arXiv subjects

Xintong Wang

Publications and source records attributed to Xintong Wang.

52 records · Page 3Linked to original sources

arXiv2024

Probing Large Language Models from A Human Behavioral Perspective

Large Language Models (LLMs) have emerged as dominant foundational models in modern NLP. However, the understanding of their prediction processes and internal mechanisms, such as feed-forward networks (FFN) and multi-head self-attention (MHSA), remains largely unexplored. In this work, we probe LLMs from a human behavioral perspective, correlating values from LLMs with eye-tracking measures, which are widely recognized as meaningful indicators of human reading patterns. Our findings reveal that LLMs exhibit a similar prediction pattern with humans but distinct from that of Shallow Language Models (SLMs). Moreover, with the escalation of LLM layers from the middle layers, the correlation coefficients also increase in FFN and MHSA, indicating that the logits within FFN increasingly encapsulate word semantics suitable for predicting tokens from the vocabulary.

cs.CL↗

arXiv2024

AnimatableDreamer: Text-Guided Non-rigid 3D Model Generation and Reconstruction with Canonical Score Distillation

Advances in 3D generation have facilitated sequential 3D model generation (a.k.a 4D generation), yet its application for animatable objects with large motion remains scarce. Our work proposes AnimatableDreamer, a text-to-4D generation framework capable of generating diverse categories of non-rigid objects on skeletons extracted from a monocular video. At its core, AnimatableDreamer is equipped with our novel optimization design dubbed Canonical Score Distillation (CSD), which lifts 2D diffusion for temporal consistent 4D generation. CSD, designed from a score gradient perspective, generates a canonical model with warp-robustness across different articulations. Notably, it also enhances the authenticity of bones and skinning by integrating inductive priors from a diffusion model. Furthermore, with multi-view distillation, CSD infers invisible regions, thereby improving the fidelity of monocular non-rigid reconstruction. Extensive experiments demonstrate the capability of our method in generating high-flexibility text-guided 3D models from the monocular video, while also showing improved reconstruction performance over existing non-rigid reconstruction methods.

cs.CV↗

arXiv2023

MHDnet: Physics-preserving learning for solving magnetohydrodynamics problems

Designing efficient and high-accuracy numerical methods for complex dynamic incompressible magnetohydrodynamics (MHD) equations remains a challenging problem in various analysis and design tasks. This is mainly due to the nonlinear coupling of the magnetic and velocity fields occurring with convection and Lorentz forces, and multiple physical constraints, which will lead to the limitations of numerical computation. In this paper, we develop the MHDnet as a physics-preserving learning approach to solve MHD problems, where three different mathematical formulations are considered and named $B$ formulation, $A_1$ formulation, and $A_2$ formulation. Then the formulations are embedded into the MHDnet that can preserve the underlying physical properties and divergence-free condition. Moreover, MHDnet is designed by the multi-modes feature merging with multiscale neural network architecture, which can accelerate the convergence of the neural networks (NN) by alleviating the interaction of magnetic fluid coupling across different frequency modes. Furthermore, the pressure fields of three formulations, as the hidden state, can be obtained without extra data and computational cost. Several numerical experiments are presented to demonstrate the performance of the proposed MHDnet compared with different NN architectures and numerical formulations.

math.NA↗

arXiv2023

Efficient Bayesian inference using physics-informed invertible neural networks for inverse problems

In this paper, we introduce an innovative approach for addressing Bayesian inverse problems through the utilization of physics-informed invertible neural networks (PI-INN). The PI-INN framework encompasses two sub-networks: an invertible neural network (INN) and a neural basis network (NB-Net). The primary role of the NB-Net lies in modeling the spatial basis functions characterizing the solution to the forward problem dictated by the underlying partial differential equation. Simultaneously, the INN is designed to partition the parameter vector linked to the input physical field into two distinct components: the expansion coefficients representing the forward problem solution and the Gaussian latent noise. If the forward mapping is precisely estimated, and the statistical independence between expansion coefficients and latent noise is well-maintained, the PI-INN offers a precise and efficient generative model for Bayesian inverse problems, yielding tractable posterior density estimates. As a particular physics-informed deep learning model, the primary training challenge for PI-INN centers on enforcing the independence constraint, which we tackle by introducing a novel independence loss based on estimated density. We support the efficacy and precision of the proposed PI-INN through a series of numerical experiments, including inverse kinematics, 1-dimensional and 2-dimensional diffusion equations, and seismic traveltime tomography. Specifically, our experimental results showcase the superior performance of the proposed independence loss in comparison to the commonly used but computationally demanding kernel-based maximum mean discrepancy loss.

math.NA↗

arXiv2023

CrossSinger: A Cross-Lingual Multi-Singer High-Fidelity Singing Voice Synthesizer Trained on Monolingual Singers

It is challenging to build a multi-singer high-fidelity singing voice synthesis system with cross-lingual ability by only using monolingual singers in the training stage. In this paper, we propose CrossSinger, which is a cross-lingual singing voice synthesizer based on Xiaoicesing2. Specifically, we utilize International Phonetic Alphabet to unify the representation for all languages of the training data. Moreover, we leverage conditional layer normalization to incorporate the language information into the model for better pronunciation when singers meet unseen languages. Additionally, gradient reversal layer (GRL) is utilized to remove singer biases included in lyrics since all singers are monolingual, which indicates singer's identity is implicitly associated with the text. The experiment is conducted on a combination of three singing voice datasets containing Japanese Kiritan dataset, English NUS-48E dataset, and one internal Chinese dataset. The result shows CrossSinger can synthesize high-fidelity songs for various singers with cross-lingual ability, including code-switch cases.

cs.SD↗

arXiv2023

Numerical analysis of pit-to-crack transition under corrosion fatigue using a stochastic pit generation algorithm

Corrosion fatigue is a major threat to the integrity of marine structures. However, the combined damaging effect of corrosion and fatigue is not well understood due to the uncertainties associated with the stochastic nature of pitting corrosion. A pitting corrosion defect could have a quite complex morphology that varies randomly from one case to another. Nevertheless, in numerical corrosion fatigue studies, the complex pit morphology is often idealized using an overly simplified geometry with a smooth surface because of the diffi-culties involved in modelling defects with the irregular and random shapes of pitting corrosion. The present study investigates the influence of such geometrical simplifications on the results of numerical corrosion fatigue analyses. For this purpose, an isolated complex-shaped pit, generated using a hierarchical stochastic algorithm scripted in Python and linked with Abaqus/CAE, is developed in a Q235 steel dog-bone specimen. The numer-ical results obtained from this model are compared with those from another model containing an idealized counterpart of the irregular pit. A discussion on the effect of pit morphology on the stress/strain history and distribution and fatigue crack initiation is presented.

math.NA↗

arXiv2023

Platform Behavior under Market Shocks: A Simulation Framework and Reinforcement-Learning Based Study

We study the behavior of an economic platform (e.g., Amazon, Uber Eats, Instacart) under shocks, such as COVID-19 lockdowns, and the effect of different regulation considerations imposed on a platform. To this end, we develop a multi-agent Gym environment of a platform economy in a dynamic, multi-period setting, with the possible occurrence of economic shocks. Buyers and sellers are modeled as economically-motivated agents, choosing whether or not to pay corresponding fees to use the platform. We formulate the platform's problem as a partially observable Markov decision process, and use deep reinforcement learning to model its fee setting and matching behavior. We consider two major types of regulation frameworks: (1) taxation policies and (2) platform fee restrictions, and offer extensive simulated experiments to characterize regulatory tradeoffs under optimal platform responses. Our results show that while many interventions are ineffective with a sophisticated platform actor, we identify a particular kind of regulation -- fixing fees to optimal, pre-shock fees while still allowing a platform to choose how to match buyer demands to sellers -- as promoting the efficiency, seller diversity, and resilience of the overall economic system.

cs.MA↗

arXiv2022

Differential Liquidity Provision in Uniswap v3 and Implications for Contract Design

Decentralized exchanges (DEXs) provide a means for users to trade pairs of assets on-chain without the need for a trusted third party to effectuate a trade. Amongst these, constant function market maker DEXs such as Uniswap handle the most volume of trades between ERC-20 tokens. With the introduction of Uniswap v3, liquidity providers can differentially allocate liquidity to trades that occur within specific price intervals. In this paper, we formalize the profit and loss that liquidity providers can earn when providing specific liquidity allocations to a v3 contract. We give a convex stochastic optimization problem for computing optimal liquidity allocation for a liquidity provider who holds a belief on how prices will evolve over time and use this to study the design question regarding how v3 contracts should partition the price space for permissible liquidity allocations. Our results show that making a greater diversity of price-space partitions available to a contract designer can simultaneously benefit both liquidity providers and traders.

cs.GT↗

arXiv2021

Federated Semi-Supervised Learning with Class Distribution Mismatch

Many existing federated learning (FL) algorithms are designed for supervised learning tasks, assuming that the local data owned by the clients are well labeled. However, in many practical situations, it could be difficult and expensive to acquire complete data labels. Federated semi-supervised learning (Fed-SSL) is an attractive solution for fully utilizing both labeled and unlabeled data. Similar to that encountered in federated supervised learning, class distribution of labeled/unlabeled data could be non-i.i.d. among clients. Besides, in each client, the class distribution of labeled data may be distinct from that of unlabeled data. Unfortunately, both can severely jeopardize the FL performance. To address such challenging issues, we introduce two proper regularization terms that can effectively alleviate the class distribution mismatch problem in Fed-SSL. In addition, to overcome the non-i.i.d. data, we leverage the variance reduction and normalized averaging techniques to develop a novel Fed-SSL algorithm. Theoretically, we prove that the proposed method has a convergence rate of $\mathcal{O}(1/\sqrt{T})$, where $T$ is the number of communication rounds, even when the data distribution are non-i.i.d. among clients. To the best of our knowledge, it is the first formal convergence result for Fed-SSL problems. Numerical experiments based on MNIST data and CIFAR-10 data show that the proposed method can greatly improve the classification accuracy compared to baselines.

cs.LG↗

arXiv2021

Designing a Combinatorial Financial Options Market

Financial options are contracts that specify the right to buy or sell an underlying asset at a strike price by an expiration date. Standard exchanges offer options of predetermined strike values and trade options of different strikes independently, even for those written on the same underlying asset. Such independent market design can introduce arbitrage opportunities and lead to the thin market problem. The paper first proposes a mechanism that consolidates and matches orders on standard options related to the same underlying asset, while providing agents the flexibility to specify any custom strike value. The mechanism generalizes the classic double auction, runs in time polynomial to the number of orders, and poses no risk to the exchange, regardless of the value of the underlying asset at expiration. Empirical analysis on real-market options data shows that the mechanism can find new matches for options of different strike prices and reduce bid-ask spreads. Extending standard options written on a single asset, we propose and define a new derivative instrument -- combinatorial financial options that offer contract holders the right to buy or sell any linear combination of multiple underlying assets. We generalize our single-asset mechanism to match options written on different combinations of assets, and prove that optimal clearing of combinatorial financial options is coNP-hard. To facilitate market operations, we propose an algorithm that finds the exact optimal match through iterative constraint generation, and evaluate its performance on synthetically generated combinatorial options markets of different scales. As option prices reveal the market's collective belief of an underlying asset's future value, a combinatorial options market enables the expression of aggregate belief about future correlations among assets.

cs.GT↗

arXiv2021

Incommensurate smectic phase in close proximity to the high-Tc superconductor FeSe/SrTiO3

Superconductivity is significantly enhanced in monolayer FeSe grown on SrTiO3, but not for multilayer films, in which large strength of nematicity develops. However, the link between the high-transition temperature superconductivity in monolayer and the correlation related nematicity in multilayer FeSe films is not well understood. Here, we use low-temperature scanning tunneling microscopy to study few-layer FeSe thin films grown by molecular beam epitaxy. We observe an incommensurate long-range smectic phase, which solely appears in bilayer FeSe films. The smectic order still locally exists and gradually fades away with increasing film thickness, while it suddenly vanishes in monolayer FeSe, indicative of an abrupt smectic phase transition. Surface alkali-metal doping can suppress the smectic phase and induce high-Tc superconductivity in bilayer FeSe. Our observations provide evidence that the monolayer FeSe is in close proximity to the smectic phase, and its superconductivity is likely enhanced by this electronic instability as well.

cond-mat.supr-con↗

arXiv2021

Log-time Prediction Markets for Interval Securities

We design a prediction market to recover a complete and fully general probability distribution over a random variable. Traders buy and sell interval securities that pay \$1 if the outcome falls into an interval and \$0 otherwise. Our market takes the form of a central automated market maker and allows traders to express interval endpoints of arbitrary precision. We present two designs in both of which market operations take time logarithmic in the number of intervals (that traders distinguish), providing the first computationally efficient market for a continuous variable. Our first design replicates the popular logarithmic market scoring rule (LMSR), but operates exponentially faster than a standard LMSR by exploiting its modularity properties to construct a balanced binary tree and decompose computations along the tree nodes. The second design consists of two or more parallel LMSR market makers that mediate submarkets of increasingly fine-grained outcome partitions. This design remains computationally efficient for all operations, including arbitrage removal across submarkets. It adds two additional benefits for the market designer: (1) the ability to express utility for information at various resolutions by assigning different liquidity values, and (2) the ability to guarantee a true constant bounded loss by appropriately decreasing the liquidity in each submarket.

cs.GT↗

arXiv2020

Electronic states and magnetic response of MnBi2Te4 by scanning tunneling microscopy and spectroscopy

Exotic quantum phenomena have been demonstrated in recently discovered intrinsic magnetic topological insulator MnBi2Te4. At its two-dimensional limit, quantum anomalous Hall (QAH) effect and axion insulator state are observed in odd and even layers of MnBi2Te4, respectively. The measured band structures exhibit intriguing and complex properties. Here we employ low-temperature scanning tunneling microscopy to study its surface states and magnetic response. The quasiparticle interference patterns indicate that the electronic structures on the topmost layer of MnBi2Te4 is different from that of the expected out-of-plane A-type antiferromagnetic phase. The topological surface states may be embedded in deeper layers beneath the topmost surface. Such novel electronic structure presumably related to the modification of crystalline structure during sample cleaving and re-orientation of magnetic moment of Mn atoms near the surface. Mn dopants substituted at the Bi site on the second atomic layer are observed. The ratio of Mn/Bi substitutions is 5%. The electronic structures are fluctuating at atomic scale on the surface, which can affect the magnetism of MnBi2Te4. Our findings shed new lights on the magnetic property of MnBi2Te4 and thus the design of magnetic topological insulators.

cond-mat.mtrl-sci↗

arXiv2019

Coulomb gap induced by electronic correlation and enlarged superconducting gap in laterally confined Pb islands grown on SrTiO3

We report high-resolution scanning tunneling microscopy (STM) study of nano-sized Pb islands grown on SrTiO3, where three distinct types of gaps with different energy scales are revealed. At low temperature, an enlarged superconducting gap (Δs) emerges while there is no enhancement in superconducting transition temperature (Tc), giving rise to a larger BCS ratio 2Δs/kBTc ~ 6.22. The strong coupling here may originate from the electron-phonon coupling on the metal-oxide interface. As the superconducting gap is suppressed under applied magnetic field or at elevated temperature, Coulomb gap and pseudogap appear, respectively. The Coulomb gap is sensitive to the lateral size of Pb islands, indicating that quantum size effect is able to influence electronic correlation, which is usually ignored in low-dimensional superconductivity. Our experimental results shall shed important light on the interplay between quantum size effect and correlations in nano-sized superconductors.

cond-mat.supr-con↗

arXiv2018

Cross-dataset Person Re-Identification Using Similarity Preserved Generative Adversarial Networks

Person re-identification (Re-ID) aims to match the image frames which contain the same person in the surveillance videos. Most of the Re-ID algorithms conduct supervised training in some small labeled datasets, so directly deploying these trained models to the real-world large camera networks may lead to a poor performance due to underfitting. The significant difference between the source training dataset and the target testing dataset makes it challenging to incrementally optimize the model. To address this challenge, we propose a novel solution by transforming the unlabeled images in the target domain to fit the original classifier by using our proposed similarity preserved generative adversarial networks model, SimPGAN. Specifically, SimPGAN adopts the generative adversarial networks with the cycle consistency constraint to transform the unlabeled images in the target domain to the style of the source domain. Meanwhile, SimPGAN uses the similarity consistency loss, which is measured by a siamese deep convolutional neural network, to preserve the similarity of the transformed images of the same person. Comprehensive experiments based on multiple real surveillance datasets are conducted, and the results show that our algorithm is better than the state-of-the-art cross-dataset unsupervised person Re-ID algorithms.

cs.CV↗

arXiv2017

T-CONV: A Convolutional Neural Network For Multi-scale Taxi Trajectory Prediction

Precise destination prediction of taxi trajectories can benefit many intelligent location based services such as accurate ad for passengers. Traditional prediction approaches, which treat trajectories as one-dimensional sequences and process them in single scale, fail to capture the diverse two-dimensional patterns of trajectories in different spatial scales. In this paper, we propose T-CONV which models trajectories as two-dimensional images, and adopts multi-layer convolutional neural networks to combine multi-scale trajectory patterns to achieve precise prediction. Furthermore, we conduct gradient analysis to visualize the multi-scale spatial patterns captured by T-CONV and extract the areas with distinct influence on the ultimate prediction. Finally, we integrate multiple local enhancement convolutional fields to explore these important areas deeply for better prediction. Comprehensive experiments based on real trajectory data show that T-CONV can achieve higher accuracy than the state-of-the-art methods.

cs.CV↗