Results for "Siwei Lyu"

total 888took 0.13s
LSTM with Working MemoryMay 06 2016May 13 2016LSTM is arguably the most successful RNN architecture for many tasks that involve sequential information. In the past few years there have been several proposed improvements to LSTM. We propose an improvement to LSTM which allows communication between ... More
Exposing Deep Fakes Using Inconsistent Head PosesNov 01 2018Nov 13 2018In this paper, we propose a new method to expose AI-generated fake face images or videos (commonly known as the Deep Fakes). Our method is based on the observations that Deep Fakes are created by splicing synthesized face region into the original image, ... More
Interpretation and Generalization of Score MatchingMay 09 2012Score matching is a recently developed parameter learning method that is particularly effective to complicated high dimensional density models with intractable partition functions. In this paper, we study two issues that have not been completely resolved ... More
Exposing DeepFake Videos By Detecting Face Warping ArtifactsNov 01 2018Mar 29 2019In this work, we describe a new deep learning based method that can effectively distinguish AI-generated fake videos (referred to as {\em DeepFake} videos hereafter) from real videos. Our method is based on the observations that current DeepFake algorithm ... More
Object-driven Text-to-Image Synthesis via Adversarial TrainingFeb 27 2019In this paper, we propose Object-driven Attentive Generative Adversarial Newtorks (Obj-GANs) that allow object-centered text-to-image synthesis for complex scenes. Following the two-step (layout-image) generation process, a novel object-driven attentive ... More
Exposing DeepFake Videos By Detecting Face Warping ArtifactsNov 01 2018In this work, we describe a new deep learning based method that can effectively distinguish AI-generated fake videos (referred to as DeepFake videos hereafter) from real videos. Our method is based on the observations that current DeepFake algorithm can ... More
Data Priming Network for Automatic Check-OutApr 10 2019Automatic Check-Out (ACO) receives increased interests in recent years. An important component of the ACO system is the visual item counting, which recognize the categories and counts of the items chosen by the customers. However, the training of such ... More
Exposing DeepFake Videos By Detecting Face Warping ArtifactsNov 01 2018May 22 2019In this work, we describe a new deep learning based method that can effectively distinguish AI-generated fake videos (referred to as {\em DeepFake} videos hereafter) from real videos. Our method is based on the observations that current DeepFake algorithm ... More
A Univariate Bound of Area Under ROCApr 16 2018May 25 2018Area under ROC (AUC) is an important metric for binary classification and bipartite ranking problems. However, it is difficult to directly optimizing AUC as a learning objective, so most existing algorithms are based on optimizing a surrogate loss to ... More
LSTM with Working MemoryMay 06 2016Mar 30 2017Previous RNN architectures have largely been superseded by LSTM, or "Long Short-Term Memory". Since its introduction, there have been many variations on this simple design. However, it is still widely used and we are not aware of a gated-RNN architecture ... More
De-identification without losing facesFeb 12 2019Training of deep learning models for computer vision requires large image or video datasets from real world. Often, in collecting such datasets, we need to protect the privacy of the people captured in the images or videos, while still preserve the useful ... More
Scale Invariant Fully Convolutional Network: Detecting Hands EfficientlyJun 11 2019Existing hand detection methods usually follow the pipeline of multiple stages with high computation cost, i.e., feature extraction, region proposal, bounding box regression, and additional layers for rotated region detection. In this paper, we propose ... More
Tagging like Humans: Diverse and Distinct Image AnnotationMar 31 2018In this work we propose a new automatic image annotation model, dubbed {\bf diverse and distinct image annotation} (D2IA). The generative model D2IA is inspired by the ensemble of human annotations, which create semantically relevant, yet distinct and ... More
Attention-driven Tree-structured Convolutional LSTM for High Dimensional Data UnderstandingJan 29 2019Modeling the sequential information of image sequences has been a vital step of various vision tasks and convolutional long short-term memory (ConvLSTM) has demonstrated its superb performance in such spatiotemporal problems. Nevertheless, the hierarchical ... More
Discussion on the origin of magic numbers in clustersNov 01 2014The distribution of the sizes of clusters is not continuous, but rather has local maxima. The numbers of atoms of those maxima distribution is called magic numbers. Two methods of determining magic numbers are firstly introduced, followed by three different ... More
Who did What at Where and When: Simultaneous Multi-Person Tracking and Activity RecognitionJul 03 2018We present a bootstrapping framework to simultaneously improve multi-person tracking and activity recognition at individual, interaction and social group activity levels. The inference consists of identifying trajectories of all pedestrian actors, individual ... More
Contrast Enhancement Estimation for Digital Image ForensicsJun 13 2017Inconsistency in contrast enhancement can be used to expose image forgeries. In this work, we describe a new method to estimate contrast enhancement from a single image. Our method takes advantage of the nature of contrast enhancement as a mapping between ... More
In Ictu Oculi: Exposing AI Generated Fake Face Videos by Detecting Eye BlinkingJun 07 2018Jun 11 2018The new developments in deep generative networks have significantly improve the quality and efficiency in generating realistically-looking fake face videos. In this work, we describe a new method to expose fake face videos generated with neural networks. ... More
STS Classification with Dual-stream CNNMay 20 2018The structured time series (STS) classification problem requires the modeling of interweaved spatiotemporal dependency. most previous STS classification methods model the spatial and temporal dependencies independently. Due to the complexity of the STS ... More
Learning with Average Top-k LossMay 24 2017Dec 20 2017In this work, we introduce the {\em average top-$k$} (\atk) loss as a new aggregate loss for supervised learning, which is the average over the $k$ largest individual losses over a training dataset. We show that the \atk loss is a natural generalization ... More
Multi-Scale Structure-Aware Network for Human Pose EstimationMar 27 2018Sep 16 2018We develop a robust multi-scale structure-aware neural network for human pose estimation. This method improves the recent deep conv-deconv hourglass models with four key improvements: (1) multi-scale supervision to strengthen contextual feature learning ... More
Learning Non-Uniform Hypergraph for Multi-Object TrackingDec 10 2018The majority of Multi-Object Tracking (MOT) algorithms based on the tracking-by-detection scheme do not use higher order dependencies among objects or tracklets, which makes them less effective in handling complex scenarios. In this work, we present a ... More
Exploring the Vulnerability of Single Shot Module in Object Detectors via Imperceptible Background PatchesSep 16 2018Nov 17 2018Recent works succeeded to generate adversarial perturbations on the entire image or the object of interests to corrupt CNN based object detectors. In this paper, we focus on exploring the vulnerability of the Single Shot Module (SSM) commonly used in ... More
Multi-Scale Supervised Network for Human Pose EstimationAug 05 2018Human pose estimation is an important topic in computer vision with many applications including gesture and activity recognition. However, pose estimation from image is challenging due to appearance variations, occlusions, clutter background, and complex ... More
Robust Adversarial Perturbation on Deep Proposal-based ModelsSep 16 2018Adversarial noises are useful tools to probe the weakness of deep learning based computer vision algorithms. In this paper, we describe a robust adversarial perturbation (R-AP) method to attack deep proposal-based object detectors and instance segmentation ... More
Improving Image Restoration with Soft-RoundingAug 20 2015Several important classes of images such as text, barcode and pattern images have the property that pixels can only take a distinct subset of values. This knowledge can benefit the restoration of such images, but it has not been widely considered in current ... More
Evolvement Constrained Adversarial Learning for Video Style TransferNov 06 2018Video style transfer is a useful component for applications such as augmented reality, non-photorealistic rendering, and interactive games. Many existing methods use optical flow to preserve the temporal smoothness of the synthesized video. However, the ... More
Multi-label Learning with Missing Labels using Mixed Dependency GraphsMar 31 2018This work focuses on the problem of multi-label learning with missing labels (MLML), which aims to label each test instance with multiple class labels given training instances that have an incomplete/partial set of these labels. The key point to handle ... More
Exposing GAN-synthesized Faces Using Landmark LocationsMar 30 2019Generative adversary networks (GANs) have recently led to highly realistic image synthesis results. In this work, we describe a new method to expose GAN-synthesized images using the locations of the facial landmark points. Our method is based on the observations ... More
Thompson Sampling for Combinatorial Semi-BanditsMar 13 2018Jun 21 2018We study the application of the Thompson sampling (TS) methodology to the stochastic combinatorial multi-armed bandit (CMAB) framework. We analyze the standard TS algorithm for the general CMAB, and obtain the first distribution-dependent regret bound ... More
Field theory for biophysical neural networksNov 05 2014The human brain is a complex system composed of a network of hundreds of billions of discrete neurons that are coupled through time dependent synapses. Simulating the entire brain is a daunting challenge. Here, we show how ideas from quantum field theory ... More
Multi-armed Bandits with CompensationNov 05 2018We propose and study the known-compensation multi-arm bandit (KCMAB) problem, where a system controller offers a set of arms to many short-term players for $T$ steps. In each step, one short-term player arrives to the system. Upon arrival, the player ... More
Graph Autoencoder-Based Unsupervised Feature Selection with Broad and Local Data Structure PreservationJan 07 2018Apr 21 2018Feature selection is a dimensionality reduction technique that selects a subset of representative features from high dimensional data by eliminating irrelevant and redundant features. Recently, feature selection combined with sparse learning has attracted ... More
Robust Localized Multi-view Subspace ClusteringMay 22 2017In multi-view clustering, different views may have different confidence levels when learning a consensus representation. Existing methods usually address this by assigning distinctive weights to different views. However, due to noisy nature of real-world ... More
Geometric Hypergraph Learning for Visual TrackingMar 18 2016Graph based representation is widely used in visual tracking field by finding correct correspondences between target parts in consecutive frames. However, most graph based trackers consider pairwise geometric relations between local parts. They do not ... More
Synchronization of finite-state pulse-coupled oscillatorsJul 04 2014Mar 30 2015We propose a novel generalized cellular automaton(GCA) model for discrete-time pulse-coupled oscillators and study the emergence of synchrony. Given a finite simple graph and an integer $n\ge 3$, each vertex is an identical oscillator of period $n$ with ... More
Knowledge-Based Sequential Decision-Making Under UncertaintyMay 16 2019Deep reinforcement learning (DRL) algorithms have achieved great success on sequential decision-making problems, yet is criticized for the lack of data-efficiency and explainability. Especially, explainability of subtasks is critical in hierarchical decision-making ... More
Phase transition in firefly cellular automata on finite treesOct 04 2016We study a one-parameter family of discrete dynamical systems called the $\kappa$-color firefly cellular automata, proposed by the author in [13] as discrete models for finite-state pulse-coupled inhibitory oscillators. At each discrete time $t$, each ... More
A Note on Graph Characteristics and Hadwiger's ConjectureMar 16 2012Nov 27 2012This is a note on three graph parameters motivated by the Euler-Poincare characteristic for simplicial complex. We show those three graph parameters of a given connected graph $G$ is greater than or equal to that of the complete graph with $\max(h(G),\chi(G))$ ... More
Phase synchronization of pulse-coupled excitable clocksApr 28 2016Consider a distributed network on a simple graph $G=(V,E)$ where each node has a phase oscillator revolving on $S^{1}=\mathbb{R}/\mathbb{Z}$ with natural period of 1 second. Pulse-coupling is a class of distributed time evolution rule for such networked ... More
Chromatic number, induced cycles, and non-separating cyclesNov 03 2012Oct 05 2016We show that connected graphs with large chromatic number must contain a large number of induced cycles, and that 3-connected graphs with large clique minor contains many induced non-separating cycles, which are known to generate the cycle space of any ... More
Sparse Laplacian Shrinkage with the Graphical Lasso Estimator for Regression ProblemsApr 09 2019This paper considers a high-dimensional linear regression problem where there are complex correlation structures among predictors. We propose a graph-constrained regularization procedure, named Sparse Laplacian Shrinkage with the Graphical Lasso Estimator ... More
Deep learning tutorial for denoisingOct 27 2018We herein introduce deep learning to seismic noise attenuation. Compared with traditional seismic noise attenuation algorithms that depend on signal models and their corresponding prior assumptions, a deep neural network is trained based on a large training ... More
Universal generation of the cylinder homomorphism of cubic hypersurfacesOct 29 2018Dec 16 2018In this article, we prove that the Chow group of algebraic cycles of a smooth cubic hypersurface $X$ over an arbitrary field $k$ is generated, via the natural cylinder homomorphism, by the algebraic cycles of its Fano variety of lines $F(X)$, under an ... More
A Local-Global LDA Model for Discovering Geographical Topics from Social MediaJul 20 2016Micro-blogging services can track users' geo-locations when users check-in their places or use geo-tagging which implicitly reveals locations. This "geo tracking" can help to find topics triggered by some events in certain regions. However, discovering ... More
A comparative study on nonlocal diffusion operators related to the fractional LaplacianNov 18 2017In this paper, we study four nonlocal diffusion operators, including the fractional Laplacian, spectral fractional Laplacian, regional fractional Laplacian, and peridynamic operator. These operators represent the infinitesimal generators of different ... More
UA-DETRAC: A New Benchmark and Protocol for Multi-Object Detection and TrackingNov 13 2015Sep 04 2016In recent years, numerous effective multi-object tracking (MOT) methods are developed because of the wide range of applications. Existing performance evaluations of MOT methods usually separate the object tracking step from the object detection step by ... More
Power-Based Direction-of-Arrival Estimation Using a Single Multi-Mode AntennaJun 29 2017Jan 30 2018Phased antenna arrays are widely used for direction-of-arrival (DoA) estimation. For low-cost applications, signal power or received signal strength indicator (RSSI) based approaches can be an alternative. However, they usually require multiple antennas, ... More
Layered Image Compression using Scalable Auto-encoderApr 01 2019This paper presents a novel convolutional neural network (CNN) based image compression framework via scalable auto-encoder (SAE). Specifically, our SAE based deep image codec consists of hierarchical coding layers, each of which is an end-to-end optimized ... More
A Group Variational Transformation Neural Network for Fractional Interpolation of Video CodingJun 19 2018Motion compensation is an important technology in video coding to remove the temporal redundancy between coded video frames. In motion compensation, fractional interpolation is used to obtain more reference blocks at sub-pixel level. Existing video coding ... More
Exploiting Image Local And Nonlocal Consistency For Mixed Gaussian-Impulse Noise RemovalAug 18 2012Most existing image denoising algorithms can only deal with a single type of noise, which violates the fact that the noisy observed images in practice are often suffered from more than one type of noise during the process of acquisition and transmission. ... More
Uniform generation of RNA-RNA interaction structures of fixed topological genusNov 04 2013Apr 12 2014Interacting RNA complexes are studied via bicellular maps using a filtration via their topological genus. Our main result is a new bijection for RNA-RNA interaction structures and linear time uniform sampling algorithm for RNA complexes of fixed topological ... More
Ferromagnetism in diluted magnetic semiconductor quantum dot arrays embedded in semiconductorsOct 29 2002We present an Anderson-type model Hamiltonian with exchange coupling between the localized spins and the confined holes in the quantum dots to study the ferromagnetism in diluted magnetic semiconductor (DMS) quantum dot arrays embedded in semiconductors. ... More
Quantitative MRI: Absolute T1, T2 and Proton Density Parameters from Deep LearningJun 19 2018Quantitative MRI is highly desirable in terms of intrinsic tissue parameters such as T1, T2 and proton density. This approach promises to minimize diagnostic variability and differentiate normal and pathological tissues by comparing tissue parameters ... More
Network-Connected UAV: 3D System Modeling and Coverage Performance AnalysisJan 19 2019Apr 25 2019With growing popularity, unmanned aerial vehicles (UAVs) are pivotally extending conventional terrestrial Internet of Things (IoT) into the sky. To enable high-performance two-way communications of UAVs with their ground pilots/users, cellular network-connected ... More
Gradient Descent Maximizes the Margin of Homogeneous Neural NetworksJun 13 2019Recent works on implicit regularization have shown that gradient descent converges to the max-margin direction for logistic regression with one-layer or multi-layer linear networks. In this paper, we generalize this result to homogeneous neural networks, ... More
Strong contraction mapping and topological non-convex optimizationJul 08 2018Mar 22 2019The strong contraction mapping, a self-mapping that the range is always a subset of the domain, admits a unique fixed-point which can be pinned down by the iteration of the mapping. We introduce a topological non-convex optimization method as an application ... More
Learning color space adaptation from synthetic to real images of cirrus cloudsOct 24 2018Training on synthetic data is becoming popular in vision due to the convenient acquisition of accurate pixel-level labels. But the domain gap between synthetic and real images significantly degrades the performance of the trained model. We propose a color ... More
Evaluating Complex Task through Crowdsourcing: Multiple Views ApproachMar 30 2017With the popularity of massive open online courses, grading through crowdsourcing has become a prevalent approach towards large scale classes. However, for getting grades for complex tasks, which require specific skills and efforts for grading, crowdsourcing ... More
Network-Connected UAV: 3D System Modeling and Coverage Performance AnalysisJan 19 2019With growing popularity, unmanned aerial vehicles (UAVs) are pivotally extending conventional terrestrial Internet of Things (IoT) into the sky. To enable high-performance two-way communications of UAVs with their ground pilots/users, cellular network-connected ... More
Median Binary-Connect Method and a Binary Convolutional Neural Nework for Word RecognitionNov 07 2018We propose and study a new projection formula for training binary weight convolutional neural networks. The projection formula measures the error in approximating a full precision (32 bit) vector by a 1-bit vector in the l_1 norm instead of the standard ... More
Blocking Probability and Spatial Throughput Characterization for Cellular-Enabled UAV Network with Directional AntennaOct 28 2017The past few years have witnessed a tremendous increase on the use of unmanned aerial vehicles (UAVs) in civilian applications, which call for high-performance communication between UAVs and their ground clients, especially when they are densely deployed. ... More
Road Segmentation Using CNN with GRUApr 14 2018This paper presents an accurate and fast algorithm for road segmentation using convolutional neural network (CNN) and gated recurrent units (GRU). For autonomous vehicles, road segmentation is a fundamental task that can provide the drivable area for ... More
A High Fraction of Double-peaked Narrow Emission Lines in Powerful Active Galactic NucleiJun 21 2016Aug 09 2016One percent of redshift z~0.1 Active Galactic Nuclei (AGNs) show velocity splitting of a few hundred km/s in the narrow emission lines in spatially integrated spectra. Such line profiles have been found to arise from the bulk motion of ionized gas clouds ... More
Word and Document Embeddings based on Neural Network ApproachesNov 18 2016Data representation is a fundamental task in machine learning. The representation of data affects the performance of the whole machine learning system. In a long history, the representation of data is done by feature engineering, and researchers aim at ... More
Wavelet-Based Semantic Features for Hyperspectral Signature DiscriminationFeb 11 2016Apr 08 2016Hyperspectral signature classification is a quantitative analysis approach for hyperspectral imagery which performs detection and classification of the constituent materials at the pixel level in the scene. The classification procedure can be operated ... More
Application of gold in the field of heterogeneous catalysisNov 09 2014Gold has been long thought as an inert metal which finds most of its use in jewelry and monetary exchange. However, catalysis by gold has rapidly become a hot topic in chemistry ever since Haruta and Hutchings found gold to be an extraordinary good heterogeneous ... More
The coherent structure of the energy cascade in shear turbulenceApr 04 2019Apr 25 2019The cascade of energy in turbulent flows, i.e., the transfer of kinetic energy from large to small flow motions (forward cascade) or vice versa (backward cascade), is the cornerstone of most theories and models of turbulence since the 1940s. Yet, understanding ... More
A brief introduction to giant magnetoresistanceDec 21 2014Giant magnetoresistance (GMR) is a quantum mechanical magnetoresistance effect observed in thin film structures composed of alternating ferromagnetic and nonmagnetic layers. The effect manifests itself as a significant decrease (typically 10-80%) in electrical ... More
StrongChain: Transparent and Collaborative Proof-of-Work ConsensusMay 23 2019Bitcoin is the most successful cryptocurrency so far. This is mainly due to its novel consensus algorithm, which is based on proof-of-work combined with a cryptographically-protected data structure and a rewarding scheme that incentivizes nodes to participate. ... More
Tunneling magnetoresistance in diluted magnetic semiconductor tunnel junctionsJan 15 2001Using the spin-polarized tunneling model and taking into account the basic physics of ferromagnetic semiconductors, we study the temperature dependence of the tunneling magnetoresistance (TMR) in the diluted magnetic semiconductor (DMS) trilayer heterostructure ... More
Remarks on Automorphism and Cohomology of Cyclic CoveringsMay 18 2017We show that the automorphism group of a smooth cyclic covering acts on its cohomology faithfully with a few well known exceptions. Firstly, we prove the faithfulness of the action in characteristic zero. The main ingredients of the proof are equivariant ... More
The phase structure of asymmetric ballistic annihilationNov 20 2018Mar 24 2019In ballistic annihilation, particles are placed throughout the real line with independent spacings and each is assigned a velocity. The particles then move at their assigned velocity and annihilate upon colliding. We develop a framework based on a mass ... More
On high order accurate schemes for space fractional diffusion equations with variable coefficientsAug 29 2016Sep 29 2016We study high order schemes for spatial fractional differential equations with variable coefficients. Approximations of fractional derivatives basing on the weighted and shifted Gr\"unwald-Letnikov formulas and weighted and shifted Lubich formulas are ... More
The phase structure of asymmetric ballistic annihilationNov 20 2018Feb 03 2019In ballistic annihilation, particles are placed throughout the real line with independent spacings and each is assigned a velocity. The particles then move at their assigned velocity and annihilate upon colliding. We develop a framework based on a mass ... More
TTR-Based Rewards for Reinforcement Learning with Implicit Model PriorsMar 23 2019Model-free reinforcement learning (RL) provides an attractive approach for learning control policies directly in high dimensional state spaces. However, many goal-oriented tasks involving sparse rewards remain difficult to solve with state-of-the-art ... More
Harnack and Shift Harnack Inequalities for Degenerate (Functional) SPDEs with Singular DriftsApr 06 2019The existence and uniqueness of the mild solutions for a class of degenerate functional SPDEs are obtained, where the drift is assumed to be H\"{o}lder-Dini continuous. Moreover, the non-explosion of the solution is proved under some reasonable conditions. ... More
On high order accurate schemes for space fractional diffusion equations with variable coefficientsAug 29 2016Oct 14 2016We study high order schemes for spatial fractional differential equations with variable coefficients. Approximations of fractional derivatives basing on the weighted and shifted Gr\"unwald-Letnikov formulas and weighted and shifted Lubich formulas are ... More
On the noise prediction for serrated leading edgesJun 14 2017An analytical model is developed for the prediction of noise radiated by an aerofoil with leading edge serration in a subsonic turbulent stream. The model makes use of the Fourier Expansion and Schwarzschild techniques in order to solve a set of coupled ... More
Experimental validation of the hybrid scattering model of installed jet noiseJul 31 2018Jet installation causes jet noise to be amplified significantly at low frequencies and its physical mechanism must be understood to develop effective aircraft noise reduction strategies. A hybrid semi-empirical prediction model has recently been developed ... More
0-1 matrices with zero trace whose squares are 0-1 matricesMar 04 2018In this paper, we determine the maximum number of nonzero entries in 0-1 matrices of order $n$ with zero trace whose squares are 0-1 matrices when $n\ge 8$. The extremal matrices attaining this maximum number are also characterized.
AMR Parsing as Graph Prediction with Latent AlignmentMay 14 2018Abstract meaning representations (AMRs) are broad-coverage sentence-level semantic representations. AMRs represent sentences as rooted labeled directed acyclic graphs. AMR parsing is challenging partly due to the lack of annotated alignments between nodes ... More
Extremal digraphs avoiding an orientation of $C_4$Feb 09 2018Let $P_{2,2}$ be the orientation of $C_4$ which consists of two 2-paths with the same initial and terminal vertices. In this paper, we determine the maximum size of $P_{2,2}$-free digraphs of order $n$ as well as the extremal digraphs attaining the maximum ... More
A Fast 2-Approximation Algorithm for Guarding Orthogonal TerrainsMay 11 2016May 17 2016Terrain Guarding Problem(TGP), which is known to be NP-complete, asks to find a smallest set of guard locations on a terrain $T$ such that every point on $T$ is visible by a guard. Here, we study this problem on 1.5D orthogonal terrains where the edges ... More
Residual Attention based Network for Hand Bone Age AssessmentDec 21 2018Computerized automatic methods have been employed to boost the productivity as well as objectiveness of hand bone age assessment. These approaches make predictions according to the whole X-ray images, which include other objects that may introduce distractions. ... More
Weak contraction map and topological non-convex optimizationJul 08 2018Nov 11 2018The definition of weak contraction map and the existence and uniqueness of the fixed-point of weak contraction map is discussed. A stochastic contour-based optimization method based on weak contraction map is proposed to achieve global minimum convergence. ... More
The temporal evolution of the energy flux across scales in homogeneous turbulenceMay 01 2015Nov 13 2015A temporal study of energy transfer across length scales is performed in 3D numerical simulations of homogeneous shear flow and isotropic turbulence. The average time taken by perturbations in the energy flux to travel between scales is measured and shown ... More
Optimal Stochastic Delivery Planning in Full-Truckload and Less-Than-Truckload DeliveryFeb 04 2018With an increasing demand from emerging logistics businesses, Vehicle Routing Problem with Private fleet and common Carrier (VRPPC) has been introduced to manage package delivery services from a supplier to customers. However, almost all of existing studies ... More
Semi-Supervised Endmember Identification In Nonlinear Spectral Mixtures Via Semantic RepresentationJan 03 2017This paper proposes a new hyperspectral unmixing method for nonlinearly mixed hyperspectral data using a semantic representation in a semi-supervised fashion, assuming the availability of a spectral reference library. Existing semi-supervised unmixing ... More
Image Restoration Using Joint Statistical Modeling in Space-Transform DomainMay 11 2014This paper presents a novel strategy for high-fidelity image restoration by characterizing both local smoothness and nonlocal self-similarity of natural images in a unified statistical manner. The main contributions are three-folds. First, from the perspective ... More
Curriculum Loss: Robust Learning and Generalization against Label CorruptionMay 24 2019Generalization is vital important for many deep network models. It becomes more challenging when high robustness is required for learning with noisy labels. The 0-1 loss has monotonic relationship between empirical adversary (reweighted) risk, and it ... More
Approximately Optimal Auctions for Selling Privacy when Costs are Correlated with DataApr 18 2012We consider a scenario in which a database stores sensitive data of users and an analyst wants to estimate statistics of the data. The users may suffer a cost when their data are used in which case they should be compensated. The analyst wishes to get ... More
Temporal stability analysis of jets of lobed geometryOct 17 2018A 2D temporal incompressible stability analysis is carried out for lobed jets. The jet base flow is assumed to be parallel and of a vortex-sheet type. The eigenfunctions of this simplified stability problem are expanded using the eigenfunctions of a round ... More
Curriculum Loss: Robust Learning and Generalization against Label CorruptionMay 24 2019May 28 2019Generalization is vital important for many deep network models. It becomes more challenging when high robustness is required for learning with noisy labels. The 0-1 loss has monotonic relationship between empirical adversary (reweighted) risk, and it ... More
Four-Dimensional Discrete-time Lotka-Volterra Models with an Application to EcologyNov 26 2012This paper presents a study of the two-predators-two-preys discrete-time Lotka-Volterra model with self- inhibition terms for preys with direct applications to ecological problems. Parameters in the model are modified so that each of them has its own ... More
Rapid noise prediction models for serrated leading and trailing edgesJun 06 2019Leading- and trailing-edge serrations have been widely used to reduce the leading- and trailing-edge noise in applications such as contra-rotating fans and large wind turbines. Recent studies show that these two noise problems can be modelled analytically ... More
The Intrinsic Far-infrared Continua of Type-1 QuasarsApr 23 2017May 25 2017The range of currently proposed active galactic nucleus (AGN) far-infrared templates results in uncertainties in retrieving host galaxy information from infrared observations and also undermines constraints on the outer part of the AGN torus. We discuss ... More
Polar Dust, Nuclear Obscuration and IR SED Diversity in Type-1 AGNsSep 10 2018Despite the hypothesized similar face-on viewing angles, the infrared emission of type-1 AGNs has diverse spectral energy distribution (SED) shapes that deviate substantially from the well-characterized quasar templates. Motivated by the commonly-seen ... More
Mode Seeking Generative Adversarial Networks for Diverse Image SynthesisMar 13 2019Apr 03 2019Most conditional generation tasks expect diverse outputs given a single conditional context. However, conditional generative adversarial networks (cGANs) often focus on the prior conditional information and ignore the input noise vectors, which contribute ... More