arXiv subjects

Bing Wang

Publications and source records attributed to Bing Wang.

At least 253 records · Page 14Linked to original sources

arXiv2021

DynaNet: Neural Kalman Dynamical Model for Motion Estimation and Prediction

Dynamical models estimate and predict the temporal evolution of physical systems. State Space Models (SSMs) in particular represent the system dynamics with many desirable properties, such as being able to model uncertainty in both the model and measurements, and optimal (in the Bayesian sense) recursive formulations e.g. the Kalman Filter. However, they require significant domain knowledge to derive the parametric form and considerable hand-tuning to correctly set all the parameters. Data driven techniques e.g. Recurrent Neural Networks have emerged as compelling alternatives to SSMs with wide success across a number of challenging tasks, in part due to their ability to extract relevant features from rich inputs. They however lack interpretability and robustness to unseen conditions. In this work, we present DynaNet, a hybrid deep learning and time-varying state-space model which can be trained end-to-end. Our neural Kalman dynamical model allows us to exploit the relative merits of each approach. We demonstrate state-of-the-art estimation and prediction on a number of physically challenging tasks, including visual odometry, sensor fusion for visual-inertial navigation and pendulum control. In addition we show how DynaNet can indicate failures through investigation of properties such as the rate of innovation (Kalman Gain).

cs.LG↗

arXiv2021

Ricci curvature integrals, local functionals, and the Ricci flow

Consider a Riemannian manifold $(M^{m}, g)$ whose volume is the same as the standard sphere $(S^{m}, g_{round})$. If $p>\frac{m}{2}$ and $\int_{M} \left\{ Rc-(m-1)g\right\}_{-}^{p} dv$ is sufficiently small, we show that the normalized Ricci flow initiated from $(M^{m}, g)$ will exist immortally and converge to the standard sphere. The choice of $p$ is optimal.

math.DG↗

arXiv2021

BossNAS: Exploring Hybrid CNN-transformers with Block-wisely Self-supervised Neural Architecture Search

A myriad of recent breakthroughs in hand-crafted neural architectures for visual recognition have highlighted the urgent need to explore hybrid architectures consisting of diversified building blocks. Meanwhile, neural architecture search methods are surging with an expectation to reduce human efforts. However, whether NAS methods can efficiently and effectively handle diversified search spaces with disparate candidates (e.g. CNNs and transformers) is still an open question. In this work, we present Block-wisely Self-supervised Neural Architecture Search (BossNAS), an unsupervised NAS method that addresses the problem of inaccurate architecture rating caused by large weight-sharing space and biased supervision in previous methods. More specifically, we factorize the search space into blocks and utilize a novel self-supervised training scheme, named ensemble bootstrapping, to train each block separately before searching them as a whole towards the population center. Additionally, we present HyTra search space, a fabric-like hybrid CNN-transformer search space with searchable down-sampling positions. On this challenging search space, our searched model, BossNet-T, achieves up to 82.5% accuracy on ImageNet, surpassing EfficientNet by 2.4% with comparable compute time. Moreover, our method achieves superior architecture rating accuracy with 0.78 and 0.76 Spearman correlation on the canonical MBConv search space with ImageNet and on NATS-Bench size search space with CIFAR-100, respectively, surpassing state-of-the-art NAS methods. Code: https://github.com/changlin31/BossNAS

cs.CV↗

arXiv2021

Transport of Self-propelled Janus Particles Confined in Corrugated Channel with Lévy Noise

The transport of self-propelled particle confined in corrugated channel with Lévy noise is investigated. The parameters of Lévy noise(i.e., the stability index, the asymmetry parameter, the scale parameter, the location parameter) and the parameters of confined corrugated channel(i.e., the compartment length, the channel width and the bottleneck size) have joint effects on the system. There exits flow reverse phenomena with increasing mean parameter. Left distribution noise will induce $-x$ directional transport and right distribution noise will induce $+x$ directional transport. The distribution skewness will effect the moving direction of the particle. The average velocity shows complex behavior with increasing stability index. The angle velocity and the angle Gaussian noise have little effects on the particle transport.

cond-mat.stat-mech↗

arXiv2021

P2-Net: Joint Description and Detection of Local Features for Pixel and Point Matching

Accurately describing and detecting 2D and 3D keypoints is crucial to establishing correspondences across images and point clouds. Despite a plethora of learning-based 2D or 3D local feature descriptors and detectors having been proposed, the derivation of a shared descriptor and joint keypoint detector that directly matches pixels and points remains under-explored by the community. This work takes the initiative to establish fine-grained correspondences between 2D images and 3D point clouds. In order to directly match pixels and points, a dual fully convolutional framework is presented that maps 2D and 3D inputs into a shared latent representation space to simultaneously describe and detect keypoints. Furthermore, an ultra-wide reception mechanism in combination with a novel loss function are designed to mitigate the intrinsic information variations between pixel and point local regions. Extensive experimental results demonstrate that our framework shows competitive performance in fine-grained matching between images and point clouds and achieves state-of-the-art results for the task of indoor visual localization. Our source code will be available at [no-name-for-blind-review].

cs.CV↗

arXiv2021

The Inhomogeneity of Composition along the Magnetic Cloud Axis

It is generally accepted that CMEs result from eruptions of magnetic flux ropes, which are dubbed as magnetic clouds in interplanetary space. The composition (including the ionic charge states and elemental abundances) is determined prior to and/or during CME eruptions in the solar atmosphere, and does not alter during magnetic cloud propagation to 1 AU and beyond. It has been known that the composition is not uniform within a cross section perpendicular to magnetic cloud axis, and the distribution of ionic charge states within a cross section provides us an important clue to investigate the formation and eruption processes of flux ropes due to the freeze-in effect. The flux rope is a three dimensional magnetic structure intrinsically, and it remains unclear whether the composition is uniform along the flux rope axis as most magnetic clouds are only detected by one spacecraft. In this paper we report a magnetic cloud that was observed by ACE near 1 AU on 1998 March 4--6 and Ulysses near 5.4 AU on March 24--28 sequentially. At these times, both spacecraft were located around the ecliptic plane, and the latitudinal and longitudinal separations between them were $\sim$2.2$^{\circ}$ and $\sim$5.5$^{\circ}$, respectively. It provides us an excellent opportunity to explore the axial inhomogeneity of flux rope composition, as both spacecraft almost intersected the cloud center at different sites along its axis. Our study shows that the average values of ionic charge states exhibit significant difference along the axis for carbon, and the differences are relatively slight but still obvious for charge states of oxygen and iron, as well as the elemental abundances of iron and helium. Besides the means, the composition profiles within the cloud measured by both spacecraft also exhibit some discrepancies. We conclude that the inhomogeneity of composition exists along the cloud axis.

astro-ph.SR↗

arXiv2021

Impact of individual behavioral changes on epidemic spreading in time-varying networks

Changs in individual behavior often entangles with the dynamic interaction of individuals, which complicates the epidemic process and brings great challenges for the understanding and control of the epidemic. In this work, we consider three kinds of typical behavioral changes in epidemic process that is, self-quarantine of infected individuals, self-protection of susceptible individuals, and social distancing between them. We connect the behavioral changes with individual's social attributes by the activity-driven network with attractiveness. A mean-field theory is established to derive an analytical estimate of epidemic threshold for SIS models with individual behavioral changes, which depends on the correlations between activity, attractiveness and the number of generative links in the susceptible and infected states. We find that individual behaviors play different roles in suppressing the epidemic. Although all the behavioral changes could delay the epidemic by increasing the epidemic threshold, self-quarantine and social distancing of infected individuals could effectively decrease the epidemic size. In addition, simultaneous changes in these behaviors and the timing of implement of them also play a key role in suppressing the epidemic. These results provide helpful significance for understanding the interaction of individual behaviors in the epidemic process.

physics.soc-ph↗

arXiv2021

Liouville theorems for semilinear differential inequalities on sub-Riemannian manifolds

In this paper, we generalize Liouville type theorems for some semilinear partial differential inequalities to sub-Riemannian manifolds satisfying a nonnegative generalized curvature-dimension inequality introduced by Baudoin and Garofalo in [5]. In particular, our results apply to all Sasakian manifolds with nonnegative horizontal Webster-Tanaka-Ricci curvature. The key ingredient is to construct a class of "good" cut-off functions. We also provide some upper bounds for lifespan to parabolic and hyperbolic inequalities.

math.DG↗

arXiv2021

PSF-LO: Parameterized Semantic Features Based Lidar Odometry

Lidar odometry (LO) is a key technology in numerous reliable and accurate localization and mapping systems of autonomous driving. The state-of-the-art LO methods generally leverage geometric information to perform point cloud registration. Furthermore, obtaining point cloud semantic information which can describe the environment more abundantly will help for the registration. We present a novel semantic lidar odometry method based on self-designed parameterized semantic features (PSFs) to achieve low-drift ego-motion estimation for autonomous vehicle in realtime. We first use a convolutional neural network-based algorithm to obtain point-wise semantics from the input laser point cloud, and then use semantic labels to separate the road, building, traffic sign and pole-like point cloud and fit them separately to obtain corresponding PSFs. A fast PSF-based matching enable us to refine geometric features (GeFs) registration, reducing the impact of blurred submap surface on the accuracy of GeFs matching. Besides, we design an efficient method to accurately recognize and remove the dynamic objects while retaining static ones in the semantic point cloud, which are beneficial to further improve the accuracy of LO. We evaluated our method, namely PSF-LO, on the public dataset KITTI Odometry Benchmark and ranked #1 among semantic lidar methods with an average translation error of 0.82% in the test dataset at the time of writing.

cs.CV↗

arXiv2021

On Ilmanen's multiplicity-one conjecture for mean curvature flow with type-I mean curvature

In this paper, we show that if the mean curvature of a closed smooth embedded mean curvature flow in R^3 is of type-I, then the rescaled flow at the first finite singular time converges smoothly to a self-shrinker flow with multiplicity one. This result confirms Ilmanen's multiplicity-one conjecture under the assumption that the mean curvature is of type-I. As a corollary, we show that the mean curvature at the first singular time of a closed smooth embedded mean curvature flow in R^3 is at least of type-I.

math.DG↗

arXiv2021

Dynamic Slimmable Network

Current dynamic networks and dynamic pruning methods have shown their promising capability in reducing theoretical computation complexity. However, dynamic sparse patterns on convolutional filters fail to achieve actual acceleration in real-world implementation, due to the extra burden of indexing, weight-copying, or zero-masking. Here, we explore a dynamic network slimming regime, named Dynamic Slimmable Network (DS-Net), which aims to achieve good hardware-efficiency via dynamically adjusting filter numbers of networks at test time with respect to different inputs, while keeping filters stored statically and contiguously in hardware to prevent the extra burden. Our DS-Net is empowered with the ability of dynamic inference by the proposed double-headed dynamic gate that comprises an attention head and a slimming head to predictively adjust network width with negligible extra computation cost. To ensure generality of each candidate architecture and the fairness of gate, we propose a disentangled two-stage training scheme inspired by one-shot NAS. In the first stage, a novel training technique for weight-sharing networks named In-place Ensemble Bootstrapping is proposed to improve the supernet training efficacy. In the second stage, Sandwich Gate Sparsification is proposed to assist the gate training by identifying easy and hard samples in an online way. Extensive experiments demonstrate our DS-Net consistently outperforms its static counterparts as well as state-of-the-art static and dynamic model compression methods by a large margin (up to 5.9%). Typically, DS-Net achieves 2-4x computation reduction and 1.62x real-world acceleration over ResNet-50 and MobileNet with minimal accuracy drops on ImageNet. Code release: https://github.com/changlin31/DS-Net .

cs.CV↗

arXiv2021

On the structure of Ricci shrinkers

We develop a structure theory for non-collapsed Ricci shrinkers without any curvature condition. As applications, we obtain some curvature estimates of the Ricci shrinkers depending only on the non-collapsing constant.

math.DG↗

arXiv2021

Tunable magnetism in ferroelectric α-In2Se3 by hole-doping

Two-dimensional (2D) multiferroics attract intensive investigations because of underlying science and their potential applications. Although many 2D systems have been observed/predicted to be ferroelectric or ferromagnetic, 2D materials with both ferroic properties are still scarce. By using first-principles calculations, we predict that hole-doping can induce robust ferromagnetism in 2D ferroelectric α-In2Se3 due to its unique flat band structure, and the Curie temperature (TC) can be much higher than room temperature. Moreover, the doping concentration, strain, and number of layers can effectively modulate the magnetic moment and the TC of the material. Interestingly, strong magnetoelectric coupling is found at the surface of hole doped multilayer α-In2Se3, which allows non-volatile electric control of magnetization. Our work provides a feasible approach for designing/searching 2D multiferroics with great potential in future device applications, such as memory devices and sensors.

cond-mat.mtrl-sci↗

arXiv2021

Photonic Floquet time crystals

The public and scientists constantly have different perspectives. While on a time crystal, they stand in line and ask: What is a time crystal? Show me a material that is spontaneously crystalline in time? This study synthesizes a photonic material of Floquet time crystals and experimentally observes its indicative period-2T beating. We explicitly reconstruct a discrete time-crystalline ground state and reveal using an appropriately-designed photonic Floquet simulator the rigid period-doubling as a signature of the spontaneous breakage of the discrete time-translational symmetry. Unlike the result of the exquisite many-body interaction, the photonic time crystal is derived from a single-particle topological phase that can be extensively accessed by many pertinent nonequilibrium and periodically-driven platforms. Our observation will drive theoretical and technological interests toward condensed matter physics and topological photonics, and demystify time crystals for the non-scientific public.

quant-ph↗

arXiv2020

Relational Deep Reinforcement Learning for Routing in Wireless Networks

While routing in wireless networks has been studied extensively, existing protocols are typically designed for a specific set of network conditions and so cannot accommodate any drastic changes in those conditions. For instance, protocols designed for connected networks cannot be easily applied to disconnected networks. In this paper, we develop a distributed routing strategy based on deep reinforcement learning that generalizes to diverse traffic patterns, congestion levels, network connectivity, and link dynamics. We make the following key innovations in our design: (i) the use of relational features as inputs to the deep neural network approximating the decision space, which enables our algorithm to generalize to diverse network conditions, (ii) the use of packet-centric decisions to transform the routing problem into an episodic task by viewing packets, rather than wireless devices, as reinforcement learning agents, which provides a natural way to propagate and model rewards accurately during learning, and (iii) the use of extended-time actions to model the time spent by a packet waiting in a queue, which reduces the amount of training data needed and allows the learning algorithm to converge more quickly. We evaluate our routing algorithm using a packet-level simulator and show that the policy our algorithm learns during training is able to generalize to larger and more congested networks, different topologies, and diverse link dynamics. Our algorithm outperforms shortest path and backpressure routing with respect to packets delivered and delay per packet.

cs.NI↗

arXiv2020

Collapsing geometry with Ricci curvature bounded below and Ricci flow smoothing

We survey some recent developments in the study of collapsing Riemannian manifolds with Ricci curvature bounded below, especially the locally bounded Ricci covering geometry and the Ricci flow smoothing techniques. We then prove that if a Calabi-Yau manifold is sufficiently volume collapsed with bounded diameter and sectional curvature, then it admits a Ricci-flat Kähler metrictogether with a compatible pure nilpotent Killing structure: this is related to an open question of Cheeger, Fukaya and Gromov.

math.DG↗

arXiv2020

3-D Motion Capture of an Unmodified Drone with Single-chip Millimeter Wave Radar

Accurate motion capture of aerial robots in 3-D is a key enabler for autonomous operation in indoor environments such as warehouses or factories, as well as driving forward research in these areas. The most commonly used solutions at present are optical motion capture (e.g. VICON) and Ultrawideband (UWB), but these are costly and cumbersome to deploy, due to their requirement of multiple cameras/sensors spaced around the tracking area. They also require the drone to be modified to carry an active or passive marker. In this work, we present an inexpensive system that can be rapidly installed, based on single-chip millimeter wave (mmWave) radar. Importantly, the drone does not need to be modified or equipped with any markers, as we exploit the Doppler signals from the rotating propellers. Furthermore, 3-D tracking is possible from a single point, greatly simplifying deployment. We develop a novel deep neural network and demonstrate decimeter level 3-D tracking at 10Hz, achieving better performance than classical baselines. Our hope is that this low-cost system will act to catalyse inexpensive drone research and increased autonomy.

cs.RO↗

arXiv2020

The local entropy along Ricci flow---Part B: the pseudo-locality theorems

We localize the entropy functionals of G. Perelman and generalize his no-local-collapsing theorem and pseudo-locality theorem. Our generalization is technically inspired by further development of Li-Yau estimates along the Ricci flow. It has various applications, including to show the continuous dependence of the Ricci flow with respect to the initial metric in Gromov-Hausdorff topology with Ricci curvature bounded below, and to show the compactness of the moduli of Kähler manifolds with bounded scalar curvature and a rough locally almost Euclidean condition.

math.DG↗