arXiv subjects

Guo Yu

Publications and source records attributed to Guo Yu.

44 records · Page 3Linked to original sources

arXiv2022

Towards Fairness-Aware Multi-Objective Optimization

Recent years have seen the rapid development of fairness-aware machine learning in mitigating unfairness or discrimination in decision-making in a wide range of applications. However, much less attention has been paid to the fairness-aware multi-objective optimization, which is indeed commonly seen in real life, such as fair resource allocation problems and data driven multi-objective optimization problems. This paper aims to illuminate and broaden our understanding of multi-objective optimization from the perspective of fairness. To this end, we start with a discussion of user preferences in multi-objective optimization and then explore its relationship to fairness in machine learning and multi-objective optimization. Following the above discussions, representative cases of fairness-aware multiobjective optimization are presented, further elaborating the importance of fairness in traditional multi-objective optimization, data-driven optimization and federated optimization. Finally, challenges and opportunities in fairness-aware multi-objective optimization are addressed. We hope that this article makes a small step forward towards understanding fairness in the context of optimization and promote research interest in fairness-aware multi-objective optimization.

math.OC↗

arXiv2022

Tiny Adversarial Mulit-Objective Oneshot Neural Architecture Search

Due to limited computational cost and energy consumption, most neural network models deployed in mobile devices are tiny. However, tiny neural networks are commonly very vulnerable to attacks. Current research has proved that larger model size can improve robustness, but little research focuses on how to enhance the robustness of tiny neural networks. Our work focuses on how to improve the robustness of tiny neural networks without seriously deteriorating of clean accuracy under mobile-level resources. To this end, we propose a multi-objective oneshot network architecture search (NAS) algorithm to obtain the best trade-off networks in terms of the adversarial accuracy, the clean accuracy and the model size. Specifically, we design a novel search space based on new tiny blocks and channels to balance model size and adversarial performance. Moreover, since the supernet significantly affects the performance of subnets in our NAS algorithm, we reveal the insights into how the supernet helps to obtain the best subnet under white-box adversarial attacks. Concretely, we explore a new adversarial training paradigm by analyzing the adversarial transferability, the width of the supernet and the difference between training the subnets from scratch and fine-tuning. Finally, we make a statistical analysis for the layer-wise combination of certain blocks and channels on the first non-dominated front, which can serve as a guideline to design tiny neural network architectures for the resilience of adversarial perturbations.

cs.LG↗

arXiv2022

Evidence for a Monolayer Excitonic Insulator

The interplay between topology and correlations can generate a variety of quantum phases, many of which remain to be explored. Recent advances have identified monolayer WTe2 as a promising material for doing so in a highly tunable fashion. The ground state of this two-dimensional (2D) crystal can be electrostatically tuned from a quantum spin Hall insulator (QSHI) to a superconductor. However, much remains unknown about the gap-opening mechanism of the insulating state. Here we report evidence that the QSHI is also an excitonic insulator (EI), arising from the spontaneous formation of electron-hole bound states (excitons). We reveal the presence of an intrinsic insulating state at the charge neutrality point (CNP) in clean samples and confirm the correlated nature of this charge-neutral insulator by tunneling spectroscopy. We provide evidence against alternative scenarios of a band insulator or a localized insulator and support the existence of an EI phase in the clean limit. These observations lay the foundation for understanding a new class of correlated insulators with nontrivial topology and identify monolayer WTe2 as a promising candidate for exploring quantum phases of ground-state excitons.

cond-mat.mes-hall↗

arXiv2021

One-Dimensional Luttinger Liquids in a Two-Dimensional Moiré Lattice

The Luttinger liquid (LL) model of one-dimensional (1D) electronic systems provides a powerful tool for understanding strongly correlated physics including phenomena such as spin-charge separation. Substantial theoretical efforts have attempted to extend the LL phenomenology to two dimensions (2D), especially in models of closely packed arrays of 1D quantum wires, each being described as a LL. Such coupled-wire models have been successfully used to construct 2D anisotropic non-Fermi liquids, quantum Hall states, topological phases, and quantum spin liquids. However, an experimental demonstration of high-quality arrays of 1D LLs suitable for realizing these models remains absent. Here we report the experimental realization of 2D arrays of 1D LLs with crystalline quality in a moiré superlattice made of twisted bilayer tungsten ditelluride (tWTe$_{2}$). Originating from the anisotropic lattice of the monolayer, the moiré pattern of tWTe$_{2}$ hosts identical, parallel 1D electronic channels, separated by a fixed nanoscale distance, which is tunable by the interlayer twist angle. At a twist angle of ~ 5 degrees, we find that hole-doped tWTe$_{2}$ exhibits exceptionally large transport anisotropy with a resistance ratio of ~ 1000 between two orthogonal in-plane directions. The across-wire conductance exhibits power-law scaling behaviors, consistent with the formation of a 2D anisotropic phase that resembles an array of LLs. Our results open the door for realizing a variety of correlated and topological quantum phases based on coupled-wire models and LL physics.

cond-mat.mes-hall↗

arXiv2021

Landau Quantization and Highly Mobile Fermions in an Insulator

In strongly correlated materials, quasiparticle excitations can carry fractional quantum numbers. An intriguing possibility is the formation of fractionalized, charge-neutral fermions, e.g., spinons and fermionic excitons, that result in neutral Fermi surfaces and Landau quantization in an insulator. While previous experiments in quantum spin liquids, topological Kondo insulators, and quantum Hall systems have hinted at charge-neutral Fermi surfaces, evidence for their existence remains far from conclusive. Here we report experimental observation of Landau quantization in a two dimensional (2D) insulator, i.e., monolayer tungsten ditelluride (WTe$_{2}$), a large gap topological insulator. Using a detection scheme that avoids edge contributions, we uncover strikingly large quantum oscillations in the monolayer insulator's magnetoresistance, with an onset field as small as ~ 0.5 tesla. Despite the huge resistance, the oscillation profile, which exhibits many periods, mimics the Shubnikov-de Haas oscillations in metals. Remarkably, at ultralow temperatures the observed oscillations evolve into discrete peaks near 1.6 tesla, above which the Landau quantized regime is fully developed. Such a low onset field of quantization is comparable to high-mobility conventional two-dimensional electron gases. Our experiments call for further investigation of the highly unusual ground state of the WTe$_{2}$ monolayer. This includes the influence of device components and the possible existence of mobile fermions and charge-neutral Fermi surfaces inside its insulating gap.

cond-mat.mes-hall↗

arXiv2020

Consistency-based Semi-supervised Active Learning: Towards Minimizing Labeling Cost

Active learning (AL) combines data labeling and model training to minimize the labeling cost by prioritizing the selection of high value data that can best improve model performance. In pool-based active learning, accessible unlabeled data are not used for model training in most conventional methods. Here, we propose to unify unlabeled sample selection and model training towards minimizing labeling cost, and make two contributions towards that end. First, we exploit both labeled and unlabeled data using semi-supervised learning (SSL) to distill information from unlabeled data during the training stage. Second, we propose a consistency-based sample selection metric that is coherent with the training objective such that the selected samples are effective at improving model performance. We conduct extensive experiments on image classification tasks. The experimental results on CIFAR-10, CIFAR-100 and ImageNet demonstrate the superior performance of our proposed method with limited labeled data, compared to the existing methods and the alternative AL and SSL combinations. Additionally, we study an important yet under-explored problem -- "When can we start learning-based AL selection?". We propose a measure that is empirically correlated with the AL target loss and is potentially useful for determining the proper starting point of learning-based AL methods.

cs.LG↗

arXiv2019

Estimating the error variance in a high-dimensional linear model

The lasso has been studied extensively as a tool for estimating the coefficient vector in the high-dimensional linear model; however, considerably less is known about estimating the error variance in this context. In this paper, we propose the natural lasso estimator for the error variance, which maximizes a penalized likelihood objective. A key aspect of the natural lasso is that the likelihood is expressed in terms of the natural parameterization of the multiparameter exponential family of a Gaussian with unknown mean and variance. The result is a remarkably simple estimator of the error variance with provably good performance in terms of mean squared error. These theoretical results do not require placing any assumptions on the design matrix or the true regression coefficients. We also propose a companion estimator, called the organic lasso, which theoretically does not require tuning of the regularization parameter. Both estimators do well empirically compared to preexisting methods, especially in settings where successful recovery of the true support of the coefficient vector is hard. Finally, we show that existing methods can do well under fewer assumptions than previously known, thus providing a fuller story about the problem of estimating the error variance in high-dimensional linear models.

stat.ME↗

arXiv2017

Learning Local Dependence In Ordered Data

In many applications, data come with a natural ordering. This ordering can often induce local dependence among nearby variables. However, in complex data, the width of this dependence may vary, making simple assumptions such as a constant neighborhood size unrealistic. We propose a framework for learning this local dependence based on estimating the inverse of the Cholesky factor of the covariance matrix. Penalized maximum likelihood estimation of this matrix yields a simple regression interpretation for local dependence in which variables are predicted by their neighbors. Our proposed method involves solving a convex, penalized Gaussian likelihood problem with a hierarchical group lasso penalty. The problem decomposes into independent subproblems which can be solved efficiently in parallel using first-order methods. Our method yields a sparse, symmetric, positive definite estimator of the precision matrix, encoding a Gaussian graphical model. We derive theoretical results not found in existing methods attaining this structure. In particular, our conditions for signed support recovery and estimation consistency rates in multiple norms are as mild as those in a regression problem. Empirical results show our method performing favorably compared to existing methods. We apply our method to genomic data to flexibly model linkage disequilibrium. Our method is also applied to improve the performance of discriminant analysis in sound recording classification.

math.ST↗