arXiv subjects

Xiang Ji

Publications and source records attributed to Xiang Ji.

At least 55 records · Page 3Linked to original sources

arXiv2024

Velocity-Resolved Ionization Mapping of Broad Line Region. I. Insights into Diverse Geometry and Kinematics

Broad emission lines of active galactic nuclei (AGNs) originate from the broad-line region (BLR), consisting of dense gas clouds in orbit around an accreting supermassive black hole. Understanding the geometry and kinematics of the region is crucial for gaining insights into the physics and evolution of AGNs. Conventional velocity-resolved reverberation mapping may face challenges in disentangling the degeneracy between intricate motion and geometry of this region. To address this challenge, new key constraints are required. Here, we report the discovery of an asymmetric BLR using a novel technique: velocity-resolved ionization mapping, which can map the distance of emitting gas clouds by measuring Hydrogen line ratios at different velocities. By analyzing spectroscopic monitoring data, we find that the Balmer decrement is anticorrelated with the continuum and correlated with the lags across broad emission line velocities. Some line ratio profiles deviate from the expectations for a symmetrically virialized BLR, suggesting that the red-shifted and blue-shifted gas clouds may not be equidistant from the supermassive black hole (SMBH). This asymmetric geometry might represent a formation imprint, provide new perspectives on the evolution of AGNs, and influence SMBH mass measurements.

astro-ph.GA↗

arXiv2024

RS-NeRF: Neural Radiance Fields from Rolling Shutter Images

Neural Radiance Fields (NeRFs) have become increasingly popular because of their impressive ability for novel view synthesis. However, their effectiveness is hindered by the Rolling Shutter (RS) effects commonly found in most camera systems. To solve this, we present RS-NeRF, a method designed to synthesize normal images from novel views using input with RS distortions. This involves a physical model that replicates the image formation process under RS conditions and jointly optimizes NeRF parameters and camera extrinsic for each image row. We further address the inherent shortcomings of the basic RS-NeRF model by delving into the RS characteristics and developing algorithms to enhance its functionality. First, we impose a smoothness regularization to better estimate trajectories and improve the synthesis quality, in line with the camera movement prior. We also identify and address a fundamental flaw in the vanilla RS model by introducing a multi-sampling algorithm. This new approach improves the model's performance by comprehensively exploiting the RGB data across different rows for each intermediate camera pose. Through rigorous experimentation, we demonstrate that RS-NeRF surpasses previous methods in both synthetic and real-world scenarios, proving its ability to correct RS-related distortions effectively. Codes and data available: https://github.com/MyNiuuu/RS-NeRF

cs.CV↗

arXiv2024

Torchtree: flexible phylogenetic model development and inference using PyTorch

Bayesian inference has predominantly relied on the Markov chain Monte Carlo (MCMC) algorithm for many years. However, MCMC is computationally laborious, especially for complex phylogenetic models of time trees. This bottleneck has led to the search for alternatives, such as variational Bayes, which can scale better to large datasets. In this paper, we introduce torchtree, a framework written in Python that allows developers to easily implement rich phylogenetic models and algorithms using a fixed tree topology. One can either use automatic differentiation, or leverage torchtree's plug-in system to compute gradients analytically for model components for which automatic differentiation is slow. We demonstrate that the torchtree variational inference framework performs similarly to BEAST in terms of speed and approximation accuracy. Furthermore, we explore the use of the forward KL divergence as an optimizing criterion for variational inference, which can handle discontinuous and non-differentiable models. Our experiments show that inference using the forward KL divergence tends to be faster per iteration compared to the evidence lower bound (ELBO) criterion, although the ELBO-based inference may converge faster in some cases. Overall, torchtree provides a flexible and efficient framework for phylogenetic model development and inference using PyTorch.

q-bio.PE↗

arXiv2024

Self-Play with Adversarial Critic: Provable and Scalable Offline Alignment for Language Models

This work studies the challenge of aligning large language models (LLMs) with offline preference data. We focus on alignment by Reinforcement Learning from Human Feedback (RLHF) in particular. While popular preference optimization methods exhibit good empirical performance in practice, they are not theoretically guaranteed to converge to the optimal policy and can provably fail when the data coverage is sparse by classical offline reinforcement learning (RL) results. On the other hand, a recent line of work has focused on theoretically motivated preference optimization methods with provable guarantees, but these are not computationally efficient for large-scale applications like LLM alignment. To bridge this gap, we propose SPAC, a new offline preference optimization method with self-play, inspired by the on-average pessimism technique from the offline RL literature, to be the first provable and scalable approach to LLM alignment. We both provide theoretical analysis for its convergence under single-policy concentrability for the general function approximation setting and demonstrate its competitive empirical performance for LLM alignment on a 7B Mistral model with Open LLM Leaderboard evaluations.

cs.LG↗

arXiv2024

Remote-contact catalysis for target-diameter semiconducting carbon nanotube array

Electrostatic catalysis has been an exciting development in chemical synthesis (beyond enzymes catalysis) in recent years, boosting reaction rates and selectively producing certain reaction products. Most of the studies to date have been focused on using external electric field (EEF) to rearrange the charge distribution in small molecule reactions such as Diels-Alder addition, carbene reaction, etc. However, in order for these EEFs to be effective, a field on the order of 1 V/nm (10 MV/cm) is required, and the direction of the EEF has to be aligned with the reaction axis. Such a large and oriented EEF will be challenging for large-scale implementation, or materials growth with multiple reaction axis or steps. Here, we demonstrate that the energy band at the tip of an individual single-walled carbon nanotube (SWCNT) can be spontaneously shifted in a high-permittivity growth environment, with its other end in contact with a low-work function electrode (e.g., hafnium carbide or titanium carbide). By adjusting the Fermi level at a point where there is a substantial disparity in the density of states (DOS) between semiconducting (s-) and metallic (m-) SWCNTs, we achieve effective electrostatic catalysis for s-SWCNT growth assisted by a weak EEF perturbation (200V/cm). This approach enables the production of high-purity (99.92%) s-SWCNT horizontal arrays with narrow diameter distribution (0.95+-0.04 nm), targeting the requirement of advanced SWCNT-based electronics for future computing. These findings highlight the potential of electrostatic catalysis in precise materials growth, especially for s-SWCNTs, and pave the way for the development of advanced SWCNT-based electronics.

physics.app-ph↗

arXiv2024

Motion Blur Decomposition with Cross-shutter Guidance

Motion blur is a frequently observed image artifact, especially under insufficient illumination where exposure time has to be prolonged so as to collect more photons for a bright enough image. Rather than simply removing such blurring effects, recent researches have aimed at decomposing a blurry image into multiple sharp images with spatial and temporal coherence. Since motion blur decomposition itself is highly ambiguous, priors from neighbouring frames or human annotation are usually needed for motion disambiguation. In this paper, inspired by the complementary exposure characteristics of a global shutter (GS) camera and a rolling shutter (RS) camera, we propose to utilize the ordered scanline-wise delay in a rolling shutter image to robustify motion decomposition of a single blurry image. To evaluate this novel dual imaging setting, we construct a triaxial system to collect realistic data, as well as a deep network architecture that explicitly addresses temporal and contextual information through reciprocal branches for cross-shutter motion blur decomposition. Experiment results have verified the effectiveness of our proposed algorithm, as well as the validity of our dual imaging setting.

cs.CV↗

arXiv2024

Generalized bulk-boundary correspondence in periodically driven non-Hermitian systems

We present a pedagogical review of the periodically driven non-Hermitian systems, particularly on the rich interplay between the non-Hermitian skin effect and the topology. We start by reviewing the non-Bloch band theory of the static non-Hermitian systems and discuss the establishment of its generalized bulk-boundary correspondence. Ultimately, we focus on the non-Bloch band theory of two typical periodically driven non-Hermitian systems: harmonically driven non-Hermitian system and periodically quenched non-Hermitian system. The non-Bloch topological invariants were defined on the generalized Brillouin zone and the real space wave functions to characterize the Floquet non-Hermtian topological phases. Then, the generalized bulk-boundary correspondence was established for the two typical periodically driven non-Hermitian systems. Additionally, we review novel phenomena in the higher-dimensional periodically driven non-Hermitian systems, including Floquet non-Hermitian higher-order topological phases and Floquet hybrid skin-topological modes. The experimental realizations and recent advances have also been surveyed. Finally, we end with a summarization and hope this pedagogical review can motivate further research on Floquet non-Hermtian topological physics.

cond-mat.mes-hall↗

arXiv2024

Sample Complexity of Preference-Based Nonparametric Off-Policy Evaluation with Deep Networks

A recently popular approach to solving reinforcement learning is with data from human preferences. In fact, human preference data are now used with classic reinforcement learning algorithms such as actor-critic methods, which involve evaluating an intermediate policy over a reward learned from human preference data with distribution shift, known as off-policy evaluation (OPE). Such algorithm includes (i) learning reward function from human preference dataset, and (ii) learning expected cumulative reward of a target policy. Despite the huge empirical success, existing OPE methods with preference data often lack theoretical understanding and rely heavily on heuristics. In this paper, we study the sample efficiency of OPE with human preference and establish a statistical guarantee for it. Specifically, we approach OPE by learning the value function by fitted-Q-evaluation with a deep neural network. By appropriately selecting the size of a ReLU network, we show that one can leverage any low-dimensional manifold structure in the Markov decision process and obtain a sample-efficient estimator without suffering from the curse of high data ambient dimensionality. Under the assumption of high reward smoothness, our results \textit{almost align with the classical OPE results with observable reward data}. To the best of our knowledge, this is the first result that establishes a \textit{provably efficient} guarantee for off-policy evaluation with RLHF.

cs.LG↗

arXiv2024

Sample Complexity of Neural Policy Mirror Descent for Policy Optimization on Low-Dimensional Manifolds

Policy gradient methods equipped with deep neural networks have achieved great success in solving high-dimensional reinforcement learning (RL) problems. However, current analyses cannot explain why they are resistant to the curse of dimensionality. In this work, we study the sample complexity of the neural policy mirror descent (NPMD) algorithm with deep convolutional neural networks (CNN). Motivated by the empirical observation that many high-dimensional environments have state spaces possessing low-dimensional structures, such as those taking images as states, we consider the state space to be a $d$-dimensional manifold embedded in the $D$-dimensional Euclidean space with intrinsic dimension $d\ll D$. We show that in each iteration of NPMD, both the value function and the policy can be well approximated by CNNs. The approximation errors are controlled by the size of the networks, and the smoothness of the previous networks can be inherited. As a result, by properly choosing the network size and hyperparameters, NPMD can find an $ε$-optimal policy with $\widetilde{O}(ε^{-\frac{d}α-2})$ samples in expectation, where $α\in(0,1]$ indicates the smoothness of environment. Compared to previous work, our result exhibits that NPMD can leverage the low-dimensional structure of state space to escape from the curse of dimensionality, explaining the efficacy of deep policy gradient algorithms.

cs.LG↗

arXiv2023

Regret-Optimal Model-Free Reinforcement Learning for Discounted MDPs with Short Burn-In Time

A crucial problem in reinforcement learning is learning the optimal policy. We study this in tabular infinite-horizon discounted Markov decision processes under the online setting. The existing algorithms either fail to achieve regret optimality or have to incur a high memory and computational cost. In addition, existing optimal algorithms all require a long burn-in time in order to achieve optimal sample efficiency, i.e., their optimality is not guaranteed unless sample size surpasses a high threshold. We address both open problems by introducing a model-free algorithm that employs variance reduction and a novel technique that switches the execution policy in a slow-yet-adaptive manner. This is the first regret-optimal model-free algorithm in the discounted setting, with the additional benefit of a low burn-in time.

cs.LG↗

arXiv2023

Half-Precision Kronecker Product SVD Preconditioner for Structured Inverse Problems

In this paper we investigate the use of half-precision Kronecker product singular value decomposition (SVD) approximations as preconditioners for large-scale Tikhonov regularized least squares problems. Half precision reduces storage requirements and has the potential to greatly speedup computations on certain GPU architectures. We consider both standard PCG and flexible PCG algorithms, and investigate, through numerical experiments on image deblurring problems, the trade-offs between potentially faster convergence with the additional cost per iteration when using this preconditioning approach. Moreover, we also investigate the use of several regularization parameter choice methods, including generalized cross validation and the discrepancy principle.

math.NA↗

arXiv2023

Provable Benefits of Policy Learning from Human Preferences in Contextual Bandit Problems

For a real-world decision-making problem, the reward function often needs to be engineered or learned. A popular approach is to utilize human feedback to learn a reward function for training. The most straightforward way to do so is to ask humans to provide ratings for state-action pairs on an absolute scale and take these ratings as reward samples directly. Another popular way is to ask humans to rank a small set of state-action pairs by preference and learn a reward function from these preference data. Recently, preference-based methods have demonstrated substantial success in empirical applications such as InstructGPT. In this work, we develop a theoretical comparison between these human feedback approaches in offline contextual bandits and show how human bias and uncertainty in feedback modelings can affect the theoretical guarantees of these approaches. Through this, our results seek to provide a theoretical explanation for the empirical successes of preference-based methods from a modeling perspective.

cs.LG↗

arXiv2023

Towards Deep Learning Models Resistant to Transfer-based Adversarial Attacks via Data-centric Robust Learning

Transfer-based adversarial attacks raise a severe threat to real-world deep learning systems since they do not require access to target models. Adversarial training (AT), which is recognized as the strongest defense against white-box attacks, has also guaranteed high robustness to (black-box) transfer-based attacks. However, AT suffers from heavy computational overhead since it optimizes the adversarial examples during the whole training process. In this paper, we demonstrate that such heavy optimization is not necessary for AT against transfer-based attacks. Instead, a one-shot adversarial augmentation prior to training is sufficient, and we name this new defense paradigm Data-centric Robust Learning (DRL). Our experimental results show that DRL outperforms widely-used AT techniques (e.g., PGD-AT, TRADES, EAT, and FAT) in terms of black-box robustness and even surpasses the top-1 defense on RobustBench when combined with diverse data augmentations and loss regularizations. We also identify other benefits of DRL, for instance, the model generalization capability and robust fairness.

cs.CR↗

arXiv2023

Random-effects substitution models for phylogenetics via scalable gradient approximations

Phylogenetic and discrete-trait evolutionary inference depend heavily on an appropriate characterization of the underlying character substitution process. In this paper, we present random-effects substitution models that extend common continuous-time Markov chain models into a richer class of processes capable of capturing a wider variety of substitution dynamics. As these random-effects substitution models often require many more parameters than their usual counterparts, inference can be both statistically and computationally challenging. Thus, we also propose an efficient approach to compute an approximation to the gradient of the data likelihood with respect to all unknown substitution model parameters. We demonstrate that this approximate gradient enables scaling of sampling-based inference, namely Bayesian inference via Hamiltonian Monte Carlo, under random-effects substitution models across large trees and state-spaces. Applied to a dataset of 583 SARS-CoV-2 sequences, an HKY model with random-effects shows strong signals of nonreversibility in the substitution process, and posterior predictive model checks clearly show that it is a more adequate model than a reversible model. When analyzing the pattern of phylogeographic spread of 1441 influenza A virus (H3N2) sequences between 14 regions, a random-effects phylogeographic substitution model infers that air travel volume adequately predicts almost all dispersal rates. A random-effects state-dependent substitution model reveals no evidence for an effect of arboreality on the swimming mode in the tree frog subfamily Hylinae. Simulations reveal that random-effects substitution models can accommodate both negligible and radical departures from the underlying base substitution model. We show that our gradient-based inference approach is over an order of magnitude more time efficient than conventional approaches.

q-bio.PE↗

arXiv2023

Hard Adversarial Example Mining for Improving Robust Fairness

Adversarial training (AT) is widely considered the state-of-the-art technique for improving the robustness of deep neural networks (DNNs) against adversarial examples (AE). Nevertheless, recent studies have revealed that adversarially trained models are prone to unfairness problems, restricting their applicability. In this paper, we empirically observe that this limitation may be attributed to serious adversarial confidence overfitting, i.e., certain adversarial examples with overconfidence. To alleviate this problem, we propose HAM, a straightforward yet effective framework via adaptive Hard Adversarial example Mining.HAM concentrates on mining hard adversarial examples while discarding the easy ones in an adaptive fashion. Specifically, HAM identifies hard AEs in terms of their step sizes needed to cross the decision boundary when calculating loss value. Besides, an early-dropping mechanism is incorporated to discard the easy examples at the initial stages of AE generation, resulting in efficient AT. Extensive experimental results on CIFAR-10, SVHN, and Imagenette demonstrate that HAM achieves significant improvement in robust fairness while reducing computational cost compared to several state-of-the-art adversarial training methods. The code will be made publicly available.

cs.LG↗

arXiv2023

Strange Quasar Candidates with Abnormal Astrometric Characteristics from Gaia EDR3 and SDSS (SQUAB-II): Optical Identifications

There are some strange quasars with multiple Gaia detections or observed with abnormal astrometric characteristics, such as with large proper motions or significant astrometric noises. Those strange quasars could be potential candidates of quasar-star pairs, dual quasars (DQs), or lensed quasars (LQs). Searching for both DQs and LQs is of great importance in many fields of astrophysics. Here in this work, we select 143 SDSS spectroscopically confirmed quasars that have multiple Gaia EDR3 detections within 1 arcsec of the SDSS quasar' position. We apply several optical identification methods to classify this sample. We firstly exclude 65 quasar-star pairs via their stellar features including their parallaxes and proper motions, stellar features in the SDSS spectra, or via the colour-colour diagram. Based on the spectral-fitting results, we find 2 DQ candidates, one of which presents a double-peaked [O III] emission line feature and the other shows a broad $H_β$ velocity offset ($\sim$ 870 $ km s^{-1} $) relative to the [O III] $λ$5007 line. Via the colour difference method, we further find 56 LQ candidates with similar colours in their multiple images. We also cross-match 143 objects with the HST archive and find 19 targets with archival HST images. Our classification results of those 19 targets are mainly consistent with previous works.

astro-ph.GA↗

arXiv2023

Discovery of Five Green Pea Galaxies with Double-peaked Narrow [OIII] Lines

Although double-peaked narrow emission-line galaxies have been studied extensively in the past years, only a few are reported with the green pea galaxies (GPs). Here we present our discovery of five GPs with double-peaked narrow [OIII] emission lines, referred to as DPGPs, selected from the LAMOST and SDSS spectroscopic surveys. We find that these five DPGPs have blueshifted narrow components more prominent than the redshifted components, with velocity offsets of [OIII]$λ$5007 lines ranging from 306 to 518 $\rm km\, s^{-1}$ and full widths at half maximums (FWHMs) of individual components ranging from 263 to 441 $\rm km\, s^{-1}$. By analyzing the spectra and the spectral energy distributions (SEDs), we find that they have larger metallicities and stellar masses compared with other GPs. The H$α$ line width, emission-line diagnostic, mid-infrared color, radio emission, and SED fitting provide evidence of the AGN activities in these DPGPs. They have the same spectral properties of Type 2 quasars. Furthermore, we discuss the possible nature of the double-peaked narrow emission-line profiles of these DPGPs and find that they are more likely to be dual AGN. These DPGP galaxies are ideal laboratories for exploring the growth mode of AGN in the extremely luminous emission-line galaxies, the co-evolution between AGN and host galaxies, and the evolution of high-redshift galaxies in the early Universe.

astro-ph.GA↗

arXiv2023

Cross-modal Place Recognition in Image Databases using Event-based Sensors

Visual place recognition is an important problem towards global localization in many robotics tasks. One of the biggest challenges is that it may suffer from illumination or appearance changes in surrounding environments. Event cameras are interesting alternatives to frame-based sensors as their high dynamic range enables robust perception in difficult illumination conditions. However, current event-based place recognition methods only rely on event information, which restricts downstream applications of VPR. In this paper, we present the first cross-modal visual place recognition framework that is capable of retrieving regular images from a database given an event query. Our method demonstrates promising results with respect to the state-of-the-art frame-based and event-based methods on the Brisbane-Event-VPR dataset under different scenarios. We also verify the effectiveness of the combination of retrieval and classification, which can boost performance by a large margin.

cs.CV↗