arXiv subjects

Yue Wang

Publications and source records attributed to Yue Wang.

At least 721 records · Page 40Linked to original sources

arXiv2021

Robust factored principal component analysis for matrix-valued outlier accommodation and detection

Principal component analysis (PCA) is a popular dimension reduction technique for vector data. Factored PCA (FPCA) is a probabilistic extension of PCA for matrix data, which can substantially reduce the number of parameters in PCA while yield satisfactory performance. However, FPCA is based on the Gaussian assumption and thereby susceptible to outliers. Although the multivariate $t$ distribution as a robust modeling tool for vector data has a very long history, its application to matrix data is very limited. The main reason is that the dimension of the vectorized matrix data is often very high and the higher the dimension, the lower the breakdown point that measures the robustness. To solve the robustness problem suffered by FPCA and make it applicable to matrix data, in this paper we propose a robust extension of FPCA (RFPCA), which is built upon a $t$-type distribution called matrix-variate $t$ distribution. Like the multivariate $t$ distribution, the matrix-variate $t$ distribution can adaptively down-weight outliers and yield robust estimates. We develop a fast EM-type algorithm for parameter estimation. Experiments on synthetic and real-world datasets reveal that RFPCA is compared favorably with several related methods and RFPCA is a simple but powerful tool for matrix-valued outlier detection.

stat.ML↗

arXiv2021

Auto-Tag: Tagging-Data-By-Example in Data Lakes

As data lakes become increasingly popular in large enterprises today, there is a growing need to tag or classify data assets (e.g., files and databases) in data lakes with additional metadata (e.g., semantic column-types), as the inferred metadata can enable a range of downstream applications like data governance (e.g., GDPR compliance), and dataset search. Given the sheer size of today's enterprise data lakes with petabytes of data and millions of data assets, it is imperative that data assets can be ``auto-tagged'', using lightweight inference algorithms and minimal user input. In this work, we develop Auto-Tag, a corpus-driven approach that automates data-tagging of \textit{custom} data types in enterprise data lakes. Using Auto-Tag, users only need to provide \textit{one} example column to demonstrate the desired data-type to tag. Leveraging an index structure built offline using a lightweight scan of the data lake, which is analogous to pre-training in machine learning, Auto-Tag can infer suitable data patterns to best ``describe'' the underlying ``domain'' of the given column at an interactive speed, which can then be used to tag additional data of the same ``type'' in data lakes. The Auto-Tag approach can adapt to custom data-types, and is shown to be both accurate and efficient. Part of Auto-Tag ships as a ``custom-classification'' feature in a cloud-based data governance and catalog solution \textit{Azure Purview}.

cs.DB↗

arXiv2021

Transition metal dichalcogenide dimer nano-antennas with ultra-small gaps

Transition metal dichalcogenides have emerged as promising materials for nano-photonic resonators due to their large refractive index, low absorption within a large portion of the visible spectrum and compatibility with a wide range of substrates. Here we use these properties to fabricate WS$_2$ double-pillar nano-antennas in a variety of geometries enabled by the anisotropy in the crystal structure. Using dark field spectroscopy, we reveal multiple Mie resonances, to which we couple WSe$_2$ monolayer photoluminescence and achieve Purcell enhancement and an increased fluorescence by factors up to 240. We introduce post-fabrication atomic force microscope repositioning and rotation of dimer nano-antennas, achieving gaps as small as 10$\pm$5 nm, opening the possibility to a host of potential applications including strong Purcell enhancement of single photon emitters and optical trapping, which we study in simulations. Our findings highlight the advantages of using transition metal dichalcogenides for nano-photonics by exploring new applications enabled by their unique properties.

physics.app-ph↗

arXiv2021

Transformer-based Network for RGB-D Saliency Detection

RGB-D saliency detection integrates information from both RGB images and depth maps to improve prediction of salient regions under challenging conditions. The key to RGB-D saliency detection is to fully mine and fuse information at multiple scales across the two modalities. Previous approaches tend to apply the multi-scale and multi-modal fusion separately via local operations, which fails to capture long-range dependencies. Here we propose a transformer-based network to address this issue. Our proposed architecture is composed of two modules: a transformer-based within-modality feature enhancement module (TWFEM) and a transformer-based feature fusion module (TFFM). TFFM conducts a sufficient feature fusion by integrating features from multiple scales and two modalities over all positions simultaneously. TWFEM enhances feature on each scale by selecting and integrating complementary information from other scales within the same modality before TFFM. We show that transformer is a uniform operation which presents great efficacy in both feature fusion and feature enhancement, and simplifies the model design. Extensive experimental results on six benchmark datasets demonstrate that our proposed network performs favorably against state-of-the-art RGB-D saliency detection methods.

cs.CV↗

arXiv2021

Anti-degenerated UWB-LiDAR Localization for Automatic Road Roller in Tunnel

The automatic road roller, as a popular type of construction robot, has attracted much interest from both the industry and the research community in recent years. However, when it comes to tunnels where the degeneration issues are prone to happen, it is still a challenging problem to provide an accurate positioning result for the robot. In this paper, we aim to deal with this problem by fusing LiDAR and UWB measurements based on optimization. In the proposed localization method, the directions of non-degeneration will be constrained and the covariance of UWB reconstruction will be introduced to improve the accuracy of localization. Apart from these, a method that can extract the feature of the inner wall of tunnels to assist positioning is also presented in this paper. To evaluate the effectiveness of the proposed method, three experiments with real road roller were carried out and the results show that our method can achieve better performance than the existing methods and can be applied to automatic road roller working inside tunnels. Finally, we discuss the feasibility of deploying the system in real applications and make several recommendations.

cs.RO↗

arXiv2021

Weakly Supervised Prototype Topic Model with Discriminative Seed Words: Modifying the Category Prior by Self-exploring Supervised Signals

Dataless text classification, i.e., a new paradigm of weakly supervised learning, refers to the task of learning with unlabeled documents and a few predefined representative words of categories, known as seed words. The recent generative dataless methods construct document-specific category priors by using seed word occurrences only, however, such category priors often contain very limited and even noisy supervised signals. To remedy this problem, in this paper we propose a novel formulation of category prior. First, for each document, we consider its label membership degree by not only counting seed word occurrences, but also using a novel prototype scheme, which captures pseudo-nearest neighboring categories. Second, for each label, we consider its frequency prior knowledge of the corpus, which is also a discriminative knowledge for classification. By incorporating the proposed category prior into the previous generative dataless method, we suggest a novel generative dataless method, namely Weakly Supervised Prototype Topic Model (WSPTM). The experimental results on real-world datasets demonstrate that WSPTM outperforms the existing baseline methods.

cs.CL↗

arXiv2021

Kernel estimation of bivariate time-varying coefficient model for longitudinal data with terminal event

We propose a nonparametric bivariate time-varying coefficient model for longitudinal measurements with the occurrence of a terminal event that is subject to right censoring. The time-varying coefficients capture the longitudinal trajectories of covariate effects along with both the followup time and the residual lifetime. The proposed model extends the parametric conditional approach given terminal event time in recent literature, and thus avoids potential model misspecification. We consider a kernel smoothing method for estimating regression coefficients in our model and use cross-validation for bandwidth selection, applying undersmoothing in the final analysis to eliminate the asymptotic bias of the kernel estimator. We show that the kernel estimates are asymptotically normal under mild regularity conditions, and provide an easily computable sandwich variance estimator. We conduct extensive simulations that show desirable performance of the proposed approach, and apply the method to analyzing the medical cost data for patients with end-stage renal disease.

stat.ME↗

arXiv2021

Incorporating NODE with Pre-trained Neural Differential Operator for Learning Dynamics

Learning dynamics governed by differential equations is crucial for predicting and controlling the systems in science and engineering. Neural Ordinary Differential Equation (NODE), a deep learning model integrated with differential equations, is popular in learning dynamics recently due to its robustness to irregular samples and its flexibility to high-dimensional input. However, the training of NODE is sensitive to the precision of the numerical solver, which makes the convergence of NODE unstable, especially for ill-conditioned dynamical systems. In this paper, to reduce the reliance on the numerical solver, we propose to enhance the supervised signal in the training of NODE. Specifically, we pre-train a neural differential operator (NDO) to output an estimation of the derivatives to serve as an additional supervised signal. The NDO is pre-trained on a class of basis functions and learns the mapping between the trajectory samples of these functions to their derivatives. To leverage both the trajectory signal and the estimated derivatives from NDO, we propose an algorithm called NDO-NODE, in which the loss function contains two terms: the fitness on the true trajectory samples and the fitness on the estimated derivatives that are outputted by the pre-trained NDO. Experiments on various kinds of dynamics show that our proposed NDO-NODE can consistently improve the forecasting accuracy with one pre-trained NDO. Especially for the stiff ODEs, we observe that NDO-NODE can capture the transitions in the dynamics more accurately compared with other regularization methods.

cs.LG↗

arXiv2021

A powerful test for differentially expressed gene pathways via graph-informed structural equation modeling

A major task in genetic studies is to identify genes related to human diseases and traits to understand functional characteristics of genetic mutations and enhance patient diagnosis. Besides marginal analyses of individual genes, identification of gene pathways, i.e., a set of genes with known interactions that collectively contribute to specific biological functions, can provide more biologically meaningful results. Such gene pathway analysis can be formulated into a high-dimensional two-sample testing problem. Due to the typically limited sample size of gene expression datasets, most existing two-sample tests may have compromised powers because they ignore or only inefficiently incorporate the auxiliary pathway information on gene interactions. We propose T2-DAG, a Hotelling's $T^2$-type test for detecting differentially expressed gene pathways, which efficiently leverages the auxiliary pathway information on gene interactions through a linear structural equation model. We establish the asymptotic distribution of the test statistic under pertinent assumptions. Simulation studies under various scenarios show that T2-DAG outperforms several representative existing methods with well-controlled type-I error rates and substantially improved powers, even with incomplete or inaccurate pathway information or unadjusted confounding effects. We also illustrate the performance of T2-DAG in an application to detect differentially expressed KEGG pathways between different stages of lung cancer.

stat.ME↗

arXiv2021

R-Drop: Regularized Dropout for Neural Networks

Dropout is a powerful and widely used technique to regularize the training of deep neural networks. In this paper, we introduce a simple regularization strategy upon dropout in model training, namely R-Drop, which forces the output distributions of different sub models generated by dropout to be consistent with each other. Specifically, for each training sample, R-Drop minimizes the bidirectional KL-divergence between the output distributions of two sub models sampled by dropout. Theoretical analysis reveals that R-Drop reduces the freedom of the model parameters and complements dropout. Experiments on $\bf{5}$ widely used deep learning tasks ($\bf{18}$ datasets in total), including neural machine translation, abstractive summarization, language understanding, language modeling, and image classification, show that R-Drop is universally effective. In particular, it yields substantial improvements when applied to fine-tune large-scale pre-trained models, e.g., ViT, RoBERTa-large, and BART, and achieves state-of-the-art (SOTA) performances with the vanilla Transformer model on WMT14 English$\to$German translation ($\bf{30.91}$ BLEU) and WMT14 English$\to$French translation ($\bf{43.95}$ BLEU), even surpassing models trained with extra large-scale data and expert-designed advanced variants of Transformer models. Our code is available at GitHub{\url{https://github.com/dropreg/R-Drop}}.

cs.LG↗

arXiv2021

Non-Asymptotic Analysis for Two Time-scale TDC with General Smooth Function Approximation

Temporal-difference learning with gradient correction (TDC) is a two time-scale algorithm for policy evaluation in reinforcement learning. This algorithm was initially proposed with linear function approximation, and was later extended to the one with general smooth function approximation. The asymptotic convergence for the on-policy setting with general smooth function approximation was established in [bhatnagar2009convergent], however, the finite-sample analysis remains unsolved due to challenges in the non-linear and two-time-scale update structure, non-convex objective function and the time-varying projection onto a tangent plane. In this paper, we develop novel techniques to explicitly characterize the finite-sample error bound for the general off-policy setting with i.i.d.\ or Markovian samples, and show that it converges as fast as $\mathcal O(1/\sqrt T)$ (up to a factor of $\mathcal O(\log T)$). Our approach can be applied to a wide range of value-based reinforcement learning algorithms with general smooth function approximation.

cs.LG↗

arXiv2021

Online Robust Reinforcement Learning with Model Uncertainty

Robust reinforcement learning (RL) is to find a policy that optimizes the worst-case performance over an uncertainty set of MDPs. In this paper, we focus on model-free robust RL, where the uncertainty set is defined to be centering at a misspecified MDP that generates a single sample trajectory sequentially and is assumed to be unknown. We develop a sample-based approach to estimate the unknown uncertainty set and design a robust Q-learning algorithm (tabular case) and robust TDC algorithm (function approximation setting), which can be implemented in an online and incremental fashion. For the robust Q-learning algorithm, we prove that it converges to the optimal robust Q function, and for the robust TDC algorithm, we prove that it converges asymptotically to some stationary points. Unlike the results in [Roy et al., 2017], our algorithms do not need any additional conditions on the discount factor to guarantee the convergence. We further characterize the finite-time error bounds of the two algorithms and show that both the robust Q-learning and robust TDC algorithms converge as fast as their vanilla counterparts(within a constant factor). Our numerical experiments further demonstrate the robustness of our algorithms. Our approach can be readily extended to robustify many other algorithms, e.g., TD, SARSA, and other GTD algorithms.

cs.LG↗

arXiv2021

Band-selective third-harmonic generation in superconducting MgB$_2$: Possible evidence for Higgs amplitude mode in the dirty limit

We report on time-resolved linear and nonlinear terahertz spectroscopy of the two-band superconductor MgB$_2$ with the superconducting transition temperature $T_c \approx 36$ K. Third-harmonic generation (THG) is observed below $T_c$ by driving the system with intense narrowband THz pulses. For the pump-pulse frequencies $f=$ 0.3, 0.4, and 0.5 THz, temperature-dependent evolution of the THG signals exhibits a resonance maximum at the temperatures with the resonance conditions $2f=2Δ_π(T)$ fulfilled, for the dirty-limit superconducting gap $2Δ_π$. In contrast, for $f=$ 0.6 and 0.7 THz with $2f>2Δ_π(T\rightarrow0)=1.03$ THz, the THG intensity increases monotonically with decreasing temperature. Moreover, for $2f<2Δ_π(T\rightarrow0)$ the THG is found nearly isotropic with respect to the pump-pulse polarization. These results suggest a predominant contribution of the driven Higgs amplitude mode of the dirty-limit $π$-band superconducting gap, pointing to the importance of scattering for observation of the Higgs mode in superconductors.

cond-mat.supr-con↗

arXiv2021

Electric Vehicle Automatic Charging System Based on Vision-force Fusion

Electric vehicles are an emerging means of transportation with environmental friendliness. The automatic charging is a hot topic in this field that is full of challenges. We introduce a complete automatic charging system based on vision-force fusion, which includes perception, planning and control for robot manipulations of the system. We design the whole system in simulation and transfer it to the real world. The experimental results prove the effectiveness of our system.

cs.RO↗

arXiv2021

Revisiting Latent-Space Interpolation via a Quantitative Evaluation Framework

Latent-space interpolation is commonly used to demonstrate the generalization ability of deep latent variable models. Various algorithms have been proposed to calculate the best trajectory between two encodings in the latent space. In this work, we show how data labeled with semantically continuous attributes can be utilized to conduct a quantitative evaluation of latent-space interpolation algorithms, for variational autoencoders. Our framework can be used to complement the standard qualitative comparison, and also enables evaluation for domains (such as graph) in which the visualization is difficult. Interestingly, our experiments reveal that the superiority of interpolation algorithms could be domain-dependent. While normalised interpolation works best for the image domain, spherical linear interpolation achieves the best performance in the graph domain. Next, we propose a simple-yet-effective method to restrict the latent space via a bottleneck structure in the encoder. We find that all interpolation algorithms evaluated in this work can benefit from this restriction. Finally, we conduct interpolation-aware training with the labeled attributes, and show that this explicit supervision can improve the interpolation performance.

cs.LG↗

arXiv2021

DETR3D: 3D Object Detection from Multi-view Images via 3D-to-2D Queries

We introduce a framework for multi-camera 3D object detection. In contrast to existing works, which estimate 3D bounding boxes directly from monocular images or use depth prediction networks to generate input for 3D object detection from 2D information, our method manipulates predictions directly in 3D space. Our architecture extracts 2D features from multiple camera images and then uses a sparse set of 3D object queries to index into these 2D features, linking 3D positions to multi-view images using camera transformation matrices. Finally, our model makes a bounding box prediction per object query, using a set-to-set loss to measure the discrepancy between the ground-truth and the prediction. This top-down approach outperforms its bottom-up counterpart in which object bounding box prediction follows per-pixel depth estimation, since it does not suffer from the compounding error introduced by a depth prediction model. Moreover, our method does not require post-processing such as non-maximum suppression, dramatically improving inference speed. We achieve state-of-the-art performance on the nuScenes autonomous driving benchmark.

cs.CV↗

arXiv2021

Object DGCNN: 3D Object Detection using Dynamic Graphs

3D object detection often involves complicated training and testing pipelines, which require substantial domain knowledge about individual datasets. Inspired by recent non-maximum suppression-free 2D object detection models, we propose a 3D object detection architecture on point clouds. Our method models 3D object detection as message passing on a dynamic graph, generalizing the DGCNN framework to predict a set of objects. In our construction, we remove the necessity of post-processing via object confidence aggregation or non-maximum suppression. To facilitate object detection from sparse point clouds, we also propose a set-to-set distillation approach customized to 3D detection. This approach aligns the outputs of the teacher model and the student model in a permutation-invariant fashion, significantly simplifying knowledge distillation for the 3D detection task. Our method achieves state-of-the-art performance on autonomous driving benchmarks. We also provide abundant analysis of the detection model and distillation framework.

cs.CV↗

arXiv2021

Machine Translation Verbosity Control for Automatic Dubbing

Automatic dubbing aims at seamlessly replacing the speech in a video document with synthetic speech in a different language. The task implies many challenges, one of which is generating translations that not only convey the original content, but also match the duration of the corresponding utterances. In this paper, we focus on the problem of controlling the verbosity of machine translation output, so that subsequent steps of our automatic dubbing pipeline can generate dubs of better quality. We propose new methods to control the verbosity of MT output and compare them against the state of the art with both intrinsic and extrinsic evaluations. For our experiments we use a public data set to dub English speeches into French, Italian, German and Spanish. Finally, we report extensive subjective tests that measure the impact of MT verbosity control on the final quality of dubbed video clips.

cs.CL↗