arXiv subjects

Tao Wang

Publications and source records attributed to Tao Wang.

At least 397 records · Page 22Linked to original sources

arXiv2024

Looking From the Future: Multi-order Iterations Can Enhance Adversarial Attack Transferability

Various methods try to enhance adversarial transferability by improving the generalization from different perspectives. In this paper, we rethink the optimization process and propose a novel sequence optimization concept, which is named Looking From the Future (LFF). LFF makes use of the original optimization process to refine the very first local optimization choice. Adapting the LFF concept to the adversarial attack task, we further propose an LFF attack as well as an MLFF attack with better generalization ability. Furthermore, guiding with the LFF concept, we propose an $LLF^{\mathcal{N}}$ attack which entends the LFF attack to a multi-order attack, further enhancing the transfer attack ability. All our proposed methods can be directly applied to the iteration-based attack methods. We evaluate our proposed method on the ImageNet1k dataset by applying several SOTA adversarial attack methods under four kinds of tasks. Experimental results show that our proposed method can greatly enhance the attack transferability. Ablation experiments are also applied to verify the effectiveness of each component. The source code will be released after this paper is accepted.

cs.CV↗

arXiv2024

The size of the Milky Way galaxy

The size of a galaxy is one of the fundamental parameters that reflects its growth and assembly history. Traditionally, the size of the Milky Way has been characterized by the scale length of the disk, based on the assumption of an exponential density profile. Earlier scale length measurements suggest the Milky Way is an overly compact galaxy, compared to similar galaxies of its mass. These size measurements, however, ignore the presence of the bulge, and the assumption of a single-exponential disk profile faces growing challenges from the recent observations. The half-light radius is an alternative size measurement that is independent of the galaxy density profile and has been widely used to quantify the size of external galaxies. Here we report the half-light radius of the Milky Way, derived from a new measurement of the age-resolved Galactic surface brightness profile in an unprecedentedly wide radial range from ${\rm R=0}$ to 17~kpc. We find a broken surface brightness profile with a nearly flat distribution between 3.5 and 7.5 kpc, which results in a half-light radius of 5.75$\pm$0.38 kpc, significantly larger than the scale-length inferred from the canonical single-exponential disk profile but in good consistency with local disk galaxies of similar mass. Because our density profile can be decomposed by stellar age and extrapolated backwards in time, we can also confirm that the size history of the Milky Way is broadly consistent with high-redshift galaxies but with systematically smaller size at each look back time. Our results suggest that the Milky Way is a typical disk galaxy regarding its size and has likely experienced inefficient secular size growth.

astro-ph.GA↗

arXiv2024

Zero-Query Adversarial Attack on Black-box Automatic Speech Recognition Systems

In recent years, extensive research has been conducted on the vulnerability of ASR systems, revealing that black-box adversarial example attacks pose significant threats to real-world ASR systems. However, most existing black-box attacks rely on queries to the target ASRs, which is impractical when queries are not permitted. In this paper, we propose ZQ-Attack, a transfer-based adversarial attack on ASR systems in the zero-query black-box setting. Through a comprehensive review and categorization of modern ASR technologies, we first meticulously select surrogate ASRs of diverse types to generate adversarial examples. Following this, ZQ-Attack initializes the adversarial perturbation with a scaled target command audio, rendering it relatively imperceptible while maintaining effectiveness. Subsequently, to achieve high transferability of adversarial perturbations, we propose a sequential ensemble optimization algorithm, which iteratively optimizes the adversarial perturbation on each surrogate model, leveraging collaborative information from other models. We conduct extensive experiments to evaluate ZQ-Attack. In the over-the-line setting, ZQ-Attack achieves a 100% success rate of attack (SRoA) with an average signal-to-noise ratio (SNR) of 21.91dB on 4 online speech recognition services, and attains an average SRoA of 100% and SNR of 19.67dB on 16 open-source ASRs. For commercial intelligent voice control devices, ZQ-Attack also achieves a 100% SRoA with an average SNR of 15.77dB in the over-the-air setting.

cs.CR↗

arXiv2024

A Multi-Stage Goal-Driven Network for Pedestrian Trajectory Prediction

Pedestrian trajectory prediction plays a pivotal role in ensuring the safety and efficiency of various applications, including autonomous vehicles and traffic management systems. This paper proposes a novel method for pedestrian trajectory prediction, called multi-stage goal-driven network (MGNet). Diverging from prior approaches relying on stepwise recursive prediction and the singular forecasting of a long-term goal, MGNet directs trajectory generation by forecasting intermediate stage goals, thereby reducing prediction errors. The network comprises three main components: a conditional variational autoencoder (CVAE), an attention module, and a multi-stage goal evaluator. Trajectories are encoded using conditional variational autoencoders to acquire knowledge about the approximate distribution of pedestrians' future trajectories, and combined with an attention mechanism to capture the temporal dependency between trajectory sequences. The pivotal module is the multi-stage goal evaluator, which utilizes the encoded feature vectors to predict intermediate goals, effectively minimizing cumulative errors in the recursive inference process. The effectiveness of MGNet is demonstrated through comprehensive experiments on the JAAD and PIE datasets. Comparative evaluations against state-of-the-art algorithms reveal significant performance improvements achieved by our proposed method.

cs.CV↗

arXiv2024

The fundamental plane of black hole activity for low-luminosity radio active galactic nuclei across 1 < z < 4

The fundamental plane of black hole activity (BHFP) describes the correlation between radio luminosity ($L_R$), X-ray luminosity ($L_X$), and black hole mass. It reflects a disc-jet connection. However, dependence of BHFP on various physical properties of active galactic nuclei (AGNs) and host galaxies is unclear, especially for low-luminosity AGNs, which is important for understanding accretion physics in AGNs. Here we explore the dependence of BHFP on radio loudness, Eddington ratio ($λ_E$), redshift, and galaxy star formation properties at 0.1 < z < 4 for radio AGNs. Based on current deep and large surveys, our studies extend to low luminosities and high redshifts. From GOODS and COSMOS fields, we constructed a homogeneous radio AGN sample including 208 objects. We divided the radio AGN sample into 141 radio-quiet (RQ) AGNs and 67 radio-loud (RL) AGNs according to the radio loudness defined by the $L_R$-to-$L_X$ ratio. This ratio shows a bimodal distribution that is well described by two Gaussian models. Their cross point corresponds to a radio loudness threshold of $\log (L_R/L_X) = -2.73$. RQ and RL AGNs show different BHFPs, indicating a dependence of BHFP on the radio loudness. For both RQ and RL AGNs, BHFP shows a dependence on $λ_E$, but no dependence on redshift. BHFP shows a dependence on the galaxy star formation properties for RQ AGNs, while for RL AGNs this dependence disappears. BHFP sheds important light on the accretion physics and X-ray emission origins. X-ray emission of RQ AGNs originates from a combination of advection-dominated accretion flow (ADAF) and synchrotron radiation from jet at $0.01 < λ_E < 0.1$, and from jet at $0.1 < λ_E < 1$. X-ray emission in RL AGNs originates from a combination of ADAF and jet at $λ_E < 0.01$, from jet at $0.01 < λ_E < 0.1$, and from a combination of standard thin disc and jet at $λ_E > 0.1$.

astro-ph.GA↗

arXiv2024

The wave-like disk oscillations of mono-age stellar populations in the Solar neighbourhood from Gaia DR3

The North-South asymmetry in the number density and bulk velocity of stars in the solar neighborhood provides valuable insights into the formation and evolution of the Milky Way disk. Our objective is to investigate the wave-like disk oscillations of mono-age stellar populations in the Solar neighbourhood using data from Gaia Data Release 3. We have selected a comprehensive sample of main sequence turn off stars. The ages of these stars can be accurately determined using isochrone fitting methods. Our findings indicate that the north-south density and mean vertical velocity asymmetries remain consistent across all age groups.The uniformity of perturbations across all subsamples suggests that all populations are responding to the same external influence, which likely affects them irrespective of their age. Moreover, the fact that these perturbations appear consistently implies they could be either ongoing or recent. Regarding vertical velocity dispersions, we observe that older stars exhibit larger dispersions.

astro-ph.GA↗

arXiv2024

GridFormer: Residual Dense Transformer with Grid Structure for Image Restoration in Adverse Weather Conditions

Image restoration in adverse weather conditions is a difficult task in computer vision. In this paper, we propose a novel transformer-based framework called GridFormer which serves as a backbone for image restoration under adverse weather conditions. GridFormer is designed in a grid structure using a residual dense transformer block, and it introduces two core designs. First, it uses an enhanced attention mechanism in the transformer layer. The mechanism includes stages of the sampler and compact self-attention to improve efficiency, and a local enhancement stage to strengthen local information. Second, we introduce a residual dense transformer block (RDTB) as the final GridFormer layer. This design further improves the network's ability to learn effective features from both preceding and current local features. The GridFormer framework achieves state-of-the-art results on five diverse image restoration tasks in adverse weather conditions, including image deraining, dehazing, deraining \& dehazing, desnowing, and multi-weather restoration. The source code and pre-trained models are available at https://github.com/TaoWangzj/GridFormer.

cs.CV↗

arXiv2024

ALMA Lensing Cluster Survey: Deep 1.2 mm Number Counts and Infrared Luminosity Functions at $z\simeq1-8$

We present a statistical study of 180 dust continuum sources identified in 33 massive cluster fields by the ALMA Lensing Cluster Survey (ALCS) over a total of 133 arcmin$^{2}$ area, homogeneously observed at 1.2 mm. ALCS enables us to detect extremely faint mm sources by lensing magnification, including near-infrared (NIR) dark objects showing no counterparts in existing {\it Hubble Space Telescope} and {\it Spitzer} images. The dust continuum sources belong to a blind sample ($N=141$) with S/N $\gtrsim$ 5.0 (a purity of $>$ 0.99) or a secondary sample ($N=39$) with S/N= $4.0-5.0$ screened by priors. With the blind sample, we securely derive 1.2-mm number counts down to $\sim7$ $μ$Jy, and find that the total integrated 1.2mm flux is 20.7$^{+8.5}_{-6.5}$ Jy deg$^{-2}$, resolving $\simeq$ 80 % of the cosmic infrared background light. The resolved fraction varies by a factor of $0.6-1.1$ due to the completeness correction depending on the spatial size of the mm emission. We also derive infrared (IR) luminosity functions (LFs) at $z=0.6-7.5$ with the $1/V_{\rm max}$ method, finding the redshift evolution of IR LFs characterized by positive luminosity and negative density evolution. The total (=UV+IR) cosmic star-formation rate density (SFRD) at $z>4$ is estimated to be $161^{+25}_{-21}$ % of the established measurements, which were almost exclusively based on optical$-$NIR surveys. Although our general understanding of the cosmic SFRD is unlikely to change beyond a factor of 2, these results add to the weight of evidence for an additional ($\approx 60$ %) SFRD component contributed by the faint-mm population, including NIR dark objects.

astro-ph.GA↗

arXiv2024

Achieving Cooling Without Repump Lasers Through Ion Motional Heating

Laser cooling typically requires one or more repump lasers to clear dark states and enable recycling transitions. Here, we have achieved cooling of Be+ ions using a single laser beam, facilitated by one-dimensional heating through micromotion. By manipulating the displacement from the trap's nodal line, we precisely controlled the ion micromotion direction and speed, reaching up to 3144 m/s, which corresponds to a 7.1 GHz Doppler frequency shift in our experiment. This approach eliminates the necessity of a 1.25 GHz offset repump laser while keeping the Be+ ions cold in the perpendicular direction. Measurements were taken using cooling laser detuning and imaging of ion trajectories. Molecular dynamics simulations, based on machine learned time-dependent electric field E(X, Y, Z, t) inside the trap, accurately reproduced the experimental observation, illuminating the relationship between the direction of micromotion and the trapping electric filed vector. This work not only provides a robust method for managing the micromotion velocity of ions but also sheds light on laser cooling complex systems that require multiple repumping lasers. Additionally, it offers a method for controlling energy in the context of ion-molecule collision investigations.

physics.atom-ph↗

arXiv2024

Understanding the Difficulty of Solving Cauchy Problems with PINNs

Physics-Informed Neural Networks (PINNs) have gained popularity in scientific computing in recent years. However, they often fail to achieve the same level of accuracy as classical methods in solving differential equations. In this paper, we identify two sources of this issue in the case of Cauchy problems: the use of $L^2$ residuals as objective functions and the approximation gap of neural networks. We show that minimizing the sum of $L^2$ residual and initial condition error is not sufficient to guarantee the true solution, as this loss function does not capture the underlying dynamics. Additionally, neural networks are not capable of capturing singularities in the solutions due to the non-compactness of their image sets. This, in turn, influences the existence of global minima and the regularity of the network. We demonstrate that when the global minimum does not exist, machine precision becomes the predominant source of achievable error in practice. We also present numerical experiments in support of our theoretical claims.

cs.LG↗

arXiv2024

State-Compute Replication: Parallelizing High-Speed Stateful Packet Processing

With the slowdown of Moore's law, CPU-oriented packet processing in software will be significantly outpaced by emerging line speeds of network interface cards (NICs). Single-core packet-processing throughput has saturated. We consider the problem of high-speed packet processing with multiple CPU cores. The key challenge is state--memory that multiple packets must read and update. The prevailing method to scale throughput with multiple cores involves state sharding, processing all packets that update the same state, i.e., flow, at the same core. However, given the heavy-tailed nature of realistic flow size distributions, this method will be untenable in the near future, since total throughput is severely limited by single core performance. This paper introduces state-compute replication, a principle to scale the throughput of a single stateful flow across multiple cores using replication. Our design leverages a packet history sequencer running on a NIC or top-of-the-rack switch to enable multiple cores to update state without explicit synchronization. Our experiments with realistic data center and wide-area Internet traces shows that state-compute replication can scale total packet-processing throughput linearly with cores, deterministically and independent of flow size distributions, across a range of realistic packet-processing programs.

cs.NI↗

arXiv2024

MINT: a Multi-modal Image and Narrative Text Dubbing Dataset for Foley Audio Content Planning and Generation

Foley audio, critical for enhancing the immersive experience in multimedia content, faces significant challenges in the AI-generated content (AIGC) landscape. Despite advancements in AIGC technologies for text and image generation, the foley audio dubbing remains rudimentary due to difficulties in cross-modal scene matching and content correlation. Current text-to-audio technology, which relies on detailed and acoustically relevant textual descriptions, falls short in practical video dubbing applications. Existing datasets like AudioSet, AudioCaps, Clotho, Sound-of-Story, and WavCaps do not fully meet the requirements for real-world foley audio dubbing task. To address this, we introduce the Multi-modal Image and Narrative Text Dubbing Dataset (MINT), designed to enhance mainstream dubbing tasks such as literary story audiobooks dubbing, image/silent video dubbing. Besides, to address the limitations of existing TTA technology in understanding and planning complex prompts, a Foley Audio Content Planning, Generation, and Alignment (CPGA) framework is proposed, which includes a content planning module leveraging large language models for complex multi-modal prompts comprehension. Additionally, the training process is optimized using Proximal Policy Optimization based reinforcement learning, significantly improving the alignment and auditory realism of generated foley audio. Experimental results demonstrate that our approach significantly advances the field of foley audio dubbing, providing robust solutions for the challenges of multi-modal dubbing. Even when utilizing the relatively lightweight GPT-2 model, our framework outperforms open-source multimodal large models such as LLaVA, DeepSeek-VL, and Moondream2. The dataset is available at https://github.com/borisfrb/MINT .

eess.AS↗

arXiv2024

Existence and uniqueness of ground state solutions for the planar Schrödinger-Newton equation on the disc

This paper is concerned with the existence and qualitative properties of positive ground state solutions for the planar Schrödinger-Newton equation on the disc. First, we prove the existence and radial symmetry of all the positive ground state solutions by employing the symmetric decreasing rearrangement and Talenti's inequality. Next, we develop Newton's theorem and then use the contraction mapping principle to establish the uniqueness of the positive ground state solution for the Schrödinger-Newton equation on the disc in the two dimensional case. Finally, we show that the unique positive ground state solution converges to the trivial solution as the radius $R$ tending to infinity, which is totally different from the higher dimensional case in \cite{Guo-Wang-Yi}.

math.AP↗

arXiv2024

PPPR: Portable Plug-in Prompt Refiner for Text to Audio Generation

Text-to-Audio (TTA) aims to generate audio that corresponds to the given text description, playing a crucial role in media production. The text descriptions in TTA datasets lack rich variations and diversity, resulting in a drop in TTA model performance when faced with complex text. To address this issue, we propose a method called Portable Plug-in Prompt Refiner, which utilizes rich knowledge about textual descriptions inherent in large language models to effectively enhance the robustness of TTA acoustic models without altering the acoustic training set. Furthermore, a Chain-of-Thought that mimics human verification is introduced to enhance the accuracy of audio descriptions, thereby improving the accuracy of generated content in practical applications. The experiments show that our method achieves a state-of-the-art Inception Score (IS) of 8.72, surpassing AudioGen, AudioLDM and Tango.

cs.SD↗

arXiv2024

Controlled Decoding from Language Models

KL-regularized reinforcement learning (RL) is a popular alignment framework to control the language model responses towards high reward outcomes. We pose a tokenwise RL objective and propose a modular solver for it, called controlled decoding (CD). CD exerts control through a separate prefix scorer module, which is trained to learn a value function for the reward. The prefix scorer is used at inference time to control the generation from a frozen base model, provably sampling from a solution to the RL objective. We empirically demonstrate that CD is effective as a control mechanism on popular benchmarks. We also show that prefix scorers for multiple rewards may be combined at inference time, effectively solving a multi-objective RL problem with no additional training. We show that the benefits of applying CD transfer to an unseen base model with no further tuning as well. Finally, we show that CD can be applied in a blockwise decoding fashion at inference-time, essentially bridging the gap between the popular best-of-K strategy and tokenwise control through reinforcement learning. This makes CD a promising approach for alignment of language models.

cs.LG↗

arXiv2024

JWST's first glimpse of a z > 2 forming cluster reveals a top-heavy stellar mass function

Clusters and their progenitors (protoclusters) at z = 2-4, the peak epoch of star formation, are ideal laboratories to study the formation process of both the clusters themselves and their member galaxies. However, a complete census of their member galaxies has been challenging due to observational difficulties. Here we present new JWST/NIRCam observations targeting the distant cluster CLJ1001 at z = 2.51 from the COSMOS-Web program, which, in combination with previous narrowband imaging targeting H-alpha emitters and deep millimeter surveys of CO emitters, provide a complete view of massive galaxy assembly in CLJ1001. In particular, JWST reveals a population of massive, extremely red cluster members in the long-wavelength bands that were invisible in previous Hubble Space Telescope (HST)/F160W imaging (HST-dark members). Based on this highly complete spectroscopic sample of member galaxies, we show that the spatial distribution of galaxies in CLJ1001 exhibits a strong central concentration, with the central galaxy density already resembling that of low-z clusters. Moreover, we reveal a "top-heavy" stellar mass function for the star-forming galaxies (SFGs), with an overabundance of massive SFGs piled up in the cluster core. These features strongly suggest that CLJ1001 is caught in a rapid transition, with many of its massive SFGs likely soon becoming quiescent. In the context of cluster formation, these findings suggest that the earliest clusters form from the inside out and top to bottom, with the massive galaxies in the core assembling first, followed by the less massive ones in the outskirts.

astro-ph.GA↗

arXiv2024

Mollification Effects of Policy Gradient Methods

Policy gradient methods have enabled deep reinforcement learning (RL) to approach challenging continuous control problems, even when the underlying systems involve highly nonlinear dynamics that generate complex non-smooth optimization landscapes. We develop a rigorous framework for understanding how policy gradient methods mollify non-smooth optimization landscapes to enable effective policy search, as well as the downside of it: while making the objective function smoother and easier to optimize, the stochastic objective deviates further from the original problem. We demonstrate the equivalence between policy gradient methods and solving backward heat equations. Following the ill-posedness of backward heat equations from PDE theory, we present a fundamental challenge to the use of policy gradient under stochasticity. Moreover, we make the connection between this limitation and the uncertainty principle in harmonic analysis to understand the effects of exploration with stochastic policies in RL. We also provide experimental results to illustrate both the positive and negative aspects of mollification effects in practice.

cs.LG↗

arXiv2024

Unlocking Data-free Low-bit Quantization with Matrix Decomposition for KV Cache Compression

Key-value~(KV) caching is an important technique to accelerate the inference of large language models~(LLMs), but incurs significant memory overhead. To compress the size of KV cache, existing methods often compromise precision or require extra data for calibration, limiting their practicality in LLM deployment. In this paper, we introduce \textbf{DecoQuant}, a novel data-free low-bit quantization technique based on tensor decomposition methods, to effectively compress KV cache. Our core idea is to adjust the outlier distribution of the original matrix by performing tensor decomposition, so that the quantization difficulties are migrated from the matrix to decomposed local tensors. Specially, we find that outliers mainly concentrate on small local tensors, while large tensors tend to have a narrower value range. Based on this finding, we propose to apply low-bit quantization to the large tensor, while maintaining high-precision representation for the small tensor. Furthermore, we utilize the proposed quantization method to compress the KV cache of LLMs to accelerate the inference and develop an efficient dequantization kernel tailored specifically for DecoQuant. Through extensive experiments, DecoQuant demonstrates remarkable efficiency gains, showcasing up to a $\sim$75\% reduction in memory footprint while maintaining comparable generation quality.

cs.CL↗