arXiv subjects

Li Wang

Publications and source records attributed to Li Wang.

At least 487 records · Page 27Linked to original sources

arXiv2019

Disentangling Dynamics and Returns: Value Function Decomposition with Future Prediction

Value functions are crucial for model-free Reinforcement Learning (RL) to obtain a policy implicitly or guide the policy updates. Value estimation heavily depends on the stochasticity of environmental dynamics and the quality of reward signals. In this paper, we propose a two-step understanding of value estimation from the perspective of future prediction, through decomposing the value function into a reward-independent future dynamics part and a policy-independent trajectory return part. We then derive a practical deep RL algorithm from the above decomposition, consisting of a convolutional trajectory representation model, a conditional variational dynamics model to predict the expected representation of future trajectory and a convex trajectory return model that maps a trajectory representation to its return. Our algorithm is evaluated in MuJoCo continuous control tasks and shows superior results under both common settings and delayed reward settings.

cs.LG↗

arXiv2019

Convolutional Restricted Boltzmann Machine Based-Radiomics for Prediction of Pathological Complete Response to Neoadjuvant Chemotherapy in Breast Cancer

We proposed a novel convolutional restricted Boltzmann machine CRBM-based radiomic method for predicting pathologic complete response (pCR) to neoadjuvant chemotherapy treatment (NACT) in breast cancer. The method consists of extracting semantic features from CRBM network, and pCR prediction. It was evaluated on the dynamic contrast-enhanced magnetic resonance imaging (DCE-MRI) data of 57 patients and using the area under the receiver operating characteristic curve (AUC). Traditional radiomics features and the semantic features learned from CRBM network were extracted from the images acquired before and after the administration of NACT. After the feature selection, the support vector machine (SVM), logistic regression (LR) and random forest (RF) were trained to predict the pCR status. Compared to traditional radiomic methods, the proposed CRBM-based radiomic method yielded an AUC of 0.92 for the prediction with the images acquired before and after NACT, and an AUC of 0.87 for the pretreatment prediction, which was increased by about 38%. The results showed that the CRBM-based radiomic method provided a potential means for accurately predicting the pCR to NACT in breast cancer before the treatment, which is very useful for making more appropriate and personalized treatment regimens.

eess.IV↗

arXiv2019

Electron-positron pair production in an oscillating Sauter potential

Electron-positron pair production in an oscillating Sauter potential is investigated in the framework of the computational quantum field theory. It is found that for a Sauter potential well with oscillating width and depth simultaneously, the phase difference between them has a great impact on the number of created electron-positron pairs. Optimal values of the phase difference corresponding to different oscillation frequencies are obtained. The optimal phase difference has a strong nonlinear dependence on oscillating frequency. When the potential well changes slowly, our results can be explained by the instantaneous bound states. For the higher frequency case, however, multiphoton effect is enhanced and the Pauli blocking effect has a strong inhibitory effect on pair creations. These results can provide a theoretical reference for the experimental creation of the electron-positron pairs.

quant-ph↗

arXiv2019

Observation of Kerr nonlinearity and Kerr-like nonlinearity induced by terahertz generation in LiNbO$_3$

Femtosecond laser pulses interacting with LiNbO3 has been widely applied in intense terahertz generation, excitation and imaging of phonon polaritons and modulating light. It is of great importance to study the nonlinearities induced by femtosecond laser pulses in LiNbO3. Here, we demonstrate that both Kerr nonlinearity and Kerr-like nonlinearity induced by terahertz wave occur under the interaction between laser pulses and LiNbO3 using optical pump-probe techniques. We show that the contribution of Kerr-like nonlinearity to pump-probe spectra varies with the interaction length because of the phase mismatching between terahertz wave and laser pulse. We also observed the excitation of the low frequency phonon polaritons. Experimental results are in consistent with theoretical calculation.

physics.optics↗

arXiv2019

Shrinking the Upper Confidence Bound: A Dynamic Product Selection Problem for Urban Warehouses

The recent rising popularity of ultra-fast delivery services on retail platforms fuels the increasing use of urban warehouses, whose proximity to customers makes fast deliveries viable. The space limit in urban warehouses poses a problem for the online retailers: the number of products (SKUs) they carry is no longer "the more, the better", yet it can still be significantly large, reaching hundreds or thousands in a product category. In this paper, we study algorithms for dynamically identifying a large number of products (i.e., SKUs) with top customer purchase probabilities on the fly, from an ocean of potential products to offer on retailers' ultra-fast delivery platforms. We distill the product selection problem into a semi-bandit model with linear generalization. There are in total $N$ different arms, each with a feature vector of dimension $d$. The player pulls $K$ arms in each period and observes the bandit feedback from each of the pulled arms. We focus on the setting where $K$ is much greater than the number of total time periods $T$ or the dimension of product features $d$. We first analyze a standard UCB algorithm and show its regret bound can be expressed as the sum of a $T$-independent part $\tilde O(K d^{3/2})$ and a $T$-dependent part $\tilde O(d\sqrt{KT})$, which we refer to as "fixed cost" and "variable cost" respectively. To reduce the fixed cost for large $K$ values, we propose a novel online learning algorithm, which iteratively shrinks the upper confidence bounds within each period, and show its fixed cost is reduced by a factor of $d$ to $\tilde O(K \sqrt{d})$. Moreover, we test the algorithms on an industrial dataset from Alibaba Group. Experimental results show that our new algorithm reduces the total regret of the standard UCB algorithm by at least 10%.

math.OC↗

arXiv2019

An immersed boundary method for fluid--structure--acoustics interactions involving large deformations and complex geometries

This paper presents an immersed boundary (IB) method for fluid--structure--acoustics interactions involving large deformations and complex geometries. In this method, the fluid dynamics is solved by a finite difference method where the temporal, viscous and convective terms are respectively discretized by the third-order Runge-Kutta scheme, the fourth-order central difference scheme and a fifth-order W/TENO (Weighted/Targeted Essentially Non-oscillation) scheme. Without loss of generality, a nonlinear flexible plate is considered here, and is solved by a finite element method based on the absolute nodal coordinate formulation. The no-slip boundary condition at the fluid--structure interface is achieved by using a diffusion-interface penalty IB method. With the above proposed method, the aeroacoustics field generated by the moving boundaries and the associated flows are inherently solved. In order to validate and verify the current method, several benchmark cases are conducted: acoustic waves scattered from a stationary cylinder in a quiescent flow, sound generation by a stationary and a rotating cylinder in a uniform flow, sound generation by an insect in hovering flight, deformation of a red blood cell induced by acoustic waves and acoustic waves scattered by a stationary sphere. The comparison of the sound scattered by a cylinder shows that the present IB--WENO scheme, a simple approach, has an excellent performance which is even better than the implicit IB--lattice Boltzmann method. For the sound scattered by a sphere, the IB--TENO scheme has a lower dissipation compared with the IB--WENO scheme. Applications of this technique to model fluid-structure-acoustics interactions of flapping foils mimicking an insect wing section during forward flight and flapping foil energy harvester are also presented, considering the effects of foil shape and flexibility.

physics.flu-dyn↗

arXiv2019

Giant Interfacial Thermal Resistance Arising From Materials With Mismatched Phonon Structures

Previous researches only reported very small interfacial thermal resistances at room temperature due to limitations in sample combinations and methods. Taking cognizance of the importance of mismatched phonon structures, we report values up to $2*10^{-4}W^{-1}m^{2}K$, thousand times larger than highest values reported to date. This enables substantial tuning of the thermal conductivity in composites, and does not constrain other characteristics. Our findings inspire new design strategies, for heat control in integrated circuits and thermoelectric composites, that harness thermal transport at interfaces.

cond-mat.mtrl-sci↗

arXiv2019

Facial Feature Embedded CycleGAN for VIS-NIR Translation

VIS-NIR face recognition remains a challenging task due to the distinction between spectral components of two modalities and insufficient paired training data. Inspired by the CycleGAN, this paper presents a method aiming to translate VIS face images into fake NIR images whose distributions are intended to approximate those of true NIR images, which is achieved by proposing a new facial feature embedded CycleGAN. Firstly, to learn the particular feature of NIR domain while preserving common facial representation between VIS and NIR domains, we employ a general facial feature extractor (FFE) to replace the encoder in the original generator of CycleGAN. For implementing the facial feature extractor, herein the MobileFaceNet is pretrained on a VIS face database, and is able to extract effective features. Secondly, the domain-invariant feature learning is enhanced by considering a new pixel consistency loss. Lastly, we establish a new WHU VIS-NIR database which varies in face rotation and expressions to enrich the training data. Experimental results on the Oulu-CASIA NIR-VIS database and the WHU VIS-NIR database show that the proposed FFE-based CycleGAN (FFE-CycleGAN) outperforms state-of-the-art VIS-NIR face recognition methods and achieves 96.5\% accuracy.

cs.CV↗

arXiv2019

A Comparison Study of Credit Card Fraud Detection: Supervised versus Unsupervised

Credit card has become popular mode of payment for both online and offline purchase, which leads to increasing daily fraud transactions. An Efficient fraud detection methodology is therefore essential to maintain the reliability of the payment system. In this study, we perform a comparison study of credit card fraud detection by using various supervised and unsupervised approaches. Specifically, 6 supervised classification models, i.e., Logistic Regression (LR), K-Nearest Neighbors (KNN), Support Vector Machines (SVM), Decision Tree (DT), Random Forest (RF), Extreme Gradient Boosting (XGB), as well as 4 unsupervised anomaly detection models, i.e., One-Class SVM (OCSVM), Auto-Encoder (AE), Restricted Boltzmann Machine (RBM), and Generative Adversarial Networks (GAN), are explored in this study. We train all these models on a public credit card transaction dataset from Kaggle website, which contains 492 frauds out of 284,807 transactions. The labels of the transactions are used for supervised learning models only. The performance of each model is evaluated through 5-fold cross validation in terms of Area Under the Receiver Operating Curves (AUROC). Within supervised approaches, XGB and RF obtain the best performance with AUROC = 0.989 and AUROC = 0.988, respectively. While for unsupervised approaches, RBM achieves the best performance with AUROC = 0.961, followed by GAN with AUROC = 0.954. The experimental results show that supervised models perform slightly better than unsupervised models in this study. Anyway, unsupervised approaches are still promising for credit card fraud transaction detection due to the insufficient annotation and the data imbalance issue in real-world applications.

cs.LG↗

arXiv2019

Attraction centers and PT-symmetric delta-functional dipoles in critical and supercritical self-focusing media

We introduce a model based on the one-dimensional nonlinear Schroedinger equation (NLSE) with the critical (quintic) or supercritical self-focusing nonlinearity. We demonstrate that a family of solitons, which are unstable in this setting against collapse, is stabilized by pinning to an attractive defect, that may also include a parity-time (PT)-symmetric gain-loss component. The model can be realized as a planar waveguide in optics, and in a super-Tonks-Girardeau bosonic gas. For the attractive defect with the delta-functional profile, a full family of the pinned solitons is found in an exact form. In the absence of the gain-loss term, the solitons' stability is investigated analytically too, by means of the Vakhitov-Kolokolov criterion; in the presence of the PT-balanced gain and loss, the stability is explored by means of numerical methods. In particular, the entire family of pinned solitons is stable in the quintic medium if the gain-loss term is absent. A stability region for the pinned solitons persists in the model with an arbitrarily high power of the self-focusing nonlinearity. A weak gain-loss component gives rise to alternations of stability and instability in the system's parameter plane. Those solitons which are unstable under the action of the supercritical self-attraction are destroyed by the collapse. If the self-attraction-driven instability is weak and the gain-loss term is present, unstable solitons spontaneously transform into localized breathers. The same outcome may be caused by a combination of the critical nonlinearity with the gain and loss. Instability of the solitons is also possible when the PT-symmetric gain-loss term is added to the subcritical nonlinearity. The system with self-repulsive nonlinearity is briefly considered too, producing completely stable families of pinned localized states.

nlin.PS↗

arXiv2019

Learning Actor Relation Graphs for Group Activity Recognition

Modeling relation between actors is important for recognizing group activity in a multi-person scene. This paper aims at learning discriminative relation between actors efficiently using deep models. To this end, we propose to build a flexible and efficient Actor Relation Graph (ARG) to simultaneously capture the appearance and position relation between actors. Thanks to the Graph Convolutional Network, the connections in ARG could be automatically learned from group activity videos in an end-to-end manner, and the inference on ARG could be efficiently performed with standard matrix operations. Furthermore, in practice, we come up with two variants to sparsify ARG for more effective modeling in videos: spatially localized ARG and temporal randomized ARG. We perform extensive experiments on two standard group activity recognition datasets: the Volleyball dataset and the Collective Activity dataset, where state-of-the-art performance is achieved on both datasets. We also visualize the learned actor graphs and relation features, which demonstrate that the proposed ARG is able to capture the discriminative relation information for group activity recognition.

cs.CV↗

arXiv2019

Road Crack Detection Using Deep Convolutional Neural Network and Adaptive Thresholding

Crack is one of the most common road distresses which may pose road safety hazards. Generally, crack detection is performed by either certified inspectors or structural engineers. This task is, however, time-consuming, subjective and labor-intensive. In this paper, we propose a novel road crack detection algorithm based on deep learning and adaptive image segmentation. Firstly, a deep convolutional neural network is trained to determine whether an image contains cracks or not. The images containing cracks are then smoothed using bilateral filtering, which greatly minimizes the number of noisy pixels. Finally, we utilize an adaptive thresholding method to extract the cracks from road surface. The experimental results illustrate that our network can classify images with an accuracy of 99.92%, and the cracks can be successfully extracted from the images using our proposed thresholding algorithm.

cs.CV↗

arXiv2019

FRNET: Flattened Residual Network for Infant MRI Skull Stripping

Skull stripping for brain MR images is a basic segmentation task. Although many methods have been proposed, most of them focused mainly on the adult MR images. Skull stripping for infant MR images is more challenging due to the small size and dynamic intensity changes of brain tissues during the early ages. In this paper, we propose a novel CNN based framework to robustly extract brain region from infant MR image without any human assistance. Specifically, we propose a simplified but more robust flattened residual network architecture (FRnet). We also introduce a new boundary loss function to highlight ambiguous and low contrast regions between brain and non-brain regions. To make the whole framework more robust to MR images with different imaging quality, we further introduce an artifact simulator for data augmentation. We have trained and tested our proposed framework on a large dataset (N=343), covering newborns to 48-month-olds, and obtained performance better than the state-of-the-art methods in all age groups.

cs.CV↗

arXiv2019

Spherical U-Net on Cortical Surfaces: Methods and Applications

Convolutional Neural Networks (CNNs) have been providing the state-of-the-art performance for learning-related problems involving 2D/3D images in Euclidean space. However, unlike in the Euclidean space, the shapes of many structures in medical imaging have a spherical topology in a manifold space, e.g., brain cortical or subcortical surfaces represented by triangular meshes, with large inter-subject and intrasubject variations in vertex number and local connectivity. Hence, there is no consistent neighborhood definition and thus no straightforward convolution/transposed convolution operations for cortical/subcortical surface data. In this paper, by leveraging the regular and consistent geometric structure of the resampled cortical surface mapped onto the spherical space, we propose a novel convolution filter analogous to the standard convolution on the image grid. Accordingly, we develop corresponding operations for convolution, pooling, and transposed convolution for spherical surface data and thus construct spherical CNNs. Specifically, we propose the Spherical U-Net architecture by replacing all operations in the standard U-Net with their spherical operation counterparts. We then apply the Spherical U-Net to two challenging and neuroscientifically important tasks in infant brains: cortical surface parcellation and cortical attribute map development prediction. Both applications demonstrate the competitive performance in the accuracy, computational efficiency, and effectiveness of our proposed Spherical U-Net, in comparison with the state-of-the-art methods.

cs.CV↗

arXiv2019

Broadband vibrational sum-frequency generation spectrometer at 100 kHz in the 950-1750 cm$^{-1}$ spectral range utilizing a LiGaS$_{2}$ optical parametric amplifier

We present a 100 kHz broadband vibrational sum-frequency generation (VSFG) spectrometer operating in the 5.5-10.5 $μ$m (950-1750 cm$^{-1}$) wavelength range. The mid-infrared beam of the system is obtained from a collinear, type-I LiGaS$_{2}$ crystal based optical parametric amplifier seeded by a supercontinuum and pumped directly by 180 fs, ~32 $μ$J, 1.03 $μ$m pulses from an Yb:KGd(WO$_{4}$)$_{2}$ laser system. Up to 0.5 $μ$J mid-infrared pulses with durations below 100 fs were obtained after dispersion compensation utilizing bulk materials. We demonstrate the utility of the spectrometer by recording high-resolution, low-noise vibrational spectra of Langmuir-Blodgett supported lipid monolayers on CaF$_{2}$. The presented VSFG spectrometer scheme offers superior signal-to-noise ratios and constitutes a high-efficiency, low-cost, easy-to-use alternative to traditional schemes relying on optical parametric amplification followed by difference frequency generation.

physics.ins-det↗

arXiv2019

CODA: Counting Objects via Scale-aware Adversarial Density Adaption

Recent advances in crowd counting have achieved promising results with increasingly complex convolutional neural network designs. However, due to the unpredictable domain shift, generalizing trained model to unseen scenarios is often suboptimal. Inspired by the observation that density maps of different scenarios share similar local structures, we propose a novel adversarial learning approach in this paper, i.e., CODA (\emph{Counting Objects via scale-aware adversarial Density Adaption}). To deal with different object scales and density distributions, we perform adversarial training with pyramid patches of multi-scales from both source- and target-domain. Along with a ranking constraint across levels of the pyramid input, consistent object counts can be produced for different scales. Extensive experiments demonstrate that our network produces much better results on unseen datasets compared with existing counting adaption models. Notably, the performance of our CODA is comparable with the state-of-the-art fully-supervised models that are trained on the target dataset. Further analysis indicates that our density adaption framework can effortlessly extend to scenarios with different objects. \emph{The code is available at https://github.com/Willy0919/CODA.}

cs.CV↗

arXiv2019

Sub-cycle Electron Dynamics in the Generation of Below Threshold Harmonics

The generation of the below threshold harmonics (BTHs) under different driving laser intensities is investigated. The linearly shifting of photon energy and resonantly enhancement of photon yield of the harmonics from 23rd (H23) to 27th (H27) are found by changing the laser intensity around 18.6 TW/cm2. It is identified that this driving laser intensity dependence is due to the transient ac Stark-shifted resonance between the first excited state and the ground state. With this transient ac Stark, the linearly shifting of the photon energy can be interpreted very well by considering the sub-cycle electron dynamics for BTH generation, which shows that the generation of the BTHs is surprisingly similar to the plateau harmonics.

physics.atom-ph↗

arXiv2019

Enhancement of Spintronic Terahertz Emission via Annealing in Ferromagnetic Heterostructures

We systematically investigate the influence of annealing effect on terahertz (THz) generation from CoFeB based magnetic nanofilms driven by femtosecond laser pulses. Three times enhancement of THz yields are achieved in W/CoFeB through annealing effect, and double boosting is obtained in Pt/CoFeB. The mechanism of annealing effect originates from the increase of hot electron mean free path induced by crystallization, which is experimentally corroborated by THz transmission measurement on time-domain spectroscopy. Comparison studies of the thickness dependent THz efficiency after annealing are also implemented, and we eventually conclude that annealing and thickness optimization are of importance for scaling up THz intensity. Our observations not only deepen understanding of the spintronic THz radiation mechanism but also provide normal platform for high speed spintronic opto-electronic devices.

physics.app-ph↗