arXiv subjects

Chen Wang

Publications and source records attributed to Chen Wang.

At least 649 records · Page 36Linked to original sources

arXiv2021

Federated Unlearning

Federated learning (FL) has recently emerged as a promising distributed machine learning (ML) paradigm. Practical needs of the "right to be forgotten" and countering data poisoning attacks call for efficient techniques that can remove, or unlearn, specific training data from the trained FL model. Existing unlearning techniques in the context of ML, however, are no longer in effect for FL, mainly due to the inherent distinction in the way how FL and ML learn from data. Therefore, how to enable efficient data removal from FL models remains largely under-explored. In this paper, we take the first step to fill this gap by presenting FedEraser, the first federated unlearning methodology that can eliminate the influence of a federated client's data on the global FL model while significantly reducing the time used for constructing the unlearned FL model.The basic idea of FedEraser is to trade the central server's storage for unlearned model's construction time, where FedEraser reconstructs the unlearned model by leveraging the historical parameter updates of federated clients that have been retained at the central server during the training process of FL. A novel calibration method is further developed to calibrate the retained updates, which are further used to promptly construct the unlearned model, yielding a significant speed-up to the reconstruction of the unlearned model while maintaining the model efficacy. Experiments on four realistic datasets demonstrate the effectiveness of FedEraser, with an expected speed-up of $4\times$ compared with retraining from the scratch. We envision our work as an early step in FL towards compliance with legal and ethical criteria in a fair and transparent manner.

cs.LG↗

arXiv2021

Possible existence of bound nuclei beyond neutron drip lines driven by deformation

Based on the relativistic calculations of the nuclear masses in the transfermium region from No $(Z=102)$ to Ds $(Z=110)$ by the deformed relativistic Hartree-Bogoliubov theory in continuum, the possible existence of the bound nuclei beyond the neutron drip lines is studied. The two-neutron and multi-neutron emission bound nuclei beyond the primary neutron drip line of $N=258$ are predicted in $Z=106,108$ and $110$ isotopes. Detailed microscopic mechanism investigation reveals that nuclear deformation plays a vital role in the existence of the bound nuclei beyond the drip line. Furthermore, not only the quadrupole deformation $β_{2}$, but also the higher orders of deformation are indispensible in the reliable description of the phenomenon of the reentrant binding.

nucl-th↗

arXiv2021

Few-Shot Meta-Learning on Point Cloud for Semantic Segmentation

The promotion of construction robots can solve the problem of human resource shortage and improve the quality of decoration. To help the construction robots obtain environmental information, we need to use 3D point cloud, which is widely used in robotics, autonomous driving, and so on. With a good understanding of environmental information, construction robots can work better. However, the dynamic changes of 3D point cloud data may bring difficulties for construction robots to understand environmental information, such as when construction robots renovate houses. The paper proposes a semantic segmentation method for point cloud based on meta-learning. The method includes a basic learning module and a meta-learning module. The basic learning module is responsible for learning data features and evaluating the model, while the meta-learning module is responsible for updating the parameters of the model and improving the model generalization ability. In our work, we pioneered the method of producing datasets for meta-learning in 3D scenes, as well as demonstrated that the Model-Agnostic Meta-Learning (MAML) algorithm can be applied to process 3D point cloud data. At the same time, experiments show that our method can allow the model to be quickly applied to new environments with a few samples. Our method has important applications.

cs.RO↗

arXiv2021

Molecular gas distribution perpendicular to the Galactic plane

We use the ~370 square degrees data from the MWISP CO survey to study the vertical distribution of the molecular clouds (MCs) toward the tangent points in the region of l=[16,52]deg and |b|<5.1deg. The molecular disk consists of two components with the layer thickness (FWHM) of ~85pc and ~280pc, respectively. In the inner Galaxy, the molecular mass in the thin disk is dominant, while the molecular mass traced by the discrete MCs with weak CO emission in the thick disk is probably <10% of the whole molecular disk. For the CO gas in the thick disk, we identified 1055 high-z MCs that are >100pc from the Galactic plane. However, only a few samples (i.e., 32 MCs) are located in the |z|>360pc region. Typically, the discrete MCs of the thick disk population have a median peak temperature of 2.1 K, a median velocity dispersion of 0.8km/s, and a median effective radius of 2.5pc. The median surface density of these MCs is 6.8 Msun/pc^2, indicating very faint CO emission for these high-z MCs. The cloud-cloud velocity dispersion is 4.9+-1.3 km/s and a linear variation with a slope of -0.4 km/s/kpc is obtained in the region of R_GC=2.2-6.4kpc. Assuming that these clouds are supported by their turbulent motions against the gravitational pull of the disk, a model of rho0(R) = 1.28exp(-R/3.2kpc) Msun/pc^3 can be used to describe the distribution of the total mass density in the Galactic midplane.

astro-ph.GA↗

arXiv2021

Towards Real-time Semantic RGB-D SLAM in Dynamic Environments

Most of the existing visual SLAM methods heavily rely on a static world assumption and easily fail in dynamic environments. Some recent works eliminate the influence of dynamic objects by introducing deep learning-based semantic information to SLAM systems. However such methods suffer from high computational cost and cannot handle unknown objects. In this paper, we propose a real-time semantic RGB-D SLAM system for dynamic environments that is capable of detecting both known and unknown moving objects. To reduce the computational cost, we only perform semantic segmentation on keyframes to remove known dynamic objects, and maintain a static map for robust camera tracking. Furthermore, we propose an efficient geometry module to detect unknown moving objects by clustering the depth image into a few regions and identifying the dynamic regions via their reprojection errors. The proposed method is evaluated on public datasets and real-world conditions. To the best of our knowledge, it is one of the first semantic RGB-D SLAM systems that run in real-time on a low-power embedded platform and provide high localization accuracy in dynamic environments.

cs.RO↗

arXiv2021

Leveraging a Joint of Phenotypic and Genetic Features on Cancer Patient Subgrouping

Cancer is responsible for millions of deaths worldwide every year. Although significant progress has been achieved in cancer medicine, many issues remain to be addressed for improving cancer therapy. Appropriate cancer patient stratification is the prerequisite for selecting appropriate treatment plan, as cancer patients are of known heterogeneous genetic make-ups and phenotypic differences. In this study, built upon deep phenotypic characterizations extractable from Mayo Clinic electronic health records (EHRs) and genetic test reports for a collection of cancer patients, we developed a system leveraging a joint of phenotypic and genetic features for cancer patient subgrouping. The workflow is roughly divided into three parts: feature preprocessing, cancer patient classification, and cancer patient clustering based. In feature preprocessing step, we performed filtering, retaining the most relevant features. In cancer patient classification, we utilized joint categorical features to build a patient-feature matrix and applied nine different machine learning models, Random Forests (RF), Decision Tree (DT), Support Vector Machine (SVM), Naive Bayes (NB), Logistic Regression (LR), Multilayer Perceptron (MLP), Gradient Boosting (GB), Convolutional Neural Network (CNN), and Feedforward Neural Network (FNN), for classification purposes. Finally, in the cancer patient clustering step, we leveraged joint embeddings features and patient-feature associations to build an undirected feature graph and then trained the cancer feature node embeddings.

cs.LG↗

arXiv2021

Directly measuring mode purity of single component in superposed optical vortices

Mode purity is an important reference for the quality of the optical vortex. In this work, we propose a self-interference method to directly measure single component in superposed optical vortices based on phase-shifting technology. This method has excellent flexibility and robustness, which can be applied to a variety of occasions and harsh conditions. Careful alignment and optimized error analysis allow us to generate and measure single component with mode purity as high as 99.997%.

physics.optics↗

arXiv2021

Supervised Contrastive Learning

Contrastive learning applied to self-supervised representation learning has seen a resurgence in recent years, leading to state of the art performance in the unsupervised training of deep image models. Modern batch contrastive approaches subsume or significantly outperform traditional contrastive losses such as triplet, max-margin and the N-pairs loss. In this work, we extend the self-supervised batch contrastive approach to the fully-supervised setting, allowing us to effectively leverage label information. Clusters of points belonging to the same class are pulled together in embedding space, while simultaneously pushing apart clusters of samples from different classes. We analyze two possible versions of the supervised contrastive (SupCon) loss, identifying the best-performing formulation of the loss. On ResNet-200, we achieve top-1 accuracy of 81.4% on the ImageNet dataset, which is 0.8% above the best number reported for this architecture. We show consistent outperformance over cross-entropy on other datasets and two ResNet variants. The loss shows benefits for robustness to natural corruptions and is more stable to hyperparameter settings such as optimizers and data augmentations. Our loss function is simple to implement, and reference TensorFlow code is released at https://t.ly/supcon.

cs.LG↗

arXiv2021

Insight-HXMT observations of jet-like corona in a black hole X-ray binary MAXI J1820+070

A black hole X-ray binary produces hard X-ray radiation from its corona and disk when the accreting matter heats up. During an outburst, the disk and corona co-evolves with each other. However, such an evolution is still unclear in both its geometry and dynamics. Here we report the unusual decrease of the reflection fraction in MAXI J1820+070, which is the ratio of the coronal intensity illuminating the disk to the coronal intensity reaching the observer, as the corona is observed to contrast during the decay phase. We postulate a jet-like corona model, in which the corona can be understood as a standing shock where the material flowing through. In this dynamical scenario, the decrease of the reflection fraction is a signature of the corona's bulk velocity. Our findings suggest that as the corona is observed to get closer to the black hole, the coronal material might be outflowing faster.

astro-ph.HE↗

arXiv2021

Decentralized Circle Formation Control for Fish-like Robots in the Real-world via Reinforcement Learning

In this paper, the circle formation control problem is addressed for a group of cooperative underactuated fish-like robots involving unknown nonlinear dynamics and disturbances. Based on the reinforcement learning and cognitive consistency theory, we propose a decentralized controller without the knowledge of the dynamics of the fish-like robots. The proposed controller can be transferred from simulation to reality. It is only trained in our established simulation environment, and the trained controller can be deployed to real robots without any manual tuning. Simulation results confirm that the proposed model-free robust formation control method is scalable with respect to the group size of the robots and outperforms other representative RL algorithms. Several experiments in the real world verify the effectiveness of our RL-based approach for circle formation control.

cs.RO↗

arXiv2021

Arbitrary controlled-phase gate on fluxonium qubits using differential ac-Stark shifts

Large scale quantum computing motivates the invention of two-qubit gate schemes that not only maximize the gate fidelity but also draw minimal resources. In the case of superconducting qubits, the weak anharmonicity of transmons imposes profound constraints on the gate design, leading to increased complexity of devices and control protocols. Here we demonstrate a resource-efficient control over the interaction of strongly-anharmonic fluxonium qubits. Namely, applying an off-resonant drive to non-computational transitions in a pair of capacitively-coupled fluxoniums induces a $\textrm{ZZ}$-interaction due to unequal ac-Stark shifts of the computational levels. With a continuous choice of frequency and amplitude, the drive can either cancel the static $\textrm{ZZ}$-term or increase it by an order of magnitude to enable a controlled-phase (CP) gate with an arbitrary programmed phase shift. The cross-entropy benchmarking of these non-Clifford operations yields a sub $1\%$ error, limited solely by incoherent processes. Our result demonstrates the advantages of strongly-anharmonic circuits over transmons in designing the next generation of quantum processors.

quant-ph↗

arXiv2021

Human-Understandable Decision Making for Visual Recognition

The widespread use of deep neural networks has achieved substantial success in many tasks. However, there still exists a huge gap between the operating mechanism of deep learning models and human-understandable decision making, so that humans cannot fully trust the predictions made by these models. To date, little work has been done on how to align the behaviors of deep learning models with human perception in order to train a human-understandable model. To fill this gap, we propose a new framework to train a deep neural network by incorporating the prior of human perception into the model learning process. Our proposed model mimics the process of perceiving conceptual parts from images and assessing their relative contributions towards the final recognition. The effectiveness of our proposed model is evaluated on two classical visual recognition tasks. The experimental results and analysis confirm our model is able to provide interpretable explanations for its predictions, but also maintain competitive recognition accuracy.

cs.AI↗

arXiv2021

Revealing spatial spillover effect in high-tech industry agglomeration from a high-skilled labor flow network perspective

Understanding the high-tech industrial agglomeration from a spatial-spillover perspective is essential for cities to gain economic and technological competitive advantages. Along with rapid urbanization and the development of fast transportation networks, socioeconomic interactions between cities have been ever-increasing, traditional spatial metrics are not enough to describe actual inter-city connections. High-skilled labor flow between cities strongly influences the high-tech industrial agglomeration, yet receives less attention. By exploiting unique large-scale datasets and tools from complex network and data mining, we construct an inter-city high-skilled labor flow network, which was integrated into spatial econometric models. Our regression results indicate that spatial-spillover effects exist in the development of high-tech industries in the Yangtze River Delta Urban Agglomeration region. Moreover, the spatial-spillover effects are stronger among cities with a higher volume of high-skilled labor flows than among cities with just stronger geographic connections. Additionally, we investigate the channels for the spillover effects and discover that inadequate local government expenses on science and technology likely hamper the high-tech industrial agglomeration, so does the inadequate local educational provision. The increasing foreign direct investments in one city likely encourages the high-tech industrial agglomeration in other cities because of the policy inertia toward traditional industries.

physics.soc-ph↗

arXiv2021

Measurement and Shaping of Perfect Optical Vortex via Cross Phase

We investigate a method for the measurement and shaping of a perfect optical vortex (POV) by using the cross-phase for the first time. Experimentally, we propose a method to measure the topological charges +3 and +-5 of POVs with the low-order cross-phase (LOCP) at the Fourier plane; we realize shaping intensity distributions of POVs at the Fourier plane with the high-order cross-phase (HOCP) and show the asymmetric modulation ability of the HOCP for an optical field. Furthermore, we discuss the influence of parameters on the optical field in the shaping process. On this basis, we propose a secondary energy control method that combines the characteristics of the POV, which ensures flexible control of the shape while reducing the impact on the orbital angular momentum of the original POV. This work provides an alternative method to measure and shape a POV, which facilitates applications in optical micro-manipulation.

physics.optics↗

arXiv2021

Intensity-SLAM: Intensity Assisted Localization and Mapping for Large Scale Environment

Simultaneous Localization And Mapping (SLAM) is a task to estimate the robot location and to reconstruct the environment based on observation from sensors such as LIght Detection And Ranging (LiDAR) and camera. It is widely used in robotic applications such as autonomous driving and drone delivery. Traditional LiDAR-based SLAM algorithms mainly leverage the geometric features from the scene context, while the intensity information from LiDAR is ignored. Some recent deep-learning-based SLAM algorithms consider intensity features and train the pose estimation network in an end-to-end manner. However, they require significant data collection effort and their generalizability to environments other than the trained one remains unclear. In this paper we introduce intensity features to a SLAM system. And we propose a novel full SLAM framework that leverages both geometry and intensity features. The proposed SLAM involves both intensity-based front-end odometry estimation and intensity-based back-end optimization. Thorough experiments are performed including both outdoor autonomous driving and indoor warehouse robot manipulation. The results show that the proposed method outperforms existing geometric-only LiDAR SLAM methods.

cs.RO↗

arXiv2021

Lightweight 3-D Localization and Mapping for Solid-State LiDAR

The LIght Detection And Ranging (LiDAR) sensor has become one of the most important perceptual devices due to its important role in simultaneous localization and mapping (SLAM). Existing SLAM methods are mainly developed for mechanical LiDAR sensors, which are often adopted by large scale robots. Recently, the solid-state LiDAR is introduced and becomes popular since it provides a cost-effective and lightweight solution for small scale robots. Compared to mechanical LiDAR, solid-state LiDAR sensors have higher update frequency and angular resolution, but also have smaller field of view (FoV), which is very challenging for existing LiDAR SLAM algorithms. Therefore, it is necessary to have a more robust and computationally efficient SLAM method for this new sensing device. To this end, we propose a new SLAM framework for solid-state LiDAR sensors, which involves feature extraction, odometry estimation, and probability map building. The proposed method is evaluated on a warehouse robot and a hand-held device. In the experiments, we demonstrate both the accuracy and efficiency of our method using an Intel L515 solid-state LiDAR. The results show that our method is able to provide precise localization and high quality mapping. We made the source codes public at \url{https://github.com/wh200720041/SSL_SLAM}.

cs.RO↗

arXiv2021

On the Beam Filling Factors of Molecular Clouds

Imaging surveys of CO and other molecular transition lines are fundamental to measuring the large-scale distribution of molecular gas in the Milky Way. Due to finite angular resolution and sensitivity, however, observational effects are inevitable in the surveys, but few studies are available on the extent of uncertainties involved. The purpose of this work is to investigate the dependence of observations on angular resolution (beam sizes), sensitivity (noise levels), distances, and molecular tracers. To this end, we use high-quality CO images of a large-scale region (25.8 <l< 49.7 deg and |b|<5 deg) mapped by the Milky Way Imaging Scroll Painting (MWISP) survey as a benchmark to simulate observations with larger beam sizes and higher noise levels, deriving corresponding beam filling and sensitivity clip factors. The sensitivity clip factor is defined to be the completeness of observed flux. Taking the entire image as a whole object, we found that 12CO has the largest beam filling and sensitivity clip factors and C18O has the lowest. For molecular cloud samples extracted from images, the beam filling factor can be described by a characteristic size, $l_{1/4}$=0.762 (in beam size), at which the beam filling factor is approximately 1/4. The sensitivity clip factor shows a similar relationship but is more correlated with the mean voxel signal-to-noise ratio of molecular clouds. This result may serve as a practical reference on beam filling and sensitivity clip factors in further analyses of the MWISP data and other observations.

astro-ph.GA↗

arXiv2021

Spin-Wave Doppler Shift by Magnon Drag in Magnetic Insulators

The Doppler shift of the quasiparticle dispersion by charge currents is responsible for the critical supercurrents in superconductors and instabilities of the magnetic ground state of metallic ferromagnets. Here we predict an analogous effect in thin films of magnetic insulators in which microwaves emitted by a proximity stripline generate coherent chiral spin currents that cause a Doppler shift in the magnon dispersion. The spin-wave instability is suppressed by magnon-magnon interactions that limit spin currents to values close to but below the threshold for the instability. The spin current limitations by the backaction of magnon currents on the magnetic order should be considered as design parameters in magnonic devices.

cond-mat.mes-hall↗