minimal quantum speed limits in addition to more common maximal ones, i.e., the lowest rate of change of a unitarily evolved quantum state. Our results establish a comprehensive experimental characterization of quantum speed limits and pave the way for their subsequent study in engineered non-unitary conditions.

quant-ph↗

arXiv2024

Phases of (2+1)D SO(5) non-linear sigma model with a topological term on a sphere: multicritical point and disorder phase

Novel critical phenomena beyond the Landau-Ginzburg-Wilson paradigm have been long sought after. Among many candidate scenarios, the deconfined quantum critical point (DQCP) constitutes the most fascinating one, and its lattice model realization has been debated over the past two decades. Here we apply the spherical Landau level regularization upon the exact (2+1)D SO(5) non-linear sigma model with a topological term to study the potential DQCP therein. We perform density matrix renormalization group (DMRG) simulation with SU(2)$_\mathrm{spin}\times$U(1)$_\mathrm{charge}\times$U(1)$_\mathrm{angular-momentum}$ symmetries explicitly implemented. Using crossing point analysis for the critical properties of the DMRG data, accompanied by quantum Monte Carlo simulations, we accurately obtain the comprehensive phase diagram of the model and find various novel quantum phases, including Néel, ferromagnet (FM), valence bond solid (VBS), valley polarized (VP) states and a gapless quantum disordered phase occupying extended area of the phase diagram. The VBS-Disorder and Néel-Disorder transitions are continuous with non-Wilson-Fisher exponents. Our results show the VBS and Néel states are separated by either a weakly first-order transition or the disordered region with a multicritical point in between, thus opening up more interesting questions on the two-decade long debate on the nature of DQCP.

cond-mat.str-el↗

arXiv2024

Multi-Agent Continuous Control with Generative Flow Networks

Generative Flow Networks (GFlowNets) aim to generate diverse trajectories from a distribution in which the final states of the trajectories are proportional to the reward, serving as a powerful alternative to reinforcement learning for exploratory control tasks. However, the individual-flow matching constraint in GFlowNets limits their applications for multi-agent systems, especially continuous joint-control problems. In this paper, we propose a novel Multi-Agent generative Continuous Flow Networks (MACFN) method to enable multiple agents to perform cooperative exploration for various compositional continuous objects. Technically, MACFN trains decentralized individual-flow-based policies in a centralized global-flow-based matching fashion. During centralized training, MACFN introduces a continuous flow decomposition network to deduce the flow contributions of each agent in the presence of only global rewards. Then agents can deliver actions solely based on their assigned local flow in a decentralized way, forming a joint policy distribution proportional to the rewards. To guarantee the expressiveness of continuous flow decomposition, we theoretically derive a consistency condition on the decomposition network. Experimental results demonstrate that the proposed method yields results superior to the state-of-the-art counterparts and better exploration capability. Our code is available at https://github.com/isluoshuang/MACFN.

cs.AI↗

arXiv2024

DPA-Net: Structured 3D Abstraction from Sparse Views via Differentiable Primitive Assembly

We present a differentiable rendering framework to learn structured 3D abstractions in the form of primitive assemblies from sparse RGB images capturing a 3D object. By leveraging differentiable volume rendering, our method does not require 3D supervision. Architecturally, our network follows the general pipeline of an image-conditioned neural radiance field (NeRF) exemplified by pixelNeRF for color prediction. As our core contribution, we introduce differential primitive assembly (DPA) into NeRF to output a 3D occupancy field in place of density prediction, where the predicted occupancies serve as opacity values for volume rendering. Our network, coined DPA-Net, produces a union of convexes, each as an intersection of convex quadric primitives, to approximate the target 3D object, subject to an abstraction loss and a masking loss, both defined in the image space upon volume rendering. With test-time adaptation and additional sampling and loss designs aimed at improving the accuracy and compactness of the obtained assemblies, our method demonstrates superior performance over state-of-the-art alternatives for 3D primitive abstraction from sparse views.

cs.CV↗

arXiv2024

Statistical AoI Guarantee Optimization for Supporting xURLLC in ISAC-enabled V2I Networks

This paper addresses the critical challenge of supporting next-generation ultra-reliable and low-latency communication (xURLLC) within integrated sensing and communication (ISAC)-enabled vehicle-to-infrastructure (V2I) networks. We incorporate channel evaluation and retransmission mechanisms for real-time reliability enhancement. Using stochastic network calculus (SNC), we establish a theoretical framework to derive upper bounds for the peak age of information violation probability (PAVP) via characterized sensing and communication moment generation functions (MGFs). By optimizing these bounds, we develop power allocation schemes that significantly reduce the statistical PAVP of sensory packets in such networks. Simulations validate our theoretical derivations and demonstrate the effectiveness of our proposed schemes.

cs.IT↗

arXiv2024

3D-UGCN: A Unified Graph Convolutional Network for Robust 3D Human Pose Estimation from Monocular RGB Images

Human pose estimation remains a multifaceted challenge in computer vision, pivotal across diverse domains such as behavior recognition, human-computer interaction, and pedestrian tracking. This paper proposes an improved method based on the spatial-temporal graph convolution net-work (UGCN) to address the issue of missing human posture skeleton sequences in single-view videos. We present the improved UGCN, which allows the network to process 3D human pose data and improves the 3D human pose skeleton sequence, thereby resolving the occlusion issue.

cs.CV↗

arXiv2024

Matching-Driven Deep Reinforcement Learning for Energy-Efficient Transmission Parameter Allocation in Multi-Gateway LoRa Networks

Long-range (LoRa) communication technology, distinguished by its low power consumption and long communication range, is widely used in the Internet of Things. Nevertheless, the LoRa MAC layer adopts pure ALOHA for medium access control, which may suffer from severe packet collisions as the network scale expands, consequently reducing the system energy efficiency (EE). To address this issue, it is critical to carefully allocate transmission parameters such as the channel (CH), transmission power (TP) and spreading factor (SF) to each end device (ED). Owing to the low duty cycle and sporadic traffic of LoRa networks, evaluating the system EE under various parameter settings proves to be time-consuming. Consequently, we propose an analytical model aimed at calculating the system EE while fully considering the impact of multiple gateways, duty cycling, quasi-orthogonal SFs and capture effects. On this basis, we investigate a joint CH, SF and TP allocation problem, with the objective of optimizing the system EE for uplink transmissions. Due to the NP-hard complexity of the problem, the optimization problem is decomposed into two subproblems: CH assignment and SF/TP assignment. First, a matching-based algorithm is introduced to address the CH assignment subproblem. Then, an attention-based multiagent reinforcement learning technique is employed to address the SF/TP assignment subproblem for EDs allocated to the same CH, which reduces the number of learning agents to achieve fast convergence. The simulation outcomes indicate that the proposed approach converges quickly under various parameter settings and obtains significantly better system EE than baseline algorithms.

cs.MA↗

arXiv2024

Hardware-Efficient Stabilization of Entanglement via Engineered Dissipation in Superconducting Circuits

Generation and preservation of quantum entanglement are among the primary tasks in quantum information processing. State stabilization via quantum bath engineering offers a resource-efficient approach to achieve this objective. However, current methods for engineering dissipative channels to stabilize target entangled states often require specialized hardware designs, complicating experimental realization and hindering their compatibility with scalable quantum computation architectures. In this work, we propose and experimentally demonstrate a stabilization protocol readily implementable in the mainstream integrated superconducting quantum circuits. The approach utilizes a Raman process involving a resonant (or nearly resonant) superconducting qubit array and their dedicated readout resonators to effectively emerge nonlocal dissipative channels. Leveraging individual controllability of the qubits and resonators, the protocol stabilizes two-qubit Bell states with a fidelity of $90.7\%$, marking the highest reported value in solid-state platforms to date. Furthermore, by extending this strategy to include three qubits, an entangled $W$ state is achieved with a fidelity of $86.2\%$, which has not been experimentally investigated before. Notably, the protocol is of practical interest since it only utilizes existing hardware common to standard operations in the underlying superconducting circuits, thereby facilitating the exploration of many-body quantum entanglement with dissipative resources.

quant-ph↗

arXiv2024

Submillimeter and Mid-Infrared Variability of Young Stellar Objects in the M17SWex Intermediate-Mass Star-Forming Region

We present a comprehensive analysis of young stellar object (YSO) variability within the M17 Southwest Extension (M17 SWex), using 3.5 years of monitoring data from the JCMT Transient Survey at sub-millimeter (sub-mm) and 9 years from the NEOWISE mission at mid-infrared (mid-IR). Our study encompasses observations of 147 bright sub-mm peaks identified within our deep JCMT co-added map as well as 156 YSOs in NEOWISE W1 and 179 in W2 that were previously identified in Spitzer surveys. We find three robust sub-mm variables: two are candidate YSOs and one is a likely extragalactic source. At mid-IR wavelengths, our analysis reveals secular and stochastic variability in 47 YSOs, with the highest fraction of secular variability occurring at the earliest evolutionary stage. This is similar to what has previously been observed for low-mass YSO variability within the Gould Belt. However, we observe less overall variability in M17SWex at both the sub-mm and mid-IR. We suspect that this lower fraction is due to the greater distance to M17 SWex. Our findings showcase the utility of multi-wavelength observations to better capture the complex variability phenomena inherent to star formation processes and demonstrate the importance of years-long monitoring of a diverse selection of star-forming environments.

astro-ph.GA↗

arXiv2024

Extraction of Weak Surface Diaphragmatic Electromyogram Using Modified Progressive FastICA Peel-Off

Diaphragmatic electromyogram (EMGdi) contains crucial information about human respiration therefore can be used to monitor respiratory condition. Although it is practical to record EMGdi noninvasively and conveniently by placing surface electrodes over chest skin, extraction of such weak surface EMGdi (sEMGdi) from great noisy environment is a challenging task, limiting its clinical use compared with esophageal EMGdi. In this paper, a novel method is presented for extracting weak sEMGdi signal from high-noise environment based on fast independent component analysis (FastICA), constrained FastICA and a peel-off strategy. It is truly a modified version of of progressive FastICA peel-off (PFP) framework, where the constrained FastICA helps to extract and refine respiration-related sEMGdi signals, while the peel-off strategy ensures the complete extraction of weaker sEMGdi components. The method was validated using both synthetic and clinical signals. It was demonstrated that our method was able to extract clean sEMGdi signals efficiently with little distortion. It outperformed state-of-the-art comparison methods in terms of sufficiently high SIR and CORR at all noise levels when tested on synthetic data, while also achieved an accuracy of 95.06% and a F2-score of 96.73% for breath identification on clinical data. The study presents a valuable solution for noninvasive extraction of sEMGdi signals, providing a convenient and valuable way of ventilator synchrony with a significant potential in aiding respiratory rehabilitation and health.

physics.med-ph↗

arXiv2024

Smooth plurisubharmonic exhaustion functions on pseudo-convex domains in infinite dimensions

In this paper, by modifying significantly the Friedrichs-Gross mollifier technique and/or using the Lasry-Lions regularization technique together with some carefully chosen cut-off functions, for the first time we construct explicitly smooth exhaustion functions on any open subset and smooth plurisubharmonic exhaustion functions on any pseudo-convex domain in a typical Hilbert space, which enjoy delicate properties needed for the $L^{2}$ method in infinite-dimensional complex analysis.

math.CV↗