arXiv subjects

Shuyang Dai

Publications and source records attributed to Shuyang Dai.

15 recordsLinked to original sources

arXiv2025

Investigating amorphization as a deformation mechanism using a novel phase field model at the mesoscale

Amorphization during severe plastic deformation has been observed in various crystalline materials, yet its underlying mechanisms remain poorly understood. This study introduces a novel phase-field model at the mesoscale, integrating elastoplastic theory with a deviatoric stress-dependent transformation strain tensor to capture stress-induced amorphization. The model enables quantitative predictions of amorphous phase nucleation and propagation under high stress, resolving distinctive microstructural patterns such as amorphous shear bands. Simulations reveal key phenomena, including avalanche-like amorphization, grain size effects, the Hall-Petch effect, and surface amorphization, consistent with experimental observations. By bridging phase-field methods with elastoplastic theory, this work provides a robust framework for studying amorphization as a deformation mechanism and offers valuable insights for designing materials resistant to extreme mechanical conditions.

cond-mat.mtrl-sci↗

arXiv2024

LLM-based Frameworks for API Argument Filling in Task-Oriented Conversational Systems

Task-orientated conversational agents interact with users and assist them via leveraging external APIs. A typical task-oriented conversational system can be broken down into three phases: external API selection, argument filling, and response generation. The focus of our work is the task of argument filling, which is in charge of accurately providing arguments required by the selected API. Upon comprehending the dialogue history and the pre-defined API schema, the argument filling task is expected to provide the external API with the necessary information to generate a desirable agent action. In this paper, we study the application of Large Language Models (LLMs) for the problem of API argument filling task. Our initial investigation reveals that LLMs require an additional grounding process to successfully perform argument filling, inspiring us to design training and prompting frameworks to ground their responses. Our experimental results demonstrate that when paired with proposed techniques, the argument filling performance of LLMs noticeably improves, paving a new way toward building an automated argument filling framework.

cs.CL↗

arXiv2024

GrounDial: Human-norm Grounded Safe Dialog Response Generation

Current conversational AI systems based on large language models (LLMs) are known to generate unsafe responses, agreeing to offensive user input or including toxic content. Previous research aimed to alleviate the toxicity, by fine-tuning LLM with manually annotated safe dialogue histories. However, the dependency on additional tuning requires substantial costs. To remove the dependency, we propose GrounDial, where response safety is achieved by grounding responses to commonsense social rules without requiring fine-tuning. A hybrid approach of in-context learning and human-norm-guided decoding of GrounDial enables the response to be quantitatively and qualitatively safer even without additional data or tuning.

cs.AI↗

arXiv2023

A Continuum Model for Dislocation Climb

Dislocation climb plays an important role in understanding plastic deformation of metallic materials at high temperature. In this paper, we present a continuum formulation for dislocation climb velocity based on densities of dislocations. The obtained continuum formulation is an accurate approximation of the Green's function based discrete dislocation dynamics method (Gu et al. J. Mech. Phys. Solids 83:319-337, 2015). The continuum dislocation climb formulation has the advantage of accounting for both the long-range effect of vacancy bulk diffusion and that of the Peach-Koehler climb force, and the two longrange effects are canceled into a short-range effect (integral with fast-decaying kernel) and in some special cases, a completely local effect. This significantly simplifies the calculation in the Green's function based discrete dislocation dynamics method, in which a linear system has to be solved over the entire system for the long-range effect of vacancy diffusion and the long-range Peach-Koehler climb force has to be calculated. This obtained continuum dislocation climb velocity can be applied in any available continuum dislocation dynamics frameworks. We also present numerical validations for this continuum climb velocity and simulation examples for implementation in continuum dislocation dynamics frameworks.

cond-mat.mtrl-sci↗

arXiv2022

New methods derived from energy minimization problems for solving two dimensional discrete dislocation dynamics

Dislocation dynamic is a typically gradient flow problem, and most of work solves it just as ODE, which means that the interacting energy of dislocations is ignored. We take the interaction energy into account and use it to introduce new methods to speed up the simulation. The non-singular stress field theory is used to make sure that the interacting energy between dislocations is finite and computational, and using this the two dimensional discrete dislocation dynamics can be rewritten into optimal problems. Based on it, the new problems from 2D dislocation dynamics can be solved by conjugate gradient method and other optimal methods. We introduce several methods into dislocation dynamics from the energy point of view and some numerical experiments are presented to compare different numerical methods, which show that the new methods are able to speed up relaxation procedures of dislocation dynamics. Those new approaches help to get the stable states of dislocations more quickly and speed up the simulations of dislocation dynamics.

cond-mat.mtrl-sci↗

arXiv2021

APo-VAE: Text Generation in Hyperbolic Space

Natural language often exhibits inherent hierarchical structure ingrained with complex syntax and semantics. However, most state-of-the-art deep generative models learn embeddings only in Euclidean vector space, without accounting for this structural property of language. In this paper, we investigate text generation in a hyperbolic latent space to learn continuous hierarchical representations. An Adversarial Poincare Variational Autoencoder (APo-VAE) is presented, where both the prior and variational posterior of latent variables are defined over a Poincare ball via wrapped normal distributions. By adopting the primal-dual formulation of KL divergence, an adversarial learning procedure is introduced to empower robust model training. Extensive experiments in language modeling and dialog-response generation tasks demonstrate the winning effectiveness of the proposed APo-VAE model over VAEs in Euclidean latent space, thanks to its superb capabilities in capturing latent language hierarchies in hyperbolic space.

cs.LG↗

arXiv2020

Contrastively Smoothed Class Alignment for Unsupervised Domain Adaptation

Recent unsupervised approaches to domain adaptation primarily focus on minimizing the gap between the source and the target domains through refining the feature generator, in order to learn a better alignment between the two domains. This minimization can be achieved via a domain classifier to detect target-domain features that are divergent from source-domain features. However, by optimizing via such domain classification discrepancy, ambiguous target samples that are not smoothly distributed on the low-dimensional data manifold are often missed. To solve this issue, we propose a novel Contrastively Smoothed Class Alignment (CoSCA) model, that explicitly incorporates both intra- and inter-class domain discrepancy to better align ambiguous target samples with the source domain. CoSCA estimates the underlying label hypothesis of target samples, and simultaneously adapts their feature representations by optimizing a proposed contrastive loss. In addition, Maximum Mean Discrepancy (MMD) is utilized to directly match features between source and target samples for better global alignment. Experiments on several benchmark datasets demonstrate that CoSCA can outperform state-of-the-art approaches for unsupervised domain adaptation by producing more discriminative features.

cs.LG↗

arXiv2020

Adaptation Across Extreme Variations using Unlabeled Domain Bridges

We tackle an unsupervised domain adaptation problem for which the domain discrepancy between labeled source and unlabeled target domains is large, due to many factors of inter and intra-domain variation. While deep domain adaptation methods have been realized by reducing the domain discrepancy, these are difficult to apply when domains are significantly unalike. In this work, we propose to decompose domain discrepancy into multiple but smaller, and thus easier to minimize, discrepancies by introducing unlabeled bridging domains that connect the source and target domains. We realize our proposal through an extension of the domain adversarial neural network with multiple discriminators, each of which accounts for reducing discrepancies between unlabeled (bridge, target) domains and a mix of all precedent domains including source. We validate the effectiveness of our method on several adaptation tasks including object recognition and semantic segmentation.

cs.CV↗

arXiv2020

Adversarial Text Generation via Feature-Mover's Distance

Generative adversarial networks (GANs) have achieved significant success in generating real-valued data. However, the discrete nature of text hinders the application of GAN to text-generation tasks. Instead of using the standard GAN objective, we propose to improve text-generation GAN via a novel approach inspired by optimal transport. Specifically, we consider matching the latent feature distributions of real and synthetic sentences using a novel metric, termed the feature-mover's distance (FMD). This formulation leads to a highly discriminative critic and easy-to-optimize objective, overcoming the mode-collapsing and brittle-training problems in existing methods. Extensive experiments are conducted on a variety of tasks to evaluate the proposed model empirically, including unconditional text generation, style transfer from non-parallel text, and unsupervised cipher cracking. The proposed model yields superior performance, demonstrating wide applicability and effectiveness.

cs.CL↗

arXiv2020

CLUB: A Contrastive Log-ratio Upper Bound of Mutual Information

Mutual information (MI) minimization has gained considerable interests in various machine learning tasks. However, estimating and minimizing MI in high-dimensional spaces remains a challenging problem, especially when only samples, rather than distribution forms, are accessible. Previous works mainly focus on MI lower bound approximation, which is not applicable to MI minimization problems. In this paper, we propose a novel Contrastive Log-ratio Upper Bound (CLUB) of mutual information. We provide a theoretical analysis of the properties of CLUB and its variational approximation. Based on this upper bound, we introduce a MI minimization training scheme and further accelerate it with a negative sampling strategy. Simulation studies on Gaussian distributions show the reliable estimation ability of CLUB. Real-world MI minimization experiments, including domain adaptation and information bottleneck, demonstrate the effectiveness of the proposed method. The code is at https://github.com/Linear95/CLUB.

cs.LG↗

arXiv2020

Bridging Maximum Likelihood and Adversarial Learning via $α$-Divergence

Maximum likelihood (ML) and adversarial learning are two popular approaches for training generative models, and from many perspectives these techniques are complementary. ML learning encourages the capture of all data modes, and it is typically characterized by stable training. However, ML learning tends to distribute probability mass diffusely over the data space, $e.g.$, yielding blurry synthetic images. Adversarial learning is well known to synthesize highly realistic natural images, despite practical challenges like mode dropping and delicate training. We propose an $α$-Bridge to unify the advantages of ML and adversarial learning, enabling the smooth transfer from one to the other via the $α$-divergence. We reveal that generalizations of the $α$-Bridge are closely related to approaches developed recently to regularize adversarial learning, providing insights into that prior work, and further understanding of why the $α$-Bridge performs well in practice.

cs.LG↗

arXiv2018

JointGAN: Multi-Domain Joint Distribution Learning with Generative Adversarial Nets

A new generative adversarial network is developed for joint distribution matching. Distinct from most existing approaches, that only learn conditional distributions, the proposed model aims to learn a joint distribution of multiple random variables (domains). This is achieved by learning to sample from conditional distributions between the domains, while simultaneously learning to sample from the marginals of each individual domain. The proposed framework consists of multiple generators and a single softmax-based critic, all jointly trained via adversarial learning. From a simple noise source, the proposed framework allows synthesis of draws from the marginals, conditional draws given observations from a subset of random variables, or complete draws from the full joint distribution. Most examples considered are for joint analysis of two domains, with examples for three domains also presented.

cs.LG↗

arXiv2018

A continuum model for distributions of dislocations incorporating short-range interactions

Dislocations are the main carriers of the permanent deformation of crystals. For simulations of engineering applications, continuum models where material microstructures are represented by continuous density distributions of dislocations are preferred. It is challenging to capture in the continuum model the short-range dislocation interactions, which vanish after the standard averaging procedure from discrete dislocation models. In this study, we consider systems of parallel straight dislocation walls and develop continuum descriptions for the short-range interactions of dislocations by using asymptotic analysis. The obtained continuum short-range interaction formulas are incorporated in the continuum model for dislocation dynamics based on a pair of dislocation density potential functions that represent continuous distributions of dislocations. This derived continuum model is able to describe the anisotropic dislocation interaction and motion. Mathematically, these short-range interaction terms ensure strong stability property of the continuum model that is possessed by the discrete dislocation dynamics model. The derived continuum model is validated by comparisons with the discrete dislocation dynamical simulation results.

cond-mat.mtrl-sci↗

arXiv2017

Symmetric Variational Autoencoder and Connections to Adversarial Learning

A new form of the variational autoencoder (VAE) is proposed, based on the symmetric Kullback-Leibler divergence. It is demonstrated that learning of the resulting symmetric VAE (sVAE) has close connections to previously developed adversarial-learning methods. This relationship helps unify the previously distinct techniques of VAE and adversarially learning, and provides insights that allow us to ameliorate shortcomings with some previously developed adversarial methods. In addition to an analysis that motivates and explains the sVAE, an extensive set of experiments validate the utility of the approach.

stat.ML↗

arXiv2015

Accurate Measurement of the Cleavage Energy of Graphite

The basal plane cleavage energy (CE) of graphite is a key material parameter for understanding many of the unusual properties of graphite, graphene, and carbon nanotubes. The CE is equal to twice the surface energy and is closely related to the interlayer binding energy and exfoliation energy of graphite. Nonetheless, a wide range of values for these properties have been reported and no consensus has yet emerged as to their magnitude. Here, we report the first direct, accurate experimental measurement of the CE of graphite using a novel method based on the recently discovered self-retraction phenomenon in graphite. The measured value, 0.37 +/- 0.01 J/m2 for the incommensurate state of bicrystal graphite, is nearly invariant with respect to temperature (from 22°C to 198°C) and bicrystal twist angle, and insensitive to impurities (from the atmosphere). The cleavage energy for the ideal ABAB graphite stacking, 0.39 +/- 0.02 J/m2, is calculated based upon a combination of the measured CE and a theoretical calculation. These experimental measurements are ideal for use in evaluating the efficacy of competing theoretical approaches.

cond-mat.mtrl-sci↗