Search arXiv⌕ Search

arXiv subjects

Fei Wang

Publications and source records attributed to Fei Wang.

At least 469 records · Page 26Linked to original sources
arXiv2023

Crowd Counting with Sparse Annotation

This paper presents a new annotation method called Sparse Annotation (SA) for crowd counting, which reduces human labeling efforts by sparsely labeling individuals in an image. We argue that sparse labeling can reduce the redundancy of full annotation and capture more diverse information from distant individuals that is not fully captured by Partial Annotation methods. Besides, we propose a point-based Progressive Point Matching network (PPM) to better explore the crowd from the whole image with sparse annotation, which includes a Proposal Matching Network (PMN) and a Performance Restoration Network (PRN). The PMN generates pseudo-point samples using a basic point classifier, while the PRN refines the point classifier with the pseudo points to maximize performance. Our experimental results show that PPM outperforms previous semi-supervised crowd counting methods with the same amount of annotation by a large margin and achieves competitive performance with state-of-the-art fully-supervised methods.

cs.CV↗
arXiv2023

Neural Multi-network Diffusion towards Social Recommendation

Graph Neural Networks (GNNs) have been widely applied on a variety of real-world applications, such as social recommendation. However, existing GNN-based models on social recommendation suffer from serious problems of generalization and oversmoothness, because of the underexplored negative sampling method and the direct implanting of the off-the-shelf GNN models. In this paper, we propose a succinct multi-network GNN-based neural model (NeMo) for social recommendation. Compared with the existing methods, the proposed model explores a generative negative sampling strategy, and leverages both the positive and negative user-item interactions for users' interest propagation. The experiments show that NeMo outperforms the state-of-the-art baselines on various real-world benchmark datasets (e.g., by up to 38.8% in terms of NDCG@15).

cs.LG↗
arXiv2023

Stochastic Graph Neural Network-based Value Decomposition for MARL in Internet of Vehicles

Autonomous driving has witnessed incredible advances in the past several decades, while Multi-Agent Reinforcement Learning (MARL) promises to satisfy the essential need of autonomous vehicle control in a wireless connected vehicle networks. In MARL, how to effectively decompose a global feedback into the relative contributions of individual agents belongs to one of the most fundamental problems. However, the environment volatility due to vehicle movement and wireless disturbance could significantly shape time-varying topological relationships among agents, thus making the Value Decomposition (VD) challenging. Therefore, in order to cope with this annoying volatility, it becomes imperative to design a dynamic VD framework. Hence, in this paper, we propose a novel Stochastic VMIX (SVMIX) methodology by taking account of dynamic topological features during the VD and incorporating the corresponding components into a multi-agent actor-critic architecture. In particular, Stochastic Graph Neural Network (SGNN) is leveraged to effectively capture underlying dynamics in topological features and improve the flexibility of VD against the environment volatility. Finally, the superiority of SVMIX is verified through extensive simulations.

cs.MA↗
arXiv2023

Evolution of the number and temperature of the remaining cold atoms in CW-laser photoionization of laser-cooled $^{87}$Rb atoms

Based on the Rb$^+$-Rb hybrid trap, we investigate the effect of ion-atom elastic collisions on the number and temperature of the remaining atoms. We measured the remaining atomic number and temperature as a function of the wavelength and intensity of the ionization laser, and whether the ion trap was turned on. Fittings with a single exponential decay function plus an offset to the number and radius of the remaining atoms are found to be in good agreement. We found a difference in the exponential factor of different wavelengths of ionization laser with the ion trap on or off. We suppose that the presence of electrons affects ion-atom collisions through disorder-induced heating. Our research contributes to a better understanding of how ultracold neutral plasma evolves, particularly the subsequent kinetics of atomic processes, which also serves as a useful reference for high-energy-density plasma.

physics.atom-ph↗
arXiv2023

FactReranker: Fact-guided Reranker for Faithful Radiology Report Summarization

Automatic radiology report summarization is a crucial clinical task, whose key challenge is to maintain factual accuracy between produced summaries and ground truth radiology findings. Existing research adopts reinforcement learning to directly optimize factual consistency metrics such as CheXBert or RadGraph score. However, their decoding method using greedy search or beam search considers no factual consistency when picking the optimal candidate, leading to limited factual consistency improvement. To address it, we propose a novel second-stage summarizing approach FactReranker, the first attempt that learns to choose the best summary from all candidates based on their estimated factual consistency score. We propose to extract medical facts of the input medical report, its gold summary, and candidate summaries based on the RadGraph schema and design the fact-guided reranker to efficiently incorporate the extracted medical facts for selecting the optimal summary. We decompose the fact-guided reranker into the factual knowledge graph generation and the factual scorer, which allows the reranker to model the mapping between the medical facts of the input text and its gold summary, thus can select the optimal summary even the gold summary can't be observed during inference. We also present a fact-based ranking metric (RadMRR) for measuring the ability of the reranker on selecting factual consistent candidates. Experimental results on two benchmark datasets demonstrate the superiority of our method in generating summaries with higher factual consistency scores when compared with existing methods.

cs.CL↗
arXiv2023

SadTalker: Learning Realistic 3D Motion Coefficients for Stylized Audio-Driven Single Image Talking Face Animation

Generating talking head videos through a face image and a piece of speech audio still contains many challenges. ie, unnatural head movement, distorted expression, and identity modification. We argue that these issues are mainly because of learning from the coupled 2D motion fields. On the other hand, explicitly using 3D information also suffers problems of stiff expression and incoherent video. We present SadTalker, which generates 3D motion coefficients (head pose, expression) of the 3DMM from audio and implicitly modulates a novel 3D-aware face render for talking head generation. To learn the realistic motion coefficients, we explicitly model the connections between audio and different types of motion coefficients individually. Precisely, we present ExpNet to learn the accurate facial expression from audio by distilling both coefficients and 3D-rendered faces. As for the head pose, we design PoseVAE via a conditional VAE to synthesize head motion in different styles. Finally, the generated 3D motion coefficients are mapped to the unsupervised 3D keypoints space of the proposed face render, and synthesize the final video. We conducted extensive experiments to demonstrate the superiority of our method in terms of motion and video quality.

cs.CV↗
arXiv2023

Towards NeuroAI: Introducing Neuronal Diversity into Artificial Neural Networks

Throughout history, the development of artificial intelligence, particularly artificial neural networks, has been open to and constantly inspired by the increasingly deepened understanding of the brain, such as the inspiration of neocognitron, which is the pioneering work of convolutional neural networks. Per the motives of the emerging field: NeuroAI, a great amount of neuroscience knowledge can help catalyze the next generation of AI by endowing a network with more powerful capabilities. As we know, the human brain has numerous morphologically and functionally different neurons, while artificial neural networks are almost exclusively built on a single neuron type. In the human brain, neuronal diversity is an enabling factor for all kinds of biological intelligent behaviors. Since an artificial network is a miniature of the human brain, introducing neuronal diversity should be valuable in terms of addressing those essential problems of artificial networks such as efficiency, interpretability, and memory. In this Primer, we first discuss the preliminaries of biological neuronal diversity and the characteristics of information transmission and processing in a biological neuron. Then, we review studies of designing new neurons for artificial networks. Next, we discuss what gains can neuronal diversity bring into artificial networks and exemplary applications in several important fields. Lastly, we discuss the challenges and future directions of neuronal diversity to explore the potential of NeuroAI.

cs.NE↗
arXiv2023

FedLP: Layer-wise Pruning Mechanism for Communication-Computation Efficient Federated Learning

Federated learning (FL) has prevailed as an efficient and privacy-preserved scheme for distributed learning. In this work, we mainly focus on the optimization of computation and communication in FL from a view of pruning. By adopting layer-wise pruning in local training and federated updating, we formulate an explicit FL pruning framework, FedLP (Federated Layer-wise Pruning), which is model-agnostic and universal for different types of deep learning models. Two specific schemes of FedLP are designed for scenarios with homogeneous local models and heterogeneous ones. Both theoretical and experimental evaluations are developed to verify that FedLP relieves the system bottlenecks of communication and computation with marginal performance decay. To the best of our knowledge, FedLP is the first framework that formally introduces the layer-wise pruning into FL. Within the scope of federated learning, more variants and combinations can be further designed based on FedLP.

cs.LG↗
arXiv2023

NetMoST: A network-based machine learning approach for subtyping schizophrenia using polygenic SNP allele biomarkers

Subtyping neuropsychiatric disorders like schizophrenia is essential for improving the diagnosis and treatment of complex diseases. Subtyping schizophrenia is challenging because it is polygenic and genetically heterogeneous, rendering the standard symptom-based diagnosis often unreliable and unrepeatable. We developed a novel network-based machine-learning approach, netMoST, to subtyping psychiatric disorders. NetMoST identifies polygenic risk SNP-allele modules from genome-wide genotyping data as polygenic haplotype biomarkers (PHBs) for disease subtyping. We applied netMoST to subtype a cohort of schizophrenia subjects into three distinct biotypes with differentiable genetic, neuroimaging and functional characteristics. The PHBs of the first biotype (36.9% of all patients) were related to neurodevelopment and cognition, the PHBs of the second biotype (28.4%) were enriched for neuroimmune functions, and the PHBs of the third biotype (34.7%) were associated with the transport of calcium ions and neurotransmitters. Neuroimaging patterns provided additional support to the new biotypes, with unique regional homogeneity (ReHo) patterns observed in the brains of each biotype compared with healthy controls. Our findings demonstrated netMoST's capability for uncovering novel biotypes of complex diseases such as schizophrenia. The results also showed the power of exploring polygenic allelic patterns that transcend the conventional GWAS approaches.

q-bio.MN↗
arXiv2023

UV Volumes for Real-time Rendering of Editable Free-view Human Performance

Neural volume rendering enables photo-realistic renderings of a human performer in free-view, a critical task in immersive VR/AR applications. But the practice is severely limited by high computational costs in the rendering process. To solve this problem, we propose the UV Volumes, a new approach that can render an editable free-view video of a human performer in real-time. It separates the high-frequency (i.e., non-smooth) human appearance from the 3D volume, and encodes them into 2D neural texture stacks (NTS). The smooth UV volumes allow much smaller and shallower neural networks to obtain densities and texture coordinates in 3D while capturing detailed appearance in 2D NTS. For editability, the mapping between the parameterized human model and the smooth texture coordinates allows us a better generalization on novel poses and shapes. Furthermore, the use of NTS enables interesting applications, e.g., retexturing. Extensive experiments on CMU Panoptic, ZJU Mocap, and H36M datasets show that our model can render 960 x 540 images in 30FPS on average with comparable photo-realism to state-of-the-art methods. The project and supplementary materials are available at https://fanegg.github.io/UV-Volumes.

cs.CV↗
arXiv2023

Local-to-Global Registration for Bundle-Adjusting Neural Radiance Fields

Neural Radiance Fields (NeRF) have achieved photorealistic novel views synthesis; however, the requirement of accurate camera poses limits its application. Despite analysis-by-synthesis extensions for jointly learning neural 3D representations and registering camera frames exist, they are susceptible to suboptimal solutions if poorly initialized. We propose L2G-NeRF, a Local-to-Global registration method for bundle-adjusting Neural Radiance Fields: first, a pixel-wise flexible alignment, followed by a frame-wise constrained parametric alignment. Pixel-wise local alignment is learned in an unsupervised way via a deep network which optimizes photometric reconstruction errors. Frame-wise global alignment is performed using differentiable parameter estimation solvers on the pixel-wise correspondences to find a global transformation. Experiments on synthetic and real-world data show that our method outperforms the current state-of-the-art in terms of high-fidelity reconstruction and resolving large camera pose misalignment. Our module is an easy-to-use plugin that can be applied to NeRF variants and other neural field applications. The Code and supplementary materials are available at https://rover-xingyu.github.io/L2G-NeRF/.

cs.CV↗
arXiv2023

CbwLoss: Constrained Bidirectional Weighted Loss for Self-supervised Learning of Depth and Pose

Photometric differences are widely used as supervision signals to train neural networks for estimating depth and camera pose from unlabeled monocular videos. However, this approach is detrimental for model optimization because occlusions and moving objects in a scene violate the underlying static scenario assumption. In addition, pixels in textureless regions or less discriminative pixels hinder model training. To solve these problems, in this paper, we deal with moving objects and occlusions utilizing the difference of the flow fields and depth structure generated by affine transformation and view synthesis, respectively. Secondly, we mitigate the effect of textureless regions on model optimization by measuring differences between features with more semantic and contextual information without adding networks. In addition, although the bidirectionality component is used in each sub-objective function, a pair of images are reasoned about only once, which helps reduce overhead. Extensive experiments and visual analysis demonstrate the effectiveness of the proposed method, which outperform existing state-of-the-art self-supervised methods under the same conditions and without introducing additional auxiliary information.

cs.CV↗
arXiv2023

Distinguishing and controlling Mottness in 1T-TaS$_2$ by ultrafast light

Distinguishing and controlling the extent of Mottness is important for materials where the energy scales of the onsite Coulomb repulsion U and the bandwidth W are comparable. Here we report the ultrafast electronic dynamics of 1T-TaS$_2$ by ultrafast time- and angle-resolved photoemission spectroscopy. A comparison of the electron dynamics for the newly-discovered intermediate phase (I-phase) as well as the low-temperature commensurate charge density wave (C-CDW) phase shows distinctive dynamics. While the I-phase is characterized by an instantaneous response and nearly time-resolution-limited fast relaxation (~200 fs), the C-CDW phase shows a delayed response and a slower relaxation (a few ps). Such distinctive dynamics refect the different relaxation mechanisms and provide nonequilibrium signatures to distinguish the Mott insulating I-phase from the C-CDW band insulating phase. Moreover, a light-induced bandwidth reduction is observed in the C-CDW phase, pushing it toward the Mott insulating phase. Our work demonstrates the power of ultrafast light-matter interaction in both distinguishing and controlling the extent of Mottness on the ultrafast timescale.

cond-mat.mes-hall↗
arXiv2023

Masked Distillation with Receptive Tokens

Distilling from the feature maps can be fairly effective for dense prediction tasks since both the feature discriminability and localization priors can be well transferred. However, not every pixel contributes equally to the performance, and a good student should learn from what really matters to the teacher. In this paper, we introduce a learnable embedding dubbed receptive token to localize those pixels of interests (PoIs) in the feature map, with a distillation mask generated via pixel-wise attention. Then the distillation will be performed on the mask via pixel-wise reconstruction. In this way, a distillation mask actually indicates a pattern of pixel dependencies within feature maps of teacher. We thus adopt multiple receptive tokens to investigate more sophisticated and informative pixel dependencies to further enhance the distillation. To obtain a group of masks, the receptive tokens are learned via the regular task loss but with teacher fixed, and we also leverage a Dice loss to enrich the diversity of learned masks. Our method dubbed MasKD is simple and practical, and needs no priors of tasks in application. Experiments show that our MasKD can achieve state-of-the-art performance consistently on object detection and semantic segmentation benchmarks. Code is available at: https://github.com/hunto/MasKD .

cs.CV↗
arXiv2023

Explaining The CDF-II W-Boson Mass Anomaly in the Georgi-Machacek Extension Models

Original Georgi-Machacek model can preserve the custodial symmetry at tree level after the electroweak symmetry breaking. Unless additional $SU(2)_c$ custodial symmetry breaking effects are significant, the new physics contributions to $Δm_W$ are always very small. Our numerical results show that ordinary GM model can contribute to $Δm_W$ a maximal amount $0.0012$ GeV, which can not explain the new CDF-II anomaly on W boson mass. We propose firstly to introduce small misalignment among the triplet VEVs to increase $Δm_W$, which can give large new physics contributions to $Δm_W$. Such slightly misaligned triplet VEVs from custodial symmetry preserving scalar potential can still be allowed. Our numerical results indicate that the resulting $Δm_W$ can easily reach the $1σ$ range of CDF-II $m_W$ data and the splitting among the triplet VEVs $Δv$($\equiv v_ξ-v_χ$) can be as small as $0.8$ GeV for $v_χ\lesssim 15$ GeV. We also propose to introduce an additional custodial symmetry breaking source by extending the GM model with a low scale RH neutrino sector, which can adopt the leptogenesis mechanism and allow moderately large coupling strength $h_{ij}$ even for triplet VEVs of order GeV. With low scale RH neutrino mass scale of order $10^2\sim 10^4$ TeV, the new physics contribution to $Δm_W$ can reach $0.03$ GeV and is much larger than that of ordinary GM model. Combining both custodial $SU(2)_c$ symmetry breaking effects, the small misalignment among the triplet VEVs and the moderately large $h_{ij}$ couplings allowed with RH neutrino sector, the value of $Δm_W$ can still easily reach the $1σ$ range of CDF-II $m_W$ data, with a minimum splitting (among the triplet VEVs) approximately $0.7$ GeV for $v_χ\lesssim 15$ GeV.

hep-ph↗
arXiv2023

Evaluate underdiagnosis and overdiagnosis bias of deep learning model on primary open-angle glaucoma diagnosis in under-served patient populations

In the United States, primary open-angle glaucoma (POAG) is the leading cause of blindness, especially among African American and Hispanic individuals. Deep learning has been widely used to detect POAG using fundus images as its performance is comparable to or even surpasses diagnosis by clinicians. However, human bias in clinical diagnosis may be reflected and amplified in the widely-used deep learning models, thus impacting their performance. Biases may cause (1) underdiagnosis, increasing the risks of delayed or inadequate treatment, and (2) overdiagnosis, which may increase individuals' stress, fear, well-being, and unnecessary/costly treatment. In this study, we examined the underdiagnosis and overdiagnosis when applying deep learning in POAG detection based on the Ocular Hypertension Treatment Study (OHTS) from 22 centers across 16 states in the United States. Our results show that the widely-used deep learning model can underdiagnose or overdiagnose underserved populations. The most underdiagnosed group is female younger (< 60 yrs) group, and the most overdiagnosed group is Black older (>=60 yrs) group. Biased diagnosis through traditional deep learning methods may delay disease detection, treatment and create burdens among under-served populations, thereby, raising ethical concerns about using deep learning models in ophthalmology clinics.

cs.CV↗
arXiv2022

Diagnosis of ultrafast ultraintense laser pulse characteristics by machine-learning-assisted electron spin

Rapid development of ultrafast ultraintense laser technologies continues to create opportunities for studying strong-field physics under extreme conditions. However, accurate determination of the spatial and temporal characteristics of a laser pulse is still a great challenge, especially when laser powers higher than hundreds of terawatts are involved. In this paper, by utilizing the radiative spin-flip effect, we find that the spin depolarization of an electron beam can be employed to diagnose characteristics of ultrafast ultraintense lasers with peak intensities around $10^{20}$-$10^{22}$~W/cm$^2$. With three shots, our machine-learning-assisted model can predict, simultaneously, the pulse duration, peak intensity, and focal radius of a focused Gaussian ultrafast ultraintense laser (in principle, the profile can be arbitrary) with relative errors of $0.1\%$-$10\%$. The underlying physics and an alternative diagnosis method (without the assistance of machine learning) are revealed by the asymptotic approximation of the final spin degree of polarization. Our proposed scheme exhibits robustness and detection accuracy with respect to fluctuations in the electron beam parameters. Accurate measurements of the ultrafast ultraintense laser parameters will lead to much higher precision in, for example, laser nuclear physics investigations and laboratory astrophysics studies. Robust machine learning techniques may also find applications in more general strong-field physics scenarios.

physics.plasm-ph↗