arXiv subjects

Pei Wang

Publications and source records attributed to Pei Wang.

At least 199 records · Page 11Linked to original sources

arXiv2021

SDTP: Semantic-aware Decoupled Transformer Pyramid for Dense Image Prediction

Although transformer has achieved great progress on computer vision tasks, the scale variation in dense image prediction is still the key challenge. Few effective multi-scale techniques are applied in transformer and there are two main limitations in the current methods. On one hand, self-attention module in vanilla transformer fails to sufficiently exploit the diversity of semantic information because of its rigid mechanism. On the other hand, it is hard to build attention and interaction among different levels due to the heavy computational burden. To alleviate this problem, we first revisit multi-scale problem in dense prediction, verifying the significance of diverse semantic representation and multi-scale interaction, and exploring the adaptation of transformer to pyramidal structure. Inspired by these findings, we propose a novel Semantic-aware Decoupled Transformer Pyramid (SDTP) for dense image prediction, consisting of Intra-level Semantic Promotion (ISP), Cross-level Decoupled Interaction (CDI) and Attention Refinement Function (ARF). ISP explores the semantic diversity in different receptive space. CDI builds the global attention and interaction among different levels in decoupled space which also solves the problem of heavy computation. Besides, ARF is further added to refine the attention in transformer. Experimental results demonstrate the validity and generality of the proposed method, which outperforms the state-of-the-art by a significant margin in dense image prediction tasks. Furthermore, the proposed components are all plug-and-play, which can be embedded in other methods.

cs.CV↗

arXiv2021

Similar Scale-invariant Behaviors between Soft Gamma-ray Repeaters and An Extreme Epoch from FRB 121102

The recent discovery of a Galactic fast radio burst (FRB) associated with a hard X-ray burst from the soft gamma-ray repeater (SGR) J1935+2154 has established the magnetar origin of at least some FRBs. In this work, we study the statistical properties of soft gamma-/hard X-ray bursts from SGRs 1806--20 and J1935+2154 and of radio bursts from the repeating FRB 121102. For SGRs, we show that the probability density functions for the differences of fluences, fluxes, and durations at different times have fat tails with a $q$-Gaussian form. The $q$ values in the $q$-Gaussian distributions are approximately steady and independent of the temporal interval scale adopted, implying a scale-invariant structure of SGRs. These features indicate that SGR bursts may be governed by a self-organizing criticality (SOC) process, confirming previous findings. Very recently, 1652 independent bursts from FRB 121102 have been detected by the Five-hundred-meter Aperture Spherical radio Telescope (FAST). Here we also investigate the scale-invariant structure of FRB 121102 based on the latest observations of FAST, and show that FRB 121102 and SGRs share similar statistical properties. Given the bimodal energy distribution of FRB 121102 bursts, we separately explore the scale-invariant behaviors of low- and high-energy bursts of FRB 121102. We find that the $q$ values of low- and high-energy bursts are different, which further strengthens the evidence of the bimodality of the energy distribution. Scale invariance in both the high-energy component of FRB 121102 and SGRs can be well explained within the same physical framework of fractal-diffusive SOC systems.

astro-ph.HE↗

arXiv2021

FAST discovery of an extremely radio-faint millisecond pulsar from the Fermi-LAT unassociated source 3FGL J0318.1+0252

High sensitivity radio searches of unassociated $γ$-ray sources have proven to be an effective way of finding new pulsars. Using the Five-hundred-meter Aperture Spherical radio Telescope (FAST) during its commissioning phase, we have carried out a number of targeted deep searches of \textit{Fermi} Large Area Telescope (LAT) $γ$-ray sources. On Feb. 27$^{th}$, 2018 we discovered an isolated millisecond pulsar (MSP), PSR J0318+0253, coincident with the unassociated $γ$-ray source 3FGL J0318.1+0252. PSR J0318+0253 has a spin period of $5.19$ milliseconds, a dispersion measure (DM) of $26$ pc cm$^{-3}$ corresponding to a DM distance of about $1.3$ kpc, and a period-averaged flux density of $\sim$11 $\pm$ 2 $μ$Jy at L-band (1.05-1.45 GHz). Among all high energy MSPs, PSR J0318+0253 is the faintest ever detected in radio bands, by a factor of at least $\sim$4 in terms of L-band fluxes. With the aid of the radio ephemeris, an analysis of 9.6 years of \textit{Fermi}-LAT data revealed that PSR J0318+0253 also displays strong $γ$-ray pulsations. Follow-up observations carried out by both Arecibo and FAST suggest a likely spectral turn-over around 350 MHz. This is the first result from the collaboration between FAST and the \textit{Fermi}-LAT teams as well as the first confirmed new MSP discovery by FAST, raising hopes for the detection of many more MSPs. Such discoveries will make a significant contribution to our understanding of the neutron star zoo while potentially contributing to the future detection of gravitational waves, via pulsar timing array (PTA) experiments.

astro-ph.HE↗

arXiv2021

Wood-leaf classification of tree point cloud based on intensity and geometrical information

Terrestrial laser scanning (TLS) can obtain tree point cloud with high precision and high density. Efficient classification of wood points and leaf points is essential to study tree structural parameters and ecological characteristics. By using both the intensity and spatial information, a three-step classification and verification method was proposed to achieve automated wood-leaf classification. Tree point cloud was classified into wood points and leaf points by using intensity threshold, neighborhood density and voxelization successively. Experiment was carried in Haidian Park, Beijing, and 24 trees were scanned by using the RIEGL VZ-400 scanner. The tree point clouds were processed by using the proposed method, whose classification results were compared with the manual classification results which were used as standard results. To evaluate the classification accuracy, three indicators were used in the experiment, which are Overall Accuracy (OA), Kappa coefficient (Kappa) and Matthews correlation coefficient (MCC). The ranges of OA, Kappa and MCC of the proposed method are from 0.9167 to 0.9872, from 0.7276 to 0.9191, and from 0.7544 to 0.9211 respectively. The average values of OA, Kappa and MCC are 0.9550, 0.8547 and 0.8627 respectively. Time cost of wood-leaf classification was also recorded to evaluate the algorithm efficiency. The average processing time are 1.4 seconds per million points. The results showed that the proposed method performed well automatically and quickly on wood-leaf classification based on the experimental dataset.

cs.CV↗

arXiv2021

IMAGINE: Image Synthesis by Image-Guided Model Inversion

We introduce an inversion based method, denoted as IMAge-Guided model INvErsion (IMAGINE), to generate high-quality and diverse images from only a single training sample. We leverage the knowledge of image semantics from a pre-trained classifier to achieve plausible generations via matching multi-level feature representations in the classifier, associated with adversarial training with an external discriminator. IMAGINE enables the synthesis procedure to simultaneously 1) enforce semantic specificity constraints during the synthesis, 2) produce realistic images without generator training, and 3) give users intuitive control over the generation process. With extensive experimental results, we demonstrate qualitatively and quantitatively that IMAGINE performs favorably against state-of-the-art GAN-based and inversion-based methods, across three different image domains (i.e., objects, scenes, and textures).

cs.CV↗

arXiv2021

Rethinking and Improving the Robustness of Image Style Transfer

Extensive research in neural style transfer methods has shown that the correlation between features extracted by a pre-trained VGG network has a remarkable ability to capture the visual style of an image. Surprisingly, however, this stylization quality is not robust and often degrades significantly when applied to features from more advanced and lightweight networks, such as those in the ResNet family. By performing extensive experiments with different network architectures, we find that residual connections, which represent the main architectural difference between VGG and ResNet, produce feature maps of small entropy, which are not suitable for style transfer. To improve the robustness of the ResNet architecture, we then propose a simple yet effective solution based on a softmax transformation of the feature activations that enhances their entropy. Experimental results demonstrate that this small magic can greatly improve the quality of stylization results, even for networks with random weights. This suggests that the architecture used for feature extraction is more important than the use of learned weights for the task of style transfer.

cs.CV↗

arXiv2021

Dynamic Transfer for Multi-Source Domain Adaptation

Recent works of multi-source domain adaptation focus on learning a domain-agnostic model, of which the parameters are static. However, such a static model is difficult to handle conflicts across multiple domains, and suffers from a performance degradation in both source domains and target domain. In this paper, we present dynamic transfer to address domain conflicts, where the model parameters are adapted to samples. The key insight is that adapting model across domains is achieved via adapting model across samples. Thus, it breaks down source domain barriers and turns multi-source domains into a single-source domain. This also simplifies the alignment between source and target domains, as it only requires the target domain to be aligned with any part of the union of source domains. Furthermore, we find dynamic transfer can be simply modeled by aggregating residual matrices and a static convolution matrix. Experimental results show that, without using domain labels, our dynamic transfer outperforms the state-of-the-art method by more than 3% on the large multi-source domain adaptation datasets -- DomainNet. Source code is at https://github.com/liyunsheng13/DRT.

cs.CV↗

arXiv2021

Search-based Kinodynamic Motion Planning for Omnidirectional Quadruped Robots

Autonomous navigation has played an increasingly significant role in quadruped robot system. However, most existing works on quadruped robots navigation using traditional search-based or sample-based methods do not consider the kinodynamic characteristics of quadruped robots, generating kinodynamically infeasible parts, that are difficult to track. In this paper, we introduce a complete navigation system considering the omnidirectional abilities of quadruped robots. First, we use kinodynamic path finding method to obtain smooth, dynamically feasible, time-optimal initial paths and add collision cost as a soft constraint to ensure safety. Then the trajectory is refined by the timed elastic band (TEB) method based on the omnidirectional model of quadruped robots. The superior performance of our work is demonstrated through simulating and real-world experiments on our quadruped robot Jueying Mini.

cs.RO↗

arXiv2021

Estimating the Number of Infected Cases in COVID-19 Pandemic

The COVID-19 pandemic has caused major disturbance to human life. An important reason behind the widespread social anxiety is the huge uncertainty about the pandemic. A fundamental uncertainty is how many or what percentage of people have been infected. There are published and frequently updated data on various statistics of the pandemic, at local, country or global level. However, due to various reasons, many cases were not included in those reported numbers. We propose a structured approach for the estimation of the number of unreported cases, where we distinguish cases that arrive late in the reported numbers and those who had mild or no symptoms and thus were not captured by any medical system at all. We use post-report data for the estimation of the former and population matching to the latter. We estimate that the reported number of infected cases in the US should be corrected by multiplying a factor of 220.54% as of Apr 20, 2020, while the infection ratio out of the US population is estimated to be 0.53%, implying a case mortality rate at 2.85% which is close to the 3.4% suggested by the WHO in Mar 2020. Towards the end of the summer of 2020, the overall infection ratio of the US rises to 2.49% while the case mortality decreases to 2.09%, and the ratio of asymptomatic cases out of all infected cases reduces from the pre-summer 35-40% to around 20-25%.

q-bio.PE↗

arXiv2021

The first evidence for three-dimensional spin-velocity alignment in pulsars

More than 50 years after the discovery of pulsars and confirmation of their association with supernova explosions, the origin of the initial spin and velocity of pulsars remains largely a mystery. The typical space velocities of several hundred km/s have been attributed to "kicks" resulting from asymmetries either in the supernova ejecta or in the neutrino emission. Observations have shown a strong tendency for alignment of the pulsar space velocity and spin axis in young pulsars but, up to now, these comparisons have been restricted to two dimensions. We report here the first evidence for three-dimensional alignment between the spin and velocity vectors, largely based on observations made with the Five-hundred-meter Aperture Spherical radio Telescope of the pulsar PSR~J0538+2817 and its associated supernova remnant S147. Analysis of these and related observations has enabled us to determine the location of the pulsar within the supernova remnant and hence its radial velocity. Current simulations of supernova explosions have difficulty producing such three-dimensional alignment. Our results, which depend on the unprecedented sensitivity of the new observations, add another dimension to the intriguing correlation between pulsar spin-axis and birth-kick directions, thus deepening the mysteries surrounding the birth of neutron stars.

astro-ph.GA↗

arXiv2021

CRAFTS for Fast Radio Bursts Extending the dispersion-fluence relation with new FRBs detected by FAST

We report three new FRBs discovered by the Five-hundred-meter Aperture Spherical radio Telescope (FAST), namely FRB 181017.J0036+11, FRB 181118 and FRB 181130, through the Commensal Radio Astronomy FAST Survey (CRAFTS). Together with FRB 181123 that was reported earlier, all four FAST-discovered FRBs share the same characteristics of low fluence ($\leq$0.2 Jy ms) and high dispersion measure (DM, $>1000$ \dmu), consistent with the anti-correlation between DM and fluence of the entire FRB population. FRB 181118 and FRB 181130 exhibit band-limited features. FRB 181130 is prominently scattered ($τ_s\simeq8$ ms) at 1.25 GHz. FRB 181017.J0036+11 has full-bandwidth emission with a fluence of 0.042 Jy ms, which is one of the faintest FRB sources detected so far. CRAFTS starts to built a new sample of FRBs that fills the region for more distant and fainter FRBs in the fluence-$\rm DM_E$ diagram, previously out of reach of other surveys. The implied all sky event rate of FRBs is $1.24^{+1.94}_{-0.90} \times 10^5$ sky$^{-1}$ day$^{-1}$ at the $95\%$ confidence interval above 0.0146 Jy ms. We also demonstrate here that the probability density function of CRAFTS FRB detections is sensitive to the assumed intrinsic FRB luminosity function and cosmological evolution, which may be further constrained with more discoveries.

astro-ph.HE↗

arXiv2021

Efficient Discretizations of Optimal Transport

Obtaining solutions to Optimal Transportation (OT) problems is typically intractable when the marginal spaces are continuous. Recent research has focused on approximating continuous solutions with discretization methods based on i.i.d. sampling, and has proven convergence as the sample size increases. However, obtaining OT solutions with large sample sizes requires intensive computation effort, that can be prohibitive in practice. In this paper, we propose an algorithm for calculating discretizations with a given number of points for marginal distributions, by minimizing the (entropy-regularized) Wasserstein distance, and result in plans that are comparable to those obtained with much larger numbers of i.i.d. samples. Moreover, a local version of such discretizations which is parallelizable for large scale applications is proposed. We prove bounds for our approximation and demonstrate performance on a wide range of problems.

math.OC↗

arXiv2021

Distributionally-Constrained Policy Optimization via Unbalanced Optimal Transport

We consider constrained policy optimization in Reinforcement Learning, where the constraints are in form of marginals on state visitations and global action executions. Given these distributions, we formulate policy optimization as unbalanced optimal transport over the space of occupancy measures. We propose a general purpose RL objective based on Bregman divergence and optimize it using Dykstra's algorithm. The approach admits an actor-critic algorithm for when the state or action space is large, and only samples from the marginals are available. We discuss applications of our approach and provide demonstrations to show the effectiveness of our algorithm.

cs.LG↗

arXiv2021

A Metamodel and Framework for Artificial General Intelligence From Theory to Practice

This paper introduces a new metamodel-based knowledge representation that significantly improves autonomous learning and adaptation. While interest in hybrid machine learning / symbolic AI systems leveraging, for example, reasoning and knowledge graphs, is gaining popularity, we find there remains a need for both a clear definition of knowledge and a metamodel to guide the creation and manipulation of knowledge. Some of the benefits of the metamodel we introduce in this paper include a solution to the symbol grounding problem, cumulative learning, and federated learning. We have applied the metamodel to problems ranging from time series analysis, computer vision, and natural language understanding and have found that the metamodel enables a wide variety of learning mechanisms ranging from machine learning, to graph network analysis and learning by reasoning engines to interoperate in a highly synergistic way. Our metamodel-based projects have consistently exhibited unprecedented accuracy, performance, and ability to generalize. This paper is inspired by the state-of-the-art approaches to AGI, recent AGI-aspiring work, the granular computing community, as well as Alfred Korzybski's general semantics. One surprising consequence of the metamodel is that it not only enables a new level of autonomous learning and optimal functioning for machine intelligences, but may also shed light on a path to better understanding how to improve human cognition.

cs.AI↗

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

Reconstruction and interpretation of photon Doppler velocimetry spectrum for ejecta particles from shock-loaded sample in vacuum

The photon Doppler velocimetry (PDV) spectrum is investigated in an attempt to reveal the particle parameters of ejecta from shock-loaded samples in a vacuum. A GPU-accelerated Monte-Carlo algorithm, which considers the multiple-scattering effects of light, is applied to reconstruct the light field of the ejecta and simulate the corresponding PDV spectrum. The influence of the velocity profile, total area mass, and particle size of the ejecta on the simulated spectra is discussed qualitatively. To facilitate a quantitative discussion, a novel theoretical optical model is proposed in which the single-scattering assumption is applied. With this model, the relationships between the particle parameters of ejecta and the peak information of the PDV spectrum are derived, enabling direct extraction of the particle parameters from the PDV spectrum. The values of the ejecta parameters estimated from the experimental spectrum are in good agreement with those measured by a piezoelectric probe.

physics.app-ph↗

arXiv2020

On Grothendieck ring of the sequence of walled Brauer algebras

In this paper, we provide a diagrammatic approach to study the branching rules for cell modules on a sequence of walled Brauer algebras. This also allows us to calculate the structure constants of multiplication over the Grothendieck ring of the sequence.

math.RT↗

arXiv2020

Semantic-Guided Representation Enhancement for Self-supervised Monocular Trained Depth Estimation

Self-supervised depth estimation has shown its great effectiveness in producing high quality depth maps given only image sequences as input. However, its performance usually drops when estimating on border areas or objects with thin structures due to the limited depth representation ability. In this paper, we address this problem by proposing a semantic-guided depth representation enhancement method, which promotes both local and global depth feature representations by leveraging rich contextual information. In stead of a single depth network as used in conventional paradigms, we propose an extra semantic segmentation branch to offer extra contextual features for depth estimation. Based on this framework, we enhance the local feature representation by sampling and feeding the point-based features that locate on the semantic edges to an individual Semantic-guided Edge Enhancement module (SEEM), which is specifically designed for promoting depth estimation on the challenging semantic borders. Then, we improve the global feature representation by proposing a semantic-guided multi-level attention mechanism, which enhances the semantic and depth features by exploring pixel-wise correlations in the multi-level depth decoding scheme. Extensive experiments validate the distinct superiority of our method in capturing highly accurate depth on the challenging image areas such as semantic category borders and thin objects. Both quantitative and qualitative experiments on KITTI show that our method outperforms the state-of-the-art methods.

cs.CV↗