arXiv subjects

Wei Sun

Publications and source records attributed to Wei Sun.

At least 361 records · Page 20Linked to original sources

arXiv2021

Learned Interpretable Residual Extragradient ISTA for Sparse Coding

Recently, the study on learned iterative shrinkage thresholding algorithm (LISTA) has attracted increasing attentions. A large number of experiments as well as some theories have proved the high efficiency of LISTA for solving sparse coding problems. However, existing LISTA methods are all serial connection. To address this issue, we propose a novel extragradient based LISTA (ELISTA), which has a residual structure and theoretical guarantees. In particular, our algorithm can also provide the interpretability for Res-Net to a certain extent. From a theoretical perspective, we prove that our method attains linear convergence. In practice, extensive empirical results verify the advantages of our method.

cs.LG↗

arXiv2021

Model Distillation for Revenue Optimization: Interpretable Personalized Pricing

Data-driven pricing strategies are becoming increasingly common, where customers are offered a personalized price based on features that are predictive of their valuation of a product. It is desirable for this pricing policy to be simple and interpretable, so it can be verified, checked for fairness, and easily implemented. However, efforts to incorporate machine learning into a pricing framework often lead to complex pricing policies which are not interpretable, resulting in slow adoption in practice. We present a customized, prescriptive tree-based algorithm that distills knowledge from a complex black-box machine learning algorithm, segments customers with similar valuations and prescribes prices in such a way that maximizes revenue while maintaining interpretability. We quantify the regret of a resulting policy and demonstrate its efficacy in applications with both synthetic and real-world datasets.

stat.ML↗

arXiv2021

Fisher-Pitman permutation tests based on nonparametric Poisson mixtures with application to single cell genomics

This paper investigates the theoretical and empirical performance of Fisher-Pitman-type permutation tests for assessing the equality of unknown Poisson mixture distributions. Building on nonparametric maximum likelihood estimators (NPMLEs) of the mixing distribution, these tests are theoretically shown to be able to adapt to complicated unspecified structures of count data and also consistent against their corresponding ANOVA-type alternatives; the latter is a result in parallel to classic claims made by Robinson (Robinson, 1973). The studied methods are then applied to a single-cell RNA-seq data obtained from different cell types from brain samples of autism subjects and healthy controls; empirically, they unveil genes that are differentially expressed between autism and control subjects yet are missed using common tests. For justifying their use, rate optimality of NPMLEs is also established in settings similar to nonparametric Gaussian (Wu and Yang, 2020a) and binomial mixtures (Tian et al., 2017; Vinayak et al., 2019).

stat.ME↗

arXiv2021

Deep Learning based Full-reference and No-reference Quality Assessment Models for Compressed UGC Videos

In this paper, we propose a deep learning based video quality assessment (VQA) framework to evaluate the quality of the compressed user's generated content (UGC) videos. The proposed VQA framework consists of three modules, the feature extraction module, the quality regression module, and the quality pooling module. For the feature extraction module, we fuse the features from intermediate layers of the convolutional neural network (CNN) network into final quality-aware feature representation, which enables the model to make full use of visual information from low-level to high-level. Specifically, the structure and texture similarities of feature maps extracted from all intermediate layers are calculated as the feature representation for the full reference (FR) VQA model, and the global mean and standard deviation of the final feature maps fused by intermediate feature maps are calculated as the feature representation for the no reference (NR) VQA model. For the quality regression module, we use the fully connected (FC) layer to regress the quality-aware features into frame-level scores. Finally, a subjectively-inspired temporal pooling strategy is adopted to pool frame-level scores into the video-level score. The proposed model achieves the best performance among the state-of-the-art FR and NR VQA models on the Compressed UGC VQA database and also achieves pretty good performance on the in-the-wild UGC VQA databases.

eess.IV↗

arXiv2021

Proactive Scheduling of Hydrogen Systems for Resilience Enhancement of Distribution Networks

Recent advances in smart grid technologies bring opportunities to better control the modern and complex power grids with renewable integration. The operation of power systems, especially distribution network (DN), is facing with preeminent challenges from cyber-physical-human (CPH) threats and natural disasters. In order to provide better response against threats and improve the resilience of power grid, proactive plans and operational schemes are required by system operators to minimize the damages caused by CPH threats. To that end, this paper proposes a proactive plan for DN operation by using hydrogen (H2) systems to enhance the resilience through cost-effective long-term energy storage. Unlike batteries, H2 energy can be stored in the storage tanks days before the extreme event, and transformed into power by fuel cell units in the post-event time to reduce load curtailment caused by CPH threats. The proposed framework is validated by testing on 33-node test feeder. Simulation results demonstrate that H2 systems can improve the resilience of DN during $N-m$ outages lasting for more than 10 hours.

eess.SY↗

arXiv2021

Gluons in charmoniumlike states

The mass components of charmoniumlike states are investigated through the decomposition of QCD energy-momentum tensor (EMT) on lattice. The quark mass contribution $\langle H_m\rangle$ and the momentum fraction $\langle x\rangle$ of valence charm quark and antiquark are calculated for conventional $1S,1P,1D$ charmonia and the exotic $1^{-+}$ charmoniumlike state, based on the $N_f=2+1$ gauge configurations generated by the RBC/UKQCD collaboration. It is found that $\langle H_m\rangle$ is close to each other and around 2.0 to 2.2 GeV for these states, which implies that the mass splittings among these states come almost from the gluon contribution of QCD trace anomaly. The $\langle x\rangle$ of the $1^{-+}$ state is only around 0.55, while that in conventional charmonia is around 0.7 to 0.8. This difference manifests that the proportion of light quarks and gluons in the $1^{-+}$ charmoniumlike state is significantly larger than conventional states.

hep-lat↗

arXiv2021

Learning Layout and Style Reconfigurable GANs for Controllable Image Synthesis

With the remarkable recent progress on learning deep generative models, it becomes increasingly interesting to develop models for controllable image synthesis from reconfigurable inputs. This paper focuses on a recent emerged task, layout-to-image, to learn generative models that are capable of synthesizing photo-realistic images from spatial layout (i.e., object bounding boxes configured in an image lattice) and style (i.e., structural and appearance variations encoded by latent vectors). This paper first proposes an intuitive paradigm for the task, layout-to-mask-to-image, to learn to unfold object masks of given bounding boxes in an input layout to bridge the gap between the input layout and synthesized images. Then, this paper presents a method built on Generative Adversarial Networks for the proposed layout-to-mask-to-image with style control at both image and mask levels. Object masks are learned from the input layout and iteratively refined along stages in the generator network. Style control at the image level is the same as in vanilla GANs, while style control at the object mask level is realized by a proposed novel feature normalization scheme, Instance-Sensitive and Layout-Aware Normalization. In experiments, the proposed method is tested in the COCO-Stuff dataset and the Visual Genome dataset with state-of-the-art performance obtained.

cs.CV↗

arXiv2021

Attentive Normalization

In state-of-the-art deep neural networks, both feature normalization and feature attention have become ubiquitous. % with significant performance improvement shown in a vast amount of tasks. They are usually studied as separate modules, however. In this paper, we propose a light-weight integration between the two schema and present Attentive Normalization (AN). Instead of learning a single affine transformation, AN learns a mixture of affine transformations and utilizes their weighted-sum as the final affine transformation applied to re-calibrate features in an instance-specific way. The weights are learned by leveraging channel-wise feature attention. In experiments, we test the proposed AN using four representative neural architectures in the ImageNet-1000 classification benchmark and the MS-COCO 2017 object detection and instance segmentation benchmark. AN obtains consistent performance improvement for different neural architectures in both benchmarks with absolute increase of top-1 accuracy in ImageNet-1000 between 0.5\% and 2.7\%, and absolute increase up to 1.8\% and 2.2\% for bounding box and mask AP in MS-COCO respectively. We observe that the proposed AN provides a strong alternative to the widely used Squeeze-and-Excitation (SE) module. The source codes are publicly available at https://github.com/iVMCL/AOGNet-v2 (the ImageNet Classification Repo) and https://github.com/iVMCL/AttentiveNorm\_Detection (the MS-COCO Detection and Segmentation Repo).

cs.CV↗

arXiv2021

Construction of explicit symplectic integrators in general relativity. III. Reissner-Nordstrom-(anti)-de Sitter black holes

We give a possible splitting method to a Hamiltonian for the description of charged particles moving around the Reissner-Nordstrom-(anti)-de Sitter black hole with an external magnetic field. This Hamiltonian can be separated into six analytical solvable pieces, whose solutions are explicit functions of proper time. In this case, second- and fourth-order explicit symplectic integrators are easily available. They exhibit excellent long-term behavior in maintaining the boundness of Hamiltonian errors regardless of ordered or chaotic orbits if appropriate step-sizes are chosen. Under some circumstances, an increase of positive cosmological constant gives rise to strengthening the extent of chaos from the global phase space; namely, chaos of charged particles occurs easily for the accelerated expansion of the universe. However, an increase of the magnitude of negative cosmological constant does not. The different contributions on chaos are because the cosmological constant acts as a repulsive force in the Reissner-Nordstrom-de Sitter black hole, but an attractive force in the Reissner-Nordstrom-anti-de Sitter black hole.

gr-qc↗

arXiv2021

Deep Consensus Learning

Both generative learning and discriminative learning have recently witnessed remarkable progress using Deep Neural Networks (DNNs). For structured input synthesis and structured output prediction problems (e.g., layout-to-image synthesis and image semantic segmentation respectively), they often are studied separately. This paper proposes deep consensus learning (DCL) for joint layout-to-image synthesis and weakly-supervised image semantic segmentation. The former is realized by a recently proposed LostGAN approach, and the latter by introducing an inference network as the third player joining the two-player game of LostGAN. Two deep consensus mappings are exploited to facilitate training the three networks end-to-end: Given an input layout (a list of object bounding boxes), the generator generates a mask (label map) and then use it to help synthesize an image. The inference network infers the mask for the synthesized image. Then, the latent consensus is measured between the mask generated by the generator and the one inferred by the inference network. For the real image corresponding to the input layout, its mask also is computed by the inference network, and then used by the generator to reconstruct the real image. Then, the data consensus is measured between the real image and its reconstructed image. The discriminator still plays the role of an adversary by computing the realness scores for a real image, its reconstructed image and a synthesized image. In experiments, our DCL is tested in the COCO-Stuff dataset. It obtains compelling layout-to-image synthesis results and weakly-supervised image semantic segmentation results.

cs.CV↗

arXiv2021

Construction of explicit symplectic integrators in general relativity. II. Reissner-Nordstrom black holes

In a previous paper, second- and fourth-order explicit symplectic integrators were designed for a Hamiltonian of the Schwarzschild black hole. Following this work, we continue to trace the possibility of the construction of explicit symplectic integrators for a Hamiltonian of charged particles moving around a Reissner-Nordstrom black hole with an external magnetic field. Such explicit symplectic methods are still available when the Hamiltonian is separated into five independently integrable parts with analytical solutions as explicit functions of proper time. Numerical tests show that the proposed algorithms share the desirable properties in their long-term stability, precision and efficiency for appropriate choices of step sizes. For the applicability of one of the new algorithms, the effects of the black hole's charge, the Coulomb part of the electromagnetic potential and the magnetic parameter on the dynamical behavior are surveyed. Under some circumstances, the extent of chaos gets strong with an increase of the magnetic parameter from a global phase-space structure. No the variation of the black hole's charge but the variation of the Coulomb part is considerably sensitive to affect the regular and chaotic dynamics of particles' orbits. A positive Coulomb part is easier to induce chaos than a negative one.

gr-qc↗

arXiv2021

TeethTap: Recognizing Discrete Teeth Gestures Using Motion and Acoustic Sensing on an Earpiece

Teeth gestures become an alternative input modality for different situations and accessibility purposes. In this paper, we present TeethTap, a novel eyes-free and hands-free input technique, which can recognize up to 13 discrete teeth tapping gestures. TeethTap adopts a wearable 3D printed earpiece with an IMU sensor and a contact microphone behind both ears, which works in tandem to detect jaw movement and sound data, respectively. TeethTap uses a support vector machine to classify gestures from noise by fusing acoustic and motion data, and implements K-Nearest-Neighbor (KNN) with a Dynamic Time Warping (DTW) distance measurement using motion data for gesture classification. A user study with 11 participants demonstrated that TeethTap could recognize 13 gestures with a real-time classification accuracy of 90.9% in a laboratory environment. We further uncovered the accuracy differences on different teeth gestures when having sensors on single vs. both sides. Moreover, we explored the activation gesture under real-world environments, including eating, speaking, walking and jumping. Based on our findings, we further discussed potential applications and practical challenges of integrating TeethTap into future devices.

cs.HC↗

arXiv2021

Construction of Explicit Symplectic Integrators in General Relativity. I. Schwarzschild Black Holes

Symplectic integrators that preserve the geometric structure of Hamiltonian flows and do not exhibit secular growth in energy errors are suitable for the long-term integration of N-body Hamiltonian systems in the solar system. However, the construction of explicit symplectic integrators is frequently difficult in general relativity because all variables are inseparable. Moreover, even if two analytically integrable splitting parts exist in a relativistic Hamiltonian, all analytical solutions are not explicit functions of proper time. Naturally, implicit symplectic integrators, such as the midpoint rule, are applicable to this case. In general, these integrators are numerically more expensive to solve than same-order explicit symplectic algorithms. To address this issue, we split the Hamiltonian of Schwarzschild spacetime geometry into four integrable parts with analytical solutions as explicit functions of proper time. In this manner, second- and fourth-order explicit symplectic integrators can be easily made available. The new algorithms are also useful for modeling the chaotic motion of charged particles around a black hole with an external magnetic field. They demonstrate excellent long-term performance in maintaining bounded Hamiltonian errors and saving computational cost when appropriate proper time steps are adopted.

gr-qc↗

arXiv2021

Sum-Rate Maximization in Distributed Intelligent Reflecting Surfaces-Aided mmWave Communications

In this paper, we focus on the sum-rate optimization in a multi-user millimeter-wave (mmWave) system with distributed intelligent reflecting surfaces (D-IRSs), where a base station (BS) communicates with users via multiple IRSs. The BS transmit beamforming, IRS switch vector, and phase shifts of the IRS are jointly optimized to maximize the sum-rate under minimum user rate, unit-modulus, and transmit power constraints. To solve the resulting non-convex optimization problem, we develop an efficient alternating optimization (AO) algorithm. Specifically, the non-convex problem is converted into three subproblems, which are solved alternatively. The solution to transmit beamforming at the BS and the phase shifts at the IRS are derived by using the successive convex approximation (SCA)-based algorithm, and a greedy algorithm is proposed to design the IRS switch vector. The complexity of the proposed AO algorithm is analyzed theoretically. Numerical results show that the D-IRSs-aided scheme can significantly improve the sum-rate and energy efficiency performance.

eess.SP↗

arXiv2021

Non-uniform Motion Deblurring with Blurry Component Divided Guidance

Blind image deblurring is a fundamental and challenging computer vision problem, which aims to recover both the blur kernel and the latent sharp image from only a blurry observation. Despite the superiority of deep learning methods in image deblurring have displayed, there still exists major challenge with various non-uniform motion blur. Previous methods simply take all the image features as the input to the decoder, which handles different degrees (e.g. large blur, small blur) simultaneously, leading to challenges for sharp image generation. To tackle the above problems, we present a deep two-branch network to deal with blurry images via a component divided module, which divides an image into two components based on the representation of blurry degree. Specifically, two component attentive blocks are employed to learn attention maps to exploit useful deblurring feature representations on both large and small blurry regions. Then, the blur-aware features are fed into two-branch reconstruction decoders respectively. In addition, a new feature fusion mechanism, orientation-based feature fusion, is proposed to merge sharp features of the two branches. Both qualitative and quantitative experimental results show that our method performs favorably against the state-of-the-art approaches.

cs.CV↗

arXiv2021

Laser Cooling of Germanium Semiconductor Nanocrystals

Laser cooling of matter through anti-Stokes photoluminescence, where the emitted frequency of light exceeds that of the impinging laser by virtue of absorption of thermal vibrational energy, has been successfully realized in condensed media, and in particular with rare earth doped systems achieving sub-100K solid state optical refrigeration. Studies suggest that laser cooling in semiconductors has the potential of achieving temperatures down to ~10K and that its direct integration can usher unique high-performance nanostructured semiconductor devices. While laser cooling of nanostructured II-VI semiconductors has been reported recently, laser cooling of indirect bandgap semiconductors such as group IV silicon and germanium remains a major challenge. Here we report on the anomalous observation of dominant anti-Stokes photoluminescence in germanium nanocrystals. We attribute this result to the confluence of ultra-high purity nanocrystal germanium, generation of high density of electron-hole plasma, the inherent degeneracy of longitudinal and transverse optical phonons in non-polar indirect bandgap semiconductors, and commensurate spatial confinement effects. At high laser intensities, laser cooling with lattice temperature as low as ~50K is inferred.

cond-mat.mes-hall↗

arXiv2020

Mapping Diffuse Emission in Lyman UV band

The CAFE (Census of warm-hot intergalactic medium, Accretion, and Feedback Explorer) and LyRIC (Lyman UV Radiation from Interstellar medium and Circum-galactic medium) have been proposed to the space agencies in China respectively. CAFE was first proposed in 2015 as a joint scientific CAS-ESA small space mission. LyRIC was proposed in 2019 as an independent external payload operating on the Chinese Space Station. Both missions are dedicated to mapping the Lyman UV emission (ionized oxygen (O VI) resonance lines at 103.2 and 103.8 nm, and Lyman series) for the diffuse sources in our Galaxy and the circum-galactic mediums of the nearby galaxies. We present the primary science objectives, mission concepts, the enabling technologies, as well as the current status.

astro-ph.IM↗

arXiv2020

Uplink Achievable Rate Maximization for Reconfigurable Intelligent Surface Aided Millimeter Wave Systems with Resolution-Adaptive ADCs

In this letter, we investigate the uplink of a reconfigurable intelligent surface (RIS)-aided millimeter-wave (mmWave) multi-user system. In the considered system, however, problems with hardware cost and power consumption arise when massive antenna arrays coupled with power-demanding analog-to-digital converters (ADCs) are employed. To account for practical hardware complexity, we consider that the access point (AP) is equipped with resolution-adaptive analog-to-digital converters (RADCs). We maximize the achievable rate under hardware constraints by jointly optimizing the ADC quantization bits, the RIS phase shifts, and the beam selection matrix. Due to the non-convexity of the feasible set and objective function, the formulated problem is non-convex and difficult to solve. To efficiently tackle this problem, a block coordinated descent (BCD)-based algorithm is proposed. Simulations demonstrate that an RIS can mitigate the hardware loss due to use of RADCs, and that the proposed BCD-based algorithm outperforms state-of-the-art algorithms.

eess.SP↗