Searcharxiv⌕ Search

arXiv subjects

Xin Liu

Publications and source records attributed to Xin Liu.

At least 649 records · Page 36Linked to original sources
arXiv2023

Recognizable Information Bottleneck

Information Bottlenecks (IBs) learn representations that generalize to unseen data by information compression. However, existing IBs are practically unable to guarantee generalization in real-world scenarios due to the vacuous generalization bound. The recent PAC-Bayes IB uses information complexity instead of information compression to establish a connection with the mutual information generalization bound. However, it requires the computation of expensive second-order curvature, which hinders its practical application. In this paper, we establish the connection between the recognizability of representations and the recent functional conditional mutual information (f-CMI) generalization bound, which is significantly easier to estimate. On this basis we propose a Recognizable Information Bottleneck (RIB) which regularizes the recognizability of representations through a recognizability critic optimized by density ratio matching under the Bregman divergence. Extensive experiments on several commonly used datasets demonstrate the effectiveness of the proposed method in regularizing the model and estimating the generalization gap.

cs.LG↗
arXiv2023

Axial correlation revivals and number factorization with structured random waves

We advance a general theory of field correlation revivals of structured random wave packets, composed of superpositions of propagation-invariant modes, at pairs of planes transverse to the packet propagation direction. We derive an elegant analytical relation between the normalized intensity autocorrelation function of thus structured paraxial light fields at a pair of points on an optical axis of the system and a Gauss sum, thereby establishing a fundamental link between statistical optics and number theory. We propose and experimentally implement a simple, robust analog random wave computer that can efficiently decompose numbers into prime factors.

physics.optics↗
arXiv2023

Sensitivity estimation for differentially private query processing

Differential privacy has become a popular privacy-preserving method in data analysis, query processing, and machine learning, which adds noise to the query result to avoid leaking privacy. Sensitivity, or the maximum impact of deleting or inserting a tuple on query results, determines the amount of noise added. Computing the sensitivity of some simple queries such as counting query is easy, however, computing the sensitivity of complex queries containing join operations is challenging. Global sensitivity of such a query is unboundedly large, which corrupts the accuracy of the query answer. Elastic sensitivity and residual sensitivity offer upper bounds of local sensitivity to reduce the noise, but they suffer from either low accuracy or high computational overhead. We propose two fast query sensitivity estimation methods based on sampling and sketch respectively, offering competitive accuracy and higher efficiency compared to the state-of-the-art methods.

cs.DB↗
arXiv2023

3D spectroscopy with GTC-MEGARA of the triple AGN candidate in SDSS J102700.40+174900.8

Triple AGN systems are expected to be the result of the hierarchical model of galaxy formation. Since there are very few of them confirmed as such, we present the results of a new study of the triple-AGN candidate SDSS J102700.40+174900.8 (center nucleus) through observations with $\it{GTC}$-$\it{MEGARA}$ Integral Field Unit. 1D and 2D analysis of the line ratios of the three nuclei allow us to locate them in the EW(H$α$) vs. [Nii] /H$α$ diagram. The central nucleus is found to be a retired galaxy (or fake AGN). The neighbors are found to be a strong AGN (southeastern nucleus, J102700.55+174900.2) compatible with a Sy2 galaxy, and a weak AGN (northern nucleus, J102700.38+174902.6) compatible with a LINER2. We find evidence that the neighbors constitute a dual AGN system (Sy2-LINER2) with a projected separation of 3.98 kpc in the optical bands. The H$α$ velocity map shows that the northern nucleus has an H$α$ emission with a velocity offset of $\sim$-500 km s$^{-1}$, whereas the southeastern nucleus has a rotating disk and H$α$ extended emission at kpc scales. Chandra archival data confirm that the neighbors have X-ray (0.5-2) keV and (2-7) keV emission, whereas the center nucleus shows no X-ray emission. A collisional ring with knots is observed in the HST images of the southeastern nucleus. These knots coincide with star formation regions that along with the ring are predicted in a head-on collision. In this case, the morphology changes are probably due to a minor merger that was produced by the passing of the northern through the southeastern nucleus.

astro-ph.GA↗
arXiv2023

Numerical Discreteness Errors in Multi-Species Cosmological N-body Simulations

We present a detailed analysis of numerical discreteness errors in two-species, gravity-only, cosmological simulations using the density power spectrum as a diagnostic probe. In a simple setup where both species are initialized with the same total matter transfer function, biased growth of power forms on small scales when the solver force resolution is finer than the mean interparticle separation. The artificial bias is more severe when individual density and velocity transfer functions are applied. In particular, significant large-scale offsets in power are measured between simulations with conventional offset grid initial conditions when compared against converged high-resolution results where the force resolution scale is matched to the interparticle separation. These offsets persist even when the cosmology is chosen so that the two particle species have the same mass, indicating that the error is sourced from discreteness in the total matter field as opposed to unequal particle mass. We further investigate two mitigation strategies to address discreteness errors: the frozen potential method and softened interspecies short-range forces. The former evolves particles under the approximately "frozen" total matter potential in linear theory at early times, while the latter filters cross-species gravitational interactions on small scales in low density regions. By modeling closer to the continuum limit, both mitigation strategies demonstrate considerable reductions in large-scale power spectrum offsets.

astro-ph.CO↗
arXiv2023

Derivation of a generalized quasi-geostrophic approximation for inviscid flows in a channel domain: The fast waves correction

This paper is devoted to investigating the rotating Boussinesq equations of inviscid, incompressible flows with both fast Rossby waves and fast internal gravity waves. The main objective is to establish a rigorous derivation and justification of a new generalized quasi-geostrophic approximation in a channel domain with no normal flow at the upper and lower solid boundaries, taking into account the resonance terms due to the fast and slow waves interactions. Under these circumstances, We are able to obtain uniform estimates and compactness without the requirement of either well-prepared initial data (as in [10]) or domain with no boundary (as in [17]). In particular, the nonlinear resonances and the new limit system, which takes into account the fast waves correction to the slow waves dynamics, are also identified without introducing Fourier series expansion. The key ingredient includes the introduction of (full) generalized potential vorticity.

math.AP↗
arXiv2023

Electrostatic effects of the MnBi2Te4-superconductor hetero-structures in chiral Majorana search

The realization of chiral Majorana modes is a challenging task. We aim to comprehend the phase diagrams and parameter control capabilities of the actual devices used in the chiral Majorana search. Beyond the well-known minimal models, we develop a numerical simulation scheme using a self-consistent Schrodinger-Poisson approach to study, as an example, the MnBi2Te4 thin film coupled to an s-wave superconductor. We show that both the superconducting proximity effect and the tunability of the chemical potential for the topological surface states are significantly influenced by the gate-induced electrostatic potential. This complicates the implementation in experiments, and the actual topological region will be narrowed in stark contrast to those predicted in the previous minimal models. Nevertheless, we demonstrate that the chiral Majorana mode still exists in a wide range of experimental parameters with practical tunability.

cond-mat.supr-con↗
arXiv2023

Incoherent mode division multiplexing for high-security information encryption

In the age of information explosion, the conventional optical communication protocols are rapidly reaching the limits of their capacity, as almost all available degrees of freedom (e.g., wavelength, polarization) for division multiplexing have been explored to date. Recent advances in coherent mode division multiplexing have greatly facilitated high-speed optical communications and secure, high-capacity information storage and transfer. However, coherent mode division multiplexing is quite vulnerable to even minute environmental disturbances which can cause significant information loss. Here, we propose and experimentally demonstrate a paradigm shift to incoherent mode division multiplexing for high-security optical information encryption by harnessing the degree of coherence of structured random light beams. In contrast to the conventional techniques, our approach does not require mode orthogonality to circumnavigate unwanted mode crosstalk. In addition, our protocol has, in principle, no upper bound on its capacity. Thanks to the extreme robustness of structured random light to external perturbations, we are able to achieve highly accurate information encryption and decryption in the adverse environment. The proposed protocol opens new horizons in an array of fields, such as optical communications and cryptography, and it can be relevant for information processing with acoustical, matter as well as other types of waves.

physics.optics↗
arXiv2023

Mode-pairing quantum key distribution with advantage distillation

Mode-pairing quantum key distribution (MP-QKD) is an easy-to-implement scheme that transcends the Pirandola--Laurenza--Ottaviani--Banchi bound without using quantum repeaters. In this paper, we present an improvement of the performance of MP-QKD using an advantage distillation method. The simulation results demonstrate that the proposed scheme extends the transmission distance significantly with a channel loss exceeding 7.6 dB. Moreover, the scheme tolerates a maximum quantum bit error rate of 8.9%, which is nearly twice that of the original MP-QKD. In particular, as the system misalignment error increases, the expandable distance of the proposed scheme also increases. The proposed system is expected to promote the practical implementation of MP-QKD in a wide range of applications, particularly in scenarios involving high channel losses and system errors.

quant-ph↗
arXiv2023

A close quasar pair in a disk-disk galaxy merger at z = 2.17

Galaxy mergers produce pairs of supermassive black holes (SMBHs), which may be witnessed as dual quasars if both SMBHs are rapidly accreting. The kiloparsec (kpc)-scale separation represents a physical regime sufficiently close for merger-induced effects to be important yet wide enough to be directly resolvable with the facilities currently available. Whereas many kpc-scale dual active galactic nuclei--the low-luminosity counterparts of quasars--have been observed in low-redshift mergers, no unambiguous dual quasar is known at cosmic noon ($z\approx2$), the peak of global star formation and quasar activity. Here we report multiwavelength observations of SDSS J0749+2255 as a kpc-scale, dual-quasar system hosted by a galaxy merger at cosmic noon ($z=2.17$). We discover extended host galaxies associated with the much brighter compact quasar nuclei (separated by 0.46" or 3.8 kpc) and low-surface-brightness tidal features as evidence for galactic interactions. Unlike its low-redshift and low-luminosity counterparts, SDSS J0749+2255 is hosted by massive compact disk-dominated galaxies. The apparent lack of stellar bulges and the fact that SDSS J0749+2255 already follows the local SMBH mass--host stellar mass relation, suggest that at least some SMBHs may have formed before their host stellar bulges. While still at kpc-scale separations where the host-galaxy gravitational potential dominates, the two SMBHs may evolve into a gravitationally bound binary system in around 0.22 Gyr.

astro-ph.CO↗
arXiv2023

A Partial Exact Penalty Function Approach for Constrained Optimization

In this paper, we focus on a class of constrained nonlinear optimization problems (NLP), where some of its equality constraints define a closed embedded submanifold $\mathcal{M}$ in $\mathbb{R}^n$. Although NLP can be solved directly by various existing approaches for constrained optimization in Euclidean space, these approaches usually fail to recognize the manifold structure of $\mathcal{M}$. To achieve better efficiency by utilizing the manifold structure of $\mathcal{M}$ in directly applying these existing optimization approaches, we propose a partial penalty function approach for NLP. In our proposed penalty function approach, we transform NLP into the corresponding constraint dissolving problem (CDP) in the Euclidean space, where the constraints that define $\mathcal{M}$ are eliminated through exact penalization. We establish the relationships on the constraint qualifications between NLP and CDP, and prove that NLP and CDP have the same stationary points and KKT points in a neighborhood of the feasible region under mild conditions. Therefore, various existing optimization approaches developed for constrained optimization in the Euclidean space can be directly applied to solve NLP through CDP. Preliminary numerical experiments demonstrate that by dissolving the constraints that define $\mathcal{M}$, CDP gains superior computational efficiency when compared to directly applying existing optimization approaches to solve NLP, especially in high dimensional scenarios.

math.OC↗
arXiv2023

Quantifying the Impact of Label Noise on Federated Learning

Federated Learning (FL) is a distributed machine learning paradigm where clients collaboratively train a model using their local (human-generated) datasets. While existing studies focus on FL algorithm development to tackle data heterogeneity across clients, the important issue of data quality (e.g., label noise) in FL is overlooked. This paper aims to fill this gap by providing a quantitative study on the impact of label noise on FL. We derive an upper bound for the generalization error that is linear in the clients' label noise level. Then we conduct experiments on MNIST and CIFAR-10 datasets using various FL algorithms. Our empirical results show that the global model accuracy linearly decreases as the noise level increases, which is consistent with our theoretical analysis. We further find that label noise slows down the convergence of FL training, and the global model tends to overfit when the noise level is high.

cs.LG↗
arXiv2023

Smoothing Gradient Tracking for Decentralized Optimization over the Stiefel Manifold with Non-smooth Regularizers

Recently, decentralized optimization over the Stiefel manifold has attacked tremendous attentions due to its wide range of applications in various fields. Existing methods rely on the gradients to update variables, which are not applicable to the objective functions with non-smooth regularizers, such as sparse PCA. In this paper, to the best of our knowledge, we propose the first decentralized algorithm for non-smooth optimization over Stiefel manifolds. Our algorithm approximates the non-smooth part of objective function by its Moreau envelope, and then existing algorithms for smooth optimization can be deployed. We establish the convergence guarantee with the iteration complexity of $\mathcal{O} (ε^{-4})$. Numerical experiments conducted under the decentralized setting demonstrate the effectiveness and efficiency of our algorithm.

math.OC↗
arXiv2023

Lifetime-based Optimization for Simulating Quantum Circuits on a New Sunway Supercomputer

High-performance classical simulator for quantum circuits, in particular the tensor network contraction algorithm, has become an important tool for the validation of noisy quantum computing. In order to address the memory limitations, the slicing technique is used to reduce the tensor dimensions, but it could also lead to additional computation overhead that greatly slows down the overall performance. This paper proposes novel lifetime-based methods to reduce the slicing overhead and improve the computing efficiency, including an interpretation method to deal with slicing overhead, an in-place slicing strategy to find the smallest slicing set and an adaptive tensor network contraction path refiner customized for Sunway architecture. Experiments show that in most cases the slicing overhead with our in-place slicing strategy would be less than the cotengra, which is the most used graph path optimization software at present. Finally, the resulting simulation time is reduced to 96.1s for the Sycamore quantum processor RQC, with a sustainable single-precision performance of 308.6Pflops using over 41M cores to generate 1M correlated samples, which is more than 5 times performance improvement compared to 60.4 Pflops in 2021 Gordon Bell Prize work.

cs.DC↗
arXiv2023

Single crystal synthesis and low-lying electronic structure of V$_3$S$_4$

We report successful growth of millimeter-sized high quality single crystals of V$_3$S$_4$, a candidate topological semimetal belonging to a low-symmetry space group and consisting of only low atomic number elements. Using density functional theory calculations and angle-resolved photoemission spectroscopy, we show that the nonmagnetic phase of monoclinic V$_3$S$_4$ hosts type-II Dirac-like quasiparticles which opens a sizable gap due to spin orbit coupling, as well as theoretical multiple nodal lines that are eliminated also by spin orbit coupling. These results suggest that relativistic effects give rise to profound modifications of the topological properties even in compounds with low-weight elements.

cond-mat.mtrl-sci↗
arXiv2023

MuxFlow: Efficient and Safe GPU Sharing in Large-Scale Production Deep Learning Clusters

Large-scale GPU clusters are widely-used to speed up both latency-critical (online) and best-effort (offline) deep learning (DL) workloads. However, most DL clusters either dedicate each GPU to one workload or share workloads in time, leading to very low GPU resource utilization. We present MuxFlow, the first production cluster system that supports efficient and safe space-sharing for DL workloads. NVIDIA MPS provides an opportunity to share multiple workloads in space on widely-deployed NVIDIA GPUs, but it cannot guarantee the performance and safety of online workloads. MuxFlow introduces a two-level protection mechanism for memory and computation to guarantee the performance of online workloads. Based on our practical error analysis, we design a mixed error-handling mechanism to guarantee the safety of online workloads. MuxFlow further proposes dynamic streaming multiprocessor (SM) allocation and matching-based scheduling to improve the efficiency of offline workloads. MuxFlow has been deployed at CompanyX's clusters with more than 20,000 GPUs. The deployment results indicate that MuxFlow substantially improves the GPU utilization from 26$\%$ to 76$\%$, SM activity from 16$\%$ to 33$\%$, and GPU memory from 42$\%$ to 48$\%$.

cs.DC↗
arXiv2023

PIAT: Parameter Interpolation based Adversarial Training for Image Classification

Adversarial training has been demonstrated to be the most effective approach to defend against adversarial attacks. However, existing adversarial training methods show apparent oscillations and overfitting issue in the training process, degrading the defense efficacy. In this work, we propose a novel framework, termed Parameter Interpolation based Adversarial Training (PIAT), that makes full use of the historical information during training. Specifically, at the end of each epoch, PIAT tunes the model parameters as the interpolation of the parameters of the previous and current epochs. Besides, we suggest to use the Normalized Mean Square Error (NMSE) to further improve the robustness by aligning the clean and adversarial examples. Compared with other regularization methods, NMSE focuses more on the relative magnitude of the logits rather than the absolute magnitude. Extensive experiments on several benchmark datasets and various networks show that our method could prominently improve the model robustness and reduce the generalization error. Moreover, our framework is general and could further boost the robust accuracy when combined with other adversarial training methods.

cs.CV↗
arXiv2023

Facial Affect Recognition based on Transformer Encoder and Audiovisual Fusion for the ABAW5 Challenge

In this paper, we present our solutions for the 5th Workshop and Competition on Affective Behavior Analysis in-the-wild (ABAW), which includes four sub-challenges of Valence-Arousal (VA) Estimation, Expression (Expr) Classification, Action Unit (AU) Detection and Emotional Reaction Intensity (ERI) Estimation. The 5th ABAW competition focuses on facial affect recognition utilizing different modalities and datasets. In our work, we extract powerful audio and visual features using a large number of sota models. These features are fused by Transformer Encoder and TEMMA. Besides, to avoid the possible impact of large dimensional differences between various features, we design an Affine Module to align different features to the same dimension. Extensive experiments demonstrate that the superiority of the proposed method. For the VA Estimation sub-challenge, our method obtains the mean Concordance Correlation Coefficient (CCC) of 0.6066. For the Expression Classification sub-challenge, the average F1 Score is 0.4055. For the AU Detection sub-challenge, the average F1 Score is 0.5296. For the Emotional Reaction Intensity Estimation sub-challenge, the average pearson's correlations coefficient on the validation set is 0.3968. All of the results of four sub-challenges outperform the baseline with a large margin.

cs.CV↗