arXiv subjects

Jun Zhang

Publications and source records attributed to Jun Zhang.

At least 541 records · Page 30Linked to original sources

arXiv2024

Draft & Verify: Lossless Large Language Model Acceleration via Self-Speculative Decoding

We present a novel inference scheme, self-speculative decoding, for accelerating Large Language Models (LLMs) without the need for an auxiliary model. This approach is characterized by a two-stage process: drafting and verification. The drafting stage generates draft tokens at a slightly lower quality but more quickly, which is achieved by selectively skipping certain intermediate layers during drafting. Subsequently, the verification stage employs the original LLM to validate those draft output tokens in one forward pass. This process ensures the final output remains identical to that produced by the unaltered LLM. Moreover, the proposed method requires no additional neural network training and no extra memory footprint, making it a plug-and-play and cost-effective solution for inference acceleration. Benchmarks with LLaMA-2 and its variants demonstrated a speedup up to 1.99$\times$.

cs.CL↗

arXiv2024

The Effect of Quantization in Federated Learning: A Rényi Differential Privacy Perspective

Federated Learning (FL) is an emerging paradigm that holds great promise for privacy-preserving machine learning using distributed data. To enhance privacy, FL can be combined with Differential Privacy (DP), which involves adding Gaussian noise to the model weights. However, FL faces a significant challenge in terms of large communication overhead when transmitting these model weights. To address this issue, quantization is commonly employed. Nevertheless, the presence of quantized Gaussian noise introduces complexities in understanding privacy protection. This research paper investigates the impact of quantization on privacy in FL systems. We examine the privacy guarantees of quantized Gaussian mechanisms using Rényi Differential Privacy (RDP). By deriving the privacy budget of quantized Gaussian mechanisms, we demonstrate that lower quantization bit levels provide improved privacy protection. To validate our theoretical findings, we employ Membership Inference Attacks (MIA), which gauge the accuracy of privacy leakage. The numerical results align with our theoretical analysis, confirming that quantization can indeed enhance privacy protection. This study not only enhances our understanding of the correlation between privacy and communication in FL but also underscores the advantages of quantization in preserving privacy.

cs.LG↗

arXiv2024

Low-Complexity Joint Azimuth-Range-Velocity Estimation for Integrated Sensing and Communication with OFDM Waveform

Integrated sensing and communication (ISAC) is a main application scenario of the sixth-generation mobile communication systems. Due to the fast-growing number of antennas and subcarriers in cellular systems, the computational complexity of joint azimuth-range-velocity estimation (JARVE) in ISAC systems is extremely high. This paper studies the JARVE problem for a monostatic ISAC system with orthogonal frequency division multiplexing (OFDM) waveform, in which a base station receives the echos of its transmitted cellular OFDM signals to sense multiple targets. The Cramer-Rao bounds are first derived for JARVE. A low-complexity algorithm is further designed for super-resolution JARVE, which utilizes the proposed iterative subspace update scheme and Levenberg-Marquardt optimization method to replace the exhaustive search of spatial spectrum in multiple-signal-classification (MUSIC) algorithm. Finally, with the practical parameters of 5G New Radio, simulation results verify that the proposed algorithm can reduce the computational complexity by three orders of magnitude and two orders of magnitude compared to the existing three-dimensional MUSIC algorithm and estimation-of-signal-parameters-using-rotational-invariance-techniques (ESPRIT) algorithm, respectively, and also improve the estimation performance.

cs.IT↗

arXiv2024

Tackling Distribution Shifts in Task-Oriented Communication with Information Bottleneck

Task-oriented communication aims to extract and transmit task-relevant information to significantly reduce the communication overhead and transmission latency. However, the unpredictable distribution shifts between training and test data, including domain shift and semantic shift, can dramatically undermine the system performance. In order to tackle these challenges, it is crucial to ensure that the encoded features can generalize to domain-shifted data and detect semanticshifted data, while remaining compact for transmission. In this paper, we propose a novel approach based on the information bottleneck (IB) principle and invariant risk minimization (IRM) framework. The proposed method aims to extract compact and informative features that possess high capability for effective domain-shift generalization and accurate semantic-shift detection without any knowledge of the test data during training. Specifically, we propose an invariant feature encoding approach based on the IB principle and IRM framework for domainshift generalization, which aims to find the causal relationship between the input data and task result by minimizing the complexity and domain dependence of the encoded feature. Furthermore, we enhance the task-oriented communication with the label-dependent feature encoding approach for semanticshift detection which achieves joint gains in IB optimization and detection performance. To avoid the intractable computation of the IB-based objective, we leverage variational approximation to derive a tractable upper bound for optimization. Extensive simulation results on image classification tasks demonstrate that the proposed scheme outperforms state-of-the-art approaches and achieves a better rate-distortion tradeoff.

eess.SP↗

arXiv2024

Benchmarking Classical and Learning-Based Multibeam Point Cloud Registration

Deep learning has shown promising results for multiple 3D point cloud registration datasets. However, in the underwater domain, most registration of multibeam echo-sounder (MBES) point cloud data are still performed using classical methods in the iterative closest point (ICP) family. In this work, we curate and release DotsonEast Dataset, a semi-synthetic MBES registration dataset constructed from an autonomous underwater vehicle in West Antarctica. Using this dataset, we systematically benchmark the performance of 2 classical and 4 learning-based methods. The experimental results show that the learning-based methods work well for coarse alignment, and are better at recovering rough transforms consistently at high overlap (20-50%). In comparison, GICP (a variant of ICP) performs well for fine alignment and is better across all metrics at extremely low overlap (10%). To the best of our knowledge, this is the first work to benchmark both learning-based and classical registration methods on an AUV-based MBES dataset. To facilitate future research, both the code and data are made available online.

cs.CV↗

arXiv2024

NeuRSS: Enhancing AUV Localization and Bathymetric Mapping with Neural Rendering for Sidescan SLAM

Implicit neural representations and neural rendering have gained increasing attention for bathymetry estimation from sidescan sonar (SSS). These methods incorporate multiple observations of the same place from SSS data to constrain the elevation estimate, converging to a globally-consistent bathymetric model. However, the quality and precision of the bathymetric estimate are limited by the positioning accuracy of the autonomous underwater vehicle (AUV) equipped with the sonar. The global positioning estimate of the AUV relying on dead reckoning (DR) has an unbounded error due to the absence of a geo-reference system like GPS underwater. To address this challenge, we propose in this letter a modern and scalable framework, NeuRSS, for SSS SLAM based on DR and loop closures (LCs) over large timescales, with an elevation prior provided by the bathymetric estimate using neural rendering from SSS. This framework is an iterative procedure that improves localization and bathymetric mapping. Initially, the bathymetry estimated from SSS using the DR estimate, though crude, can provide an important elevation prior in the nonlinear least-squares (NLS) optimization that estimates the relative pose between two loop-closure vertices in a pose graph. Subsequently, the global pose estimate from the SLAM component improves the positioning estimate of the vehicle, thus improving the bathymetry estimation. We validate our localization and mapping approach on two large surveys collected with a surface vessel and an AUV, respectively. We evaluate their localization results against the ground truth and compare the bathymetry estimation against data collected with multibeam echo sounders (MBES).

cs.RO↗

arXiv2024

Formation and Evolution of Transient Prominence Bubbles Driven by Erupting Mini-filaments

Prominence bubbles, the dark arch-shaped "voids" below quiescent prominences, are generally believed to be caused by the interaction between the prominences and the slowly-emerging or quasi-stable underlying magnetic loops. However, this scenario could not explain some short-lived bubbles with extremely dynamic properties of evolution. Based on high-resolution H$α$ observations, here we propose that bubbles should be classified into two categories according to their dynamic properties: quasi-steady (Type-I) bubbles and transient (Type-II) bubbles. Type-I bubbles could remain relatively stable and last for several hours, indicating the existence of a quasi-stable magnetic topology, while Type-II bubbles grow and collapse quickly within one hour without stability duration, which are usually associated with erupting mini-filaments. Analysis of several typical Type-II bubbles from different views, especially including an on-disk event, reveals that Type-II bubbles quickly appear and expand at a velocity of $\thicksim$5--25 km s$^{-1}$ accompanied by an erupting mini-filament below. The mini-filament's rising velocity is slightly larger than that of the Type-II bubbles' boundary, which will lead to the collision with each other in a short time, subsequent collapse of Type-II bubbles, and formation of a large plume into the above prominence. We also speculate that only if the angle between the axis of the erupting mini-filament and the line-of-sight is large enough, the interaction between the erupting mini-filament and the overlying prominence could trigger a Type-II bubble with a typical arch-shaped but quickly-expanding bright boundary.

astro-ph.SR↗

arXiv2024

Some Computational Results on Koszul-Vinberg Cochain Complexes

An affine connection is said to be flat if its curvature tensor vanishes identically. Koszul-Vinberg (KV for abbreviation) cohomology has been invoked to study the deformation theory of flat and torsion-free affine connections on tangent bundle. In this Note, we compute explicitly the differentials of various specific KV cochains, and study their relation to classical objects in information geometry, including deformations associated with projective and dual-projective transformations of a flat and torsion-free affine connection. As an application, we also give a simple yet non-trivial example of a KV algebra of which second cohomology group does not vanish.

math.DG↗

arXiv2024

A Hierarchical PSF Reconstruction Method

Reconstruction of the point spread function (PSF) plays an important role in many areas of astronomy, including photometry, astrometry, galaxy morphology, and shear measurement. The atmospheric and instrumental effects are the two main contributors to the PSF, both of which may exhibit complex spatial features. Current PSF reconstruction schemes typically rely on individual exposures, and its ability of reproducing the complicated features of the PSF distribution is therefore limited by the number of stars. Interestingly, in conventional methods, after stacking the model residuals of the PSF ellipticities and (relative) sizes from a large number of exposures, one can often observe some stable and nontrivial spatial patterns on the entire focal plane, which could be quite detrimental to, e.g., weak lensing measurements. These PSF residual patterns are caused by instrumental effects as they consistently appear in different exposures. Taking this as an advantage, we propose a multi-layer PSF reconstruction method to remove such PSF residuals, the second and third layers of which make use of all available exposures together. We test our method on the i-band data of the second release of Hyper Suprime-Cam. Our method successfully eliminates most of the PSF residuals. Using the Fourier\_Quad shear measurement method, we further test the performance of the resulting PSF fields on shear recovery using the field distortion effect. The PSF residuals have strong correlations with the shear residuals, and our new multi-layer PSF reconstruction method can remove most of such systematic errors related to PSF, leading to much smaller shear biases.

astro-ph.IM↗

arXiv2024

Beyond MMSE: Rank-1 Subspace Channel Estimator for Massive MIMO Systems

To glean the benefits offered by massive multi-input multi-output (MIMO) systems, channel state information must be accurately acquired. Despite the high accuracy, the computational complexity of classical linear minimum mean squared error (MMSE) estimator becomes prohibitively high in the context of massive MIMO, while the other low-complexity methods degrade the estimation accuracy seriously. In this paper, we develop a novel rank-1 subspace channel estimator to approximate the maximum likelihood (ML) estimator, which outperforms the linear MMSE estimator, but incurs a surprisingly low computational complexity. Our method first acquires the highly accurate angle-of-arrival (AoA) information via a constructed space-embedding matrix and the rank-1 subspace method. Then, it adopts the post-reception beamforming to acquire the unbiased estimate of channel gains. Furthermore, a fast method is designed to implement our new estimator. Theoretical analysis shows that the extra gain achieved by our method over the linear MMSE estimator grows according to the rule of O($\log_{10}M$), while its computational complexity is linearly scalable to the number of antennas $M$. Numerical simulations also validate the theoretical results. Our new method substantially extends the accuracy-complexity region and constitutes a promising channel estimation solution to the emerging massive MIMO communications.

cs.IT↗

arXiv2024

Task-Aware Encoder Control for Deep Video Compression

Prior research on deep video compression (DVC) for machine tasks typically necessitates training a unique codec for each specific task, mandating a dedicated decoder per task. In contrast, traditional video codecs employ a flexible encoder controller, enabling the adaptation of a single codec to different tasks through mechanisms like mode prediction. Drawing inspiration from this, we introduce an innovative encoder controller for deep video compression for machines. This controller features a mode prediction and a Group of Pictures (GoP) selection module. Our approach centralizes control at the encoding stage, allowing for adaptable encoder adjustments across different tasks, such as detection and tracking, while maintaining compatibility with a standard pre-trained DVC decoder. Empirical evidence demonstrates that our method is applicable across multiple tasks with various existing pre-trained DVCs. Moreover, extensive experiments demonstrate that our method outperforms previous DVC by about 25% bitrate for different tasks, with only one pre-trained decoder.

eess.IV↗

arXiv2024

Generating High-Precision Force Fields for Molecular Dynamics Simulations to Study Chemical Reaction Mechanisms using Molecular Configuration Transformer

Theoretical studies on chemical reaction mechanisms have been crucial in organic chemistry. Traditionally, calculating the manually constructed molecular conformations of transition states for chemical reactions using quantum chemical calculations is the most commonly used method. However, this way is heavily dependent on individual experience and chemical intuition. In our previous study, we proposed a research paradigm that uses enhanced sampling in molecular dynamics simulations to study chemical reactions. This approach can directly simulate the entire process of a chemical reaction. However, the computational speed limits the use of high-precision potential energy functions for simulations. To address this issue, we present a scheme for training high-precision force fields for molecular modeling using a previously developed graph-neural-network-based molecular model, molecular configuration transformer. This potential energy function allows for highly accurate simulations at a low computational cost, leading to more precise calculations of the mechanism of chemical reactions. We applied this approach to study a Claisen rearrangement reaction and a Carbonyl insertion reaction catalyzed by Manganese.

physics.chem-ph↗

arXiv2024

Floquet engineering Rydberg sub-THz frequency comb spectroscopy

Engineering a Terahertz (THz) frequency comb spectroscopy at atomic level advances the precisely measurement in spectroscopy and sensing. Current progresses on THz frequency comb rely on difference-frequency generation, optical parametric oscillation, and other methods. Generating a THz frequency comb poses challenges in source stability and achieving a narrow bandwidth, which traditional THz devices are difficult to achieve. Furthermore, accurately measuring the generated THz frequency comb necessitates a high-performance THz detector. Rydberg atoms are well-suited for electric field sensing due to their ultra-wide radio frequency transition energy levels, making them especially sensitive to external electric fields in the DC to THz bandwidth. However, there have been no reports about generating THz frequency comb spectroscopy at the atomic level until now. This work presents a THz frequency comb spectroscopy with Rydberg atoms, in which a Floquet comb-like transition is engineered through a time-periodic drive field. Our approach simplifies the setup required for THz frequency comb spectroscopy while extending the working bandwidth for Rydberg atomic sensors. The THz frequency comb spectroscopy at the atomic level reported in this article shows great potential for various applications in astronomy, remote sensing, spectral detection of biological samples, and other related fields.

physics.atom-ph↗

arXiv2024

Cavity-enhanced Rydberg atom microwave receiver

Developing microwave electric field sensing based on Rydberg atom has received significant attention due to its unique advantages. However, achieving effective coupling between Rydberg atom and the microwave electric field in the sensing process is a challenging problem that greatly impacts the sensitivity. To address this, we propose the use of a microwave resonant cavity to enhance the effective coupling between the Rydberg atoms and the microwave electric field. In our experiment, we use a three-photon excitation scheme to prepare Rydberg atoms, make measurements of electric fields without and with a microwave cavity in which the vapor cell is put inside. Through experimental testing, we achieve an 18 dB enhancement of power sensitivity. The experiment shows an effective enhancement in electric field pulse signal detection. This result provides a promising direction for enhancing the sensitivity of Rydberg atomic electric field sensors and paves the way for their application in precision electric field measurements.

physics.atom-ph↗

arXiv2024

Strong Arnold chord conjecture via normalized capacities

We show that every dynamically convex toric domain in $\mathbb R^4$ admits a $C^1$-neighborhood $\mathcal U$ within the space of star-shaped domains of $\mathbb R^4$ with the following property: for any $X \in \mathcal U$, every Legendrian knot in $\partial X$ admits a Reeb chord with distinct endpoints. A higher dimensional analog is also discussed.

math.SG↗

arXiv2024

MimiC: Combating Client Dropouts in Federated Learning by Mimicking Central Updates

Federated learning (FL) is a promising framework for privacy-preserving collaborative learning, where model training tasks are distributed to clients and only the model updates need to be collected at a server. However, when being deployed at mobile edge networks, clients may have unpredictable availability and drop out of the training process, which hinders the convergence of FL. This paper tackles such a critical challenge. Specifically, we first investigate the convergence of the classical FedAvg algorithm with arbitrary client dropouts. We find that with the common choice of a decaying learning rate, FedAvg oscillates around a stationary point of the global loss function, which is caused by the divergence between the aggregated and desired central update. Motivated by this new observation, we then design a novel training algorithm named MimiC, where the server modifies each received model update based on the previous ones. The proposed modification of the received model updates mimics the imaginary central update irrespective of dropout clients. The theoretical analysis of MimiC shows that divergence between the aggregated and central update diminishes with proper learning rates, leading to its convergence. Simulation results further demonstrate that MimiC maintains stable convergence performance and learns better models than the baseline methods.

cs.LG↗

arXiv2024

Data-Driven Online Resource Allocation for User Experience Improvement in Mobile Edge Clouds

As the cloud is pushed to the edge of the network, resource allocation for user experience improvement in mobile edge clouds (MEC) is increasingly important and faces multiple challenges. This paper studies quality of experience (QoE)-oriented resource allocation in MEC while considering user diversity, limited resources, and the complex relationship between allocated resources and user experience. We introduce a closed-loop online resource allocation (CORA) framework to tackle this problem. It learns the objective function of resource allocation from the historical dataset and updates the learned model using the online testing results. Due to the learned objective model is typically non-convex and challenging to solve in real-time, we leverage the Lyapunov optimization to decouple the long-term average constraint and apply the prime-dual method to solve this decoupled resource allocation problem. Thereafter, we put forth a data-driven optimal online queue resource allocation (OOQRA) algorithm and a data-driven robust OQRA (ROQRA) algorithm for homogenous and heterogeneous user cases, respectively. Moreover, we provide a rigorous convergence analysis for the OOQRA algorithm. We conduct extensive experiments to evaluate the proposed algorithms using the synthesis and YouTube datasets. Numerical results validate the theoretical analysis and demonstrate that the user complaint rate is reduced by up to 100% and 18% in the synthesis and YouTube datasets, respectively.

eess.SP↗

arXiv2024

Nonreciprocal interactions in crowd dynamics: investigating the impact of moving threats on pedestrian speed preferences

Nonreciprocal interaction crowd systems, such as human-human, human-vehicle, and human-robot systems, often have serious impacts on pedestrian safety and social order. A more comprehensive understanding of these systems is needed to optimize system stability and efficiency. Despite the importance of these interactions, empirical research in this area remains limited. Thus, in our study we explore this underresearched area, focusing on scenarios where nonreciprocity plays a critical role, such as mass stabbings, which pose a substantial risk to public safety. We conducted the first experiments on this system and analysed high-accuracy data obtained from these experiments. The extent of the direct threat zone is determined by the speed of the moving threat and the radius of danger occurrence. We further categorize potential threats into direct, adjacent, and rear-view zones, quantifying the level of threat for pedestrians. Our study revealed that a pedestrian's desired velocity correlated positively with potential threat intensity, increasing until near the direct threat zone. An emerging steady state is observed when escape routes are blocked by moving threats. This deviation affects the density-velocity relationship, making it distinct from the general relationship. This deviation signifies unique pedestrian behaviour in the presence of moving threats. Additionally, the rate of change in the angle for pedestrian motion in various desired directions is synchronized. This indicates the emergence of collective intelligence in nonreciprocal interaction crowd systems. As a result, our study may constitute a pioneering step towards understanding nonreciprocal interactions in crowd systems through laboratory experiments. These findings may enhance pedestrian safety and inform not only government crowd management strategies but also individual self-protection measures.

physics.soc-ph↗