BZPEERReading, trust and collaboration for research
Menu

Discover

GraphFollow connections around a paper, topic or saved view.

Workspaces

HomePick up where your research flow left off.InboxHandle requests, replies and notifications from one place.CollectionsCurate reading lists, evidence sets and shareable dossiers.ResearchSee your private provenance graph, trail and imports.CollaborationMove from discovery into focused discussion rooms.PublishDraft, preview and publish full-text research articles.

Network

ExpertsFind specialists worth following or asking for help.CommunitiesJoin research circles organized around shared work.PartnersSee external organizations active around the same topics.

Opportunities

ListingsBrowse roles, calls and collaborations with clearer fit signals.

Account

Log inReturn to your reading and collaboration workspace.Create accountStart saving work, following people and joining teams.
Log inCreate account

Researcher profile

Chao Ma

Chao Ma contributes to research discovery and scholarly infrastructure.

ResearcherAffiliation not importedOpen to collaborate
Computer VisionMachine Learningeess.IVmath.OCmath.STStatistics Theoryphysics.app-phphysics.med-phquant-phArtificial IntelligenceComputational Engineering, Finance, and ScienceComputer Science and Game Theorycond-mat.mes-hallcond-mat.mtrl-scicond-mat.otherDistributed, Parallel, and Cluster Computingeess.ASeess.SPGraphicsmath.NAMethodologyNetworking and Internet ArchitectureNumerical Analysisphysics.chem-phphysics.opticsSound

Trust snapshot

Quick read

Trust 21 - EmergingVerification L1Unclaimed author
42works
0followers
26topics
4close collaborators

Actions

Decide how to stay connected

Follow researcher0

Identity and collaboration

How to connect with this researcher

Claiming links this public author record to a researcher profile and unlocks direct collaboration workflows.

Log in to claim

Direct collaboration

Open a focused conversation when the fit is right

Claim this author entity first to unlock direct invitations.

Research graph

See the researcher in context

Open full explorer

Inspect adjacent work, topics, institutions and collaborators without jumping out to a separate graph page.

Building this graph slice

BZPEER is loading the nearby papers, people, topics and institutions for this page.

Published work

42 published item(s)

preprint2026arXiv

Unified Map Prior Encoder for Mapping and Planning

Online mapping and end-to-end (E2E) planning in autonomous driving remain largely sensor-centric, leaving rich map priors, including HD/SD vector maps, rasterized SD maps, and satellite imagery, underused because of heterogeneity, pose drift, and inconsistent availability at test time. We present UMPE, a Unified Map Prior Encoder that can ingest any subset of four priors and fuse them with BEV features for both mapping and planning. UMPE has two branches. The vector encoder pre-aligns HD/SD polylines with a frame-wise SE(2) correction, encodes points via multi-frequency sinusoidal features, and produces polyline tokens with confidence scores. BEV queries then apply cross-attention with confidence bias, followed by normalized channel-wise gating to avoid length imbalance and softly down-weight uncertain sources. The raster encoder shares a ResNet-18 backbone conditioned by FiLM with scaling and shift at every stage, performs SE(2) micro-alignment, and injects priors through zero-initialized residual fusion, so the network starts from a do-no-harm baseline and learns to add only useful prior evidence. A vector-then-raster fusion order reflects the inductive bias of geometry first, appearance second. On nuScenes mapping, UMPE lifts MapTRv2 from 61.5 to 67.4 mAP (+5.9) and MapQR from 66.4 to 71.7 mAP (+5.3). On Argoverse2, UMPE adds +4.1 mAP over strong baselines. UMPE is compositional: when trained with all priors, it outperforms single-prior models even when only one prior is available at test time, demonstrating powerset robustness. For E2E planning with the VAD backbone on nuScenes, UMPE reduces trajectory error from 0.72 to 0.42 m L2 on average (-0.30 m) and collision rate from 0.22% to 0.12% (-0.10%), surpassing recent prior-injection methods. These results show that a unified, alignment-aware treatment of heterogeneous map priors yields better mapping and better planning.

0 citations0 reviewsNo code linkedOpen access
preprint2022arXiv

AiATrack: Attention in Attention for Transformer Visual Tracking

Transformer trackers have achieved impressive advancements recently, where the attention mechanism plays an important role. However, the independent correlation computation in the attention mechanism could result in noisy and ambiguous attention weights, which inhibits further performance improvement. To address this issue, we propose an attention in attention (AiA) module, which enhances appropriate correlations and suppresses erroneous ones by seeking consensus among all correlation vectors. Our AiA module can be readily applied to both self-attention blocks and cross-attention blocks to facilitate feature aggregation and information propagation for visual tracking. Moreover, we propose a streamlined Transformer tracking framework, dubbed AiATrack, by introducing efficient feature reuse and target-background embeddings to make full use of temporal references. Experiments show that our tracker achieves state-of-the-art performance on six tracking benchmarks while running at a real-time speed.

0 citations0 reviewsNo code linkedOpen access
preprint2022arXiv

Beyond the Quadratic Approximation: the Multiscale Structure of Neural Network Loss Landscapes

A quadratic approximation of neural network loss landscapes has been extensively used to study the optimization process of these networks. Though, it usually holds in a very small neighborhood of the minimum, it cannot explain many phenomena observed during the optimization process. In this work, we study the structure of neural network loss functions and its implication on optimization in a region beyond the reach of a good quadratic approximation. Numerically, we observe that neural network loss functions possesses a multiscale structure, manifested in two ways: (1) in a neighborhood of minima, the loss mixes a continuum of scales and grows subquadratically, and (2) in a larger region, the loss shows several separate scales clearly. Using the subquadratic growth, we are able to explain the Edge of Stability phenomenon [5] observed for the gradient descent (GD) method. Using the separate scales, we explain the working mechanism of learning rate decay by simple examples. Finally, we study the origin of the multiscale structure and propose that the non-convexity of the models and the non-uniformity of training data is one of the causes. By constructing a two-layer neural network problem we show that training data with different magnitudes give rise to different scales of the loss function, producing subquadratic growth and multiple separate scales.

0 citations0 reviewsNo code linkedOpen access
preprint2022arXiv

BSODA: A Bipartite Scalable Framework for Online Disease Diagnosis

A growing number of people are seeking healthcare advice online. Usually, they diagnose their medical conditions based on the symptoms they are experiencing, which is also known as self-diagnosis. From the machine learning perspective, online disease diagnosis is a sequential feature (symptom) selection and classification problem. Reinforcement learning (RL) methods are the standard approaches to this type of tasks. Generally, they perform well when the feature space is small, but frequently become inefficient in tasks with a large number of features, such as the self-diagnosis. To address the challenge, we propose a non-RL Bipartite Scalable framework for Online Disease diAgnosis, called BSODA. BSODA is composed of two cooperative branches that handle symptom-inquiry and disease-diagnosis, respectively. The inquiry branch determines which symptom to collect next by an information-theoretic reward. We employ a Product-of-Experts encoder to significantly improve the handling of partial observations of a large number of features. Besides, we propose several approximation methods to substantially reduce the computational cost of the reward to a level that is acceptable for online services. Additionally, we leverage the diagnosis model to estimate the reward more precisely. For the diagnosis branch, we use a knowledge-guided self-attention model to perform predictions. In particular, BSODA determines when to stop inquiry and output predictions using both the inquiry and diagnosis models. We demonstrate that BSODA outperforms the state-of-the-art methods on several public datasets. Moreover, we propose a novel evaluation method to test the transferability of symptom checking methods from synthetic to real-world tasks. Compared to existing RL baselines, BSODA is more effectively scalable to large search spaces.

0 citations0 reviewsNo code linkedOpen access
preprint2022arXiv

Continual Learning for Blind Image Quality Assessment

The explosive growth of image data facilitates the fast development of image processing and computer vision methods for emerging visual applications, meanwhile introducing novel distortions to the processed images. This poses a grand challenge to existing blind image quality assessment (BIQA) models, failing to continually adapt to such subpopulation shift. Recent work suggests training BIQA methods on the combination of all available human-rated IQA datasets. However, this type of approach is not scalable to a large number of datasets, and is cumbersome to incorporate a newly created dataset as well. In this paper, we formulate continual learning for BIQA, where a model learns continually from a stream of IQA datasets, building on what was learned from previously seen data. We first identify five desiderata in the new setting with a measure to quantify the plasticity-stability trade-off. We then propose a simple yet effective method for learning BIQA models continually. Specifically, based on a shared backbone network, we add a prediction head for a new dataset, and enforce a regularizer to allow all prediction heads to evolve with new data while being resistant to catastrophic forgetting of old data. We compute the quality score by an adaptive weighted summation of estimates from all prediction heads. Extensive experiments demonstrate the promise of the proposed continual learning method in comparison to standard training techniques for BIQA. We made the code publicly available at https://github.com/zwx8981/BIQA_CL.

0 citations0 reviewsNo code linkedOpen access
preprint2022arXiv

Correcting Convexity Bias in Function and Functional Estimate

A general framework with a series of different methods is proposed to improve the estimate of convex function (or functional) values when only noisy observations of the true input are available. Technically, our methods catch the bias introduced by the convexity and remove this bias from a baseline estimate. Theoretical analysis are conducted to show that the proposed methods can strictly reduce the expected estimate error under mild conditions. When applied, the methods require no specific knowledge about the problem except the convexity and the evaluation of the function. Therefore, they can serve as off-the-shelf tools to obtain good estimate for a wide range of problems, including optimization problems with random objective functions or constraints, and functionals of probability distributions such as the entropy and the Wasserstein distance. Numerical experiments on a wide variety of problems show that our methods can significantly improve the quality of the estimate compared with the baseline method.

0 citations0 reviewsNo code linkedOpen access
preprint2022arXiv

Coverage Analysis for Cellular-Connected Random 3D Mobile UAVs with Directional Antennas

This letter proposes an analytical framework to evaluate the coverage performance of a cellular-connected unmanned aerial vehicle (UAV) network in which UAV user equipments (UAV-UEs) are equipped with directional antennas and move according to a three-dimensional (3D) mobility model. The ground base stations (GBSs) equipped with practical down-tilted antennas are distributed according to a Poisson point process (PPP). With tools from stochastic geometry, we derive the handover probability and coverage probability of a random UAV-UE under the strongest average received signal strength (RSS) association strategy. The proposed analytical framework allows to investigate the effect of UAV-UE antenna beamwidth, mobility speed, cell association, and vertical motions on both the handover probability and coverage probability. We conclude that the optimal UAV-UE antenna beamwidth decreases with the GBS density, and the omnidirectional antenna model is preferred in the sparse network scenario. What's more, the superiority of the strongest average RSS association over the nearest association diminishes with the increment of GBS density.

0 citations0 reviewsNo code linkedOpen access
preprint2022arXiv

Deep End-to-end Causal Inference

Causal inference is essential for data-driven decision making across domains such as business engagement, medical treatment and policy making. However, research on causal discovery has evolved separately from inference methods, preventing straight-forward combination of methods from both fields. In this work, we develop Deep End-to-end Causal Inference (DECI), a single flow-based non-linear additive noise model that takes in observational data and can perform both causal discovery and inference, including conditional average treatment effect (CATE) estimation. We provide a theoretical guarantee that DECI can recover the ground truth causal graph under standard causal discovery assumptions. Motivated by application impact, we extend this model to heterogeneous, mixed-type data with missing values, allowing for both continuous and discrete treatment decisions. Our results show the competitive performance of DECI when compared to relevant baselines for both causal discovery and (C)ATE estimation in over a thousand experiments on both synthetic datasets and causal machine learning benchmarks across data-types and levels of missingness.

0 citations0 reviewsNo code linkedOpen access
preprint2022arXiv

Depth-Adapted CNNs for RGB-D Semantic Segmentation

Recent RGB-D semantic segmentation has motivated research interest thanks to the accessibility of complementary modalities from the input side. Existing works often adopt a two-stream architecture that processes photometric and geometric information in parallel, with few methods explicitly leveraging the contribution of depth cues to adjust the sampling position on RGB images. In this paper, we propose a novel framework to incorporate the depth information in the RGB convolutional neural network (CNN), termed Z-ACN (Depth-Adapted CNN). Specifically, our Z-ACN generates a 2D depth-adapted offset which is fully constrained by low-level features to guide the feature extraction on RGB images. With the generated offset, we introduce two intuitive and effective operations to replace basic CNN operators: depth-adapted convolution and depth-adapted average pooling. Extensive experiments on both indoor and outdoor semantic segmentation tasks demonstrate the effectiveness of our approach.

0 citations0 reviewsNo code linkedOpen access
preprint2022arXiv

Direct Measurement of Topological Number by Quench Dynamics

The measurement of topological number is crucial in the research of topological systems. Recently, the relations between the topological number and the dynamics are built. But a direct method to read out the topological number via the dynamics is still lacking. In this work, we propose a new dynamical protocol to directly measure the topological number of an unknown system. Different from common quench operations, we change the Hamiltonian of the unknown system to another one with known topological properties. After the quench, different initial states result in different particle number distributions on the post-quench final Bloch bands. Such distributions depend on the wavefunction overlap between the initial Bloch state and the final Bloch state, which is a complex number depending on the momentum. We prove a theorem that when the momentum varies by $2π$, the phase of the wavefunction overlap change by $Δnπ$ where $Δn$ is the topological number difference between the initial Bloch band and the final Bloch band. Based on this and the known topological number of the final Bloch band, we can directly deduce the topological number of the initial state from the particle number distribution and need not track the evolution of the system nor measure the spin texture. Two experimental schemes are also proposed as well. These schemes provide a convenient and robust measurement method and also deepens the understanding of the relation between topology and dynamics.

0 citations0 reviewsNo code linkedOpen access
preprint2022arXiv

Exploring Frequency Adversarial Attacks for Face Forgery Detection

Various facial manipulation techniques have drawn serious public concerns in morality, security, and privacy. Although existing face forgery classifiers achieve promising performance on detecting fake images, these methods are vulnerable to adversarial examples with injected imperceptible perturbations on the pixels. Meanwhile, many face forgery detectors always utilize the frequency diversity between real and fake faces as a crucial clue. In this paper, instead of injecting adversarial perturbations into the spatial domain, we propose a frequency adversarial attack method against face forgery detectors. Concretely, we apply discrete cosine transform (DCT) on the input images and introduce a fusion module to capture the salient region of adversary in the frequency domain. Compared with existing adversarial attacks (e.g. FGSM, PGD) in the spatial domain, our method is more imperceptible to human observers and does not degrade the visual quality of the original images. Moreover, inspired by the idea of meta-learning, we also propose a hybrid adversarial attack that performs attacks in both the spatial and frequency domains. Extensive experiments indicate that the proposed method fools not only the spatial-based detectors but also the state-of-the-art frequency-based detectors effectively. In addition, the proposed frequency attack enhances the transferability across face forgery detectors as black-box attacks.

0 citations0 reviewsNo code linkedOpen access
preprint2022arXiv

Facial Geometric Detail Recovery via Implicit Representation

Learning a dense 3D model with fine-scale details from a single facial image is highly challenging and ill-posed. To address this problem, many approaches fit smooth geometries through facial prior while learning details as additional displacement maps or personalized basis. However, these techniques typically require vast datasets of paired multi-view data or 3D scans, whereas such datasets are scarce and expensive. To alleviate heavy data dependency, we present a robust texture-guided geometric detail recovery approach using only a single in-the-wild facial image. More specifically, our method combines high-quality texture completion with the powerful expressiveness of implicit surfaces. Initially, we inpaint occluded facial parts, generate complete textures, and build an accurate multi-view dataset of the same subject. In order to estimate the detailed geometry, we define an implicit signed distance function and employ a physically-based implicit renderer to reconstruct fine geometric details from the generated multi-view images. Our method not only recovers accurate facial details but also decomposes normals, albedos, and shading parts in a self-supervised way. Finally, we register the implicit shape details to a 3D Morphable Model template, which can be used in traditional modeling and rendering pipelines. Extensive experiments demonstrate that the proposed approach can reconstruct impressive facial details from a single image, especially when compared with state-of-the-art methods trained on large datasets.

0 citations0 reviewsNo code linkedOpen access
preprint2022arXiv

H2-Stereo: High-Speed, High-Resolution Stereoscopic Video System

High-speed, high-resolution stereoscopic (H2-Stereo) video allows us to perceive dynamic 3D content at fine granularity. The acquisition of H2-Stereo video, however, remains challenging with commodity cameras. Existing spatial super-resolution or temporal frame interpolation methods provide compromised solutions that lack temporal or spatial details, respectively. To alleviate this problem, we propose a dual camera system, in which one camera captures high-spatial-resolution low-frame-rate (HSR-LFR) videos with rich spatial details, and the other captures low-spatial-resolution high-frame-rate (LSR-HFR) videos with smooth temporal details. We then devise a Learned Information Fusion network (LIFnet) that exploits the cross-camera redundancies to enhance both camera views to high spatiotemporal resolution (HSTR) for reconstructing the H2-Stereo video effectively. We utilize a disparity network to transfer spatiotemporal information across views even in large disparity scenes, based on which, we propose disparity-guided flow-based warping for LSR-HFR view and complementary warping for HSR-LFR view. A multi-scale fusion method in feature domain is proposed to minimize occlusion-induced warping ghosts and holes in HSR-LFR view. The LIFnet is trained in an end-to-end manner using our collected high-quality Stereo Video dataset from YouTube. Extensive experiments demonstrate that our model outperforms existing state-of-the-art methods for both views on synthetic data and camera-captured real data with large disparity. Ablation studies explore various aspects, including spatiotemporal resolution, camera baseline, camera desynchronization, long/short exposures and applications, of our system to fully understand its capability for potential applications.

0 citations0 reviewsNo code linkedOpen access
preprint2022arXiv

PillarNet: Real-Time and High-Performance Pillar-based 3D Object Detection

Real-time and high-performance 3D object detection is of critical importance for autonomous driving. Recent top-performing 3D object detectors mainly rely on point-based or 3D voxel-based convolutions, which are both computationally inefficient for onboard deployment. In contrast, pillar-based methods use solely 2D convolutions, which consume less computation resources, but they lag far behind their voxel-based counterparts in detection accuracy. In this paper, by examining the primary performance gap between pillar- and voxel-based detectors, we develop a real-time and high-performance pillar-based detector, dubbed PillarNet.The proposed PillarNet consists of a powerful encoder network for effective pillar feature learning, a neck network for spatial-semantic feature fusion and the commonly used detect head. Using only 2D convolutions, PillarNet is flexible to an optional pillar size and compatible with classical 2D CNN backbones, such as VGGNet and ResNet. Additionally, PillarNet benefits from our designed orientation-decoupled IoU regression loss along with the IoU-aware prediction branch. Extensive experimental results on the large-scale nuScenes Dataset and Waymo Open Dataset demonstrate that the proposed PillarNet performs well over state-of-the-art 3D detectors in terms of effectiveness and efficiency. Code is available at \url{https://github.com/agent-sgs/PillarNet}.

0 citations0 reviewsNo code linkedOpen access
preprint2022arXiv

Provably convergent quasistatic dynamics for mean-field two-player zero-sum games

In this paper, we study the problem of finding mixed Nash equilibrium for mean-field two-player zero-sum games. Solving this problem requires optimizing over two probability distributions. We consider a quasistatic Wasserstein gradient flow dynamics in which one probability distribution follows the Wasserstein gradient flow, while the other one is always at the equilibrium. Theoretical analysis are conducted on this dynamics, showing its convergence to the mixed Nash equilibrium under mild conditions. Inspired by the continuous dynamics of probability distributions, we derive a quasistatic Langevin gradient descent method with inner-outer iterations, and test the method on different problems, including training mixture of GANs.

0 citations0 reviewsNo code linkedOpen access
preprint2022arXiv

Removing Rain Streaks via Task Transfer Learning

Due to the difficulty in collecting paired real-world training data, image deraining is currently dominated by supervised learning with synthesized data generated by e.g., Photoshop rendering. However, the generalization to real rainy scenes is usually limited due to the gap between synthetic and real-world data. In this paper, we first statistically explore why the supervised deraining models cannot generalize well to real rainy cases, and find the substantial difference of synthetic and real rainy data. Inspired by our studies, we propose to remove rain by learning favorable deraining representations from other connected tasks. In connected tasks, the label for real data can be easily obtained. Hence, our core idea is to learn representations from real data through task transfer to improve deraining generalization. We thus term our learning strategy as \textit{task transfer learning}. If there are more than one connected tasks, we propose to reduce model size by knowledge distillation. The pretrained models for the connected tasks are treated as teachers, all their knowledge is distilled to a student network, so that we reduce the model size, meanwhile preserve effective prior representations from all the connected tasks. At last, the student network is fine-tuned with minority of paired synthetic rainy data to guide the pretrained prior representations to remove rain. Extensive experiments demonstrate that proposed task transfer learning strategy is surprisingly successful and compares favorably with state-of-the-art supervised learning methods and apparently surpass other semi-supervised deraining methods on synthetic data. Particularly, it shows superior generalization over them to real-world scenes.

0 citations0 reviewsNo code linkedOpen access
preprint2022arXiv

Temperature-induced dephasing in high-order harmonic generation from solids

High harmonic generation (HHG) in solid and gaseous targets has been proven to be a powerful avenue for the generation of attosecond pulses, whereas the influence of electron-phonon scattering on HHG is a critical outstanding problem. Here we first introduce a temperature dependent lattice vibration model by characterizing the spacing fluctuation. Our results reveal that (i) structural disorder induced by lattice vibration does not lead to generation of even-order harmonics; (ii) dephasing of HHG occurs as the lattice temperature is growing; (iii) an open-trajectory picture predicts the maximal photon energy in the temperature-dependent HHG spectra. Moreover, a formula assessing dephasing time with lattice temperature is proposed to identify the timescale of electron-phonon scattering. This work paves a way to study non-Born-Oppenheimer effect in solids driven by strong field.

0 citations0 reviewsNo code linkedOpen access
preprint2022arXiv

Wireless Real-Time Capacitance Readout Based on Perturbed Nonlinear Parity-Time Symmetry

In this article, we report a vector-network-analyzer-free and real-time LC wireless capacitance readout system based on perturbed nonlinear parity-time (PT) symmetry. The system is composed of two inductively coupled reader-sensor parallel RLC resonators with gain and loss respectively. By searching for the real mode that requires the minimum saturation gain, the steady-state frequency evolution as a function of the sensor capacitance perturbation is analytically deduced. The proposed system can work in different modes by setting different perturbation point. In particular, at the exceptional point of PT symmetry, the system exhibits high sensitivity. Experimental demonstrations revealed the viability of the proposed readout mechanism by measuring the steady-state frequency of the reader resonator in response to the change of trimmer capacitor on the sensor side. Our findings could impact many emerging applications such as implantable medical device for health monitoring, parameter detection in harsh environment and sealed food packages, etc.

0 citations0 reviewsNo code linkedOpen access
preprint2021arXiv

Deep learning-based GTV contouring modeling inter- and intra- observer variability in sarcomas

Background and purpose: The delineation of the gross tumor volume (GTV) is a critical step for radiation therapy treatment planning. The delineation procedure is typically performed manually which exposes two major issues: cost and reproducibility. Delineation is a time-consuming process that is subject to inter- and intra-observer variability. While methods have been proposed to predict GTV contours, typical approaches ignore variability and therefore fail to utilize the valuable confidence information offered by multiple contours. Materials and methods: In this work we propose an automatic GTV contouring method for soft-tissue sarcomas from X-ray computed tomography (CT) images, using deep learning by integrating inter- and intra-observer variability in the learned model. Sixty-eight patients with soft tissue and bone sarcomas were considered in this evaluation, all underwent pre-operative CT imaging used to perform GTV delineation. Four radiation oncologists and radiologists performed three contouring trials each for all patients. We quantify variability by defining confidence levels based on the frequency of inclusion of a given voxel into the GTV and use a deep convolutional neural network to learn GTV confidence maps. Results: Results were compared to confidence maps from the four readers as well as ground-truth consensus contours established jointly by all readers. The resulting continuous Dice score between predicted and true confidence maps was 87% and the Hausdorff distance was 14 mm. Conclusion: Results demonstrate the ability of the proposed method to predict accurate contours while utilizing variability and as such it can be used to improve clinical workflow.

0 citations0 reviewsNo code linkedOpen access
preprint2021arXiv

The Generalization Error of the Minimum-norm Solutions for Over-parameterized Neural Networks

We study the generalization properties of minimum-norm solutions for three over-parametrized machine learning models including the random feature model, the two-layer neural network model and the residual network model. We proved that for all three models, the generalization error for the minimum-norm solution is comparable to the Monte Carlo rate, up to some logarithmic terms, as long as the models are sufficiently over-parametrized.

0 citations0 reviewsNo code linkedOpen access
preprint2020arXiv

A Comparative Analysis of the Optimization and Generalization Property of Two-layer Neural Network and Random Feature Models Under Gradient Descent Dynamics

A fairly comprehensive analysis is presented for the gradient descent dynamics for training two-layer neural network models in the situation when the parameters in both layers are updated. General initialization schemes as well as general regimes for the network width and training data size are considered. In the over-parametrized regime, it is shown that gradient descent dynamics can achieve zero training loss exponentially fast regardless of the quality of the labels. In addition, it is proved that throughout the training process the functions represented by the neural network model are uniformly close to that of a kernel method. For general values of the network width and training data size, sharp estimates of the generalization error is established for target functions in the appropriate reproducing kernel Hilbert space.

0 citations0 reviewsNo code linkedOpen access
preprint2020arXiv

A Mean-field Analysis of Deep ResNet and Beyond: Towards Provable Optimization Via Overparameterization From Depth

Training deep neural networks with stochastic gradient descent (SGD) can often achieve zero training loss on real-world tasks although the optimization landscape is known to be highly non-convex. To understand the success of SGD for training deep neural networks, this work presents a mean-field analysis of deep residual networks, based on a line of works that interpret the continuum limit of the deep residual network as an ordinary differential equation when the network capacity tends to infinity. Specifically, we propose a new continuum limit of deep residual networks, which enjoys a good landscape in the sense that every local minimizer is global. This characterization enables us to derive the first global convergence result for multilayer neural networks in the mean-field regime. Furthermore, without assuming the convexity of the loss landscape, our proof relies on a zero-loss assumption at the global minimizer that can be achieved when the model shares a universal approximation property. Key to our result is the observation that a deep residual network resembles a shallow network ensemble, i.e. a two-layer network. We bound the difference between the shallow network and our ResNet model via the adjoint sensitivity method, which enables us to apply existing mean-field analyses of two-layer networks to deep networks. Furthermore, we propose several novel training schemes based on the new continuous model, including one training procedure that switches the order of the residual blocks and results in strong empirical performance on the benchmark datasets.

0 citations0 reviewsNo code linkedOpen access
preprint2020arXiv

A Note on Parallel Distinguishability of two Quantum Operations

We consider a homogeneous system of linear equations of the form $A_α^{\otimes N} {\bf x} = 0$ arising from the distinguishability of two quantum operations by $N$ uses in parallel, where the coefficient matrix $A_α$ depends on a real parameter $α$. It was conjectured by Duan et al. that the system has a non-trivial nonnegative solution if and only if $α$ lies in a certain interval $R_N$ depending on $N$. We affirm the necessity part of the conjecture and establish the sufficiency of the conjecture for $N\leq 10$ by presenting explicit non-trivial nonnegative solutions for the linear system.

0 citations0 reviewsNo code linkedOpen access
preprint2020arXiv

A Priori Estimates of the Population Risk for Two-layer Neural Networks

New estimates for the population risk are established for two-layer neural networks. These estimates are nearly optimal in the sense that the error rates scale in the same way as the Monte Carlo error rates. They are equally effective in the over-parametrized regime when the network size is much larger than the size of the dataset. These new estimates are a priori in nature in the sense that the bounds depend only on some norms of the underlying functions to be fitted, not the parameters in the model, in contrast with most existing results which are a posteriori in nature. Using these a priori estimates, we provide a perspective for understanding why two-layer neural networks perform better than the related kernel methods.

0 citations0 reviewsNo code linkedOpen access
preprint2020arXiv

Accelerating MRI Reconstruction on TPUs

The advanced magnetic resonance (MR) image reconstructions such as the compressed sensing and subspace-based imaging are considered as large-scale, iterative, optimization problems. Given the large number of reconstructions required by the practical clinical usage, the computation time of these advanced reconstruction methods is often unacceptable. In this work, we propose using Google's Tensor Processing Units (TPUs) to accelerate the MR image reconstruction. TPU is an application-specific integrated circuit (ASIC) for machine learning applications, which has recently been used to solve large-scale scientific computing problems. As proof-of-concept, we implement the alternating direction method of multipliers (ADMM) in TensorFlow to reconstruct images on TPUs. The reconstruction is based on multi-channel, sparsely sampled, and radial-trajectory $k$-space data with sparsity constraints. The forward and inverse non-uniform Fourier transform operations are formulated in terms of matrix multiplications as in the discrete Fourier transform. The sparsifying transform and its adjoint operations are formulated as convolutions. The data decomposition is applied to the measured $k$-space data such that the aforementioned tensor operations are localized within individual TPU cores. The data decomposition and the inter-core communication strategy are designed in accordance with the TPU interconnect network topology in order to minimize the communication time. The accuracy and the high parallel efficiency of the proposed TPU-based image reconstruction method are demonstrated through numerical examples.

0 citations0 reviewsNo code linkedOpen access
preprint2020arXiv

Complexity Measures for Neural Networks with General Activation Functions Using Path-based Norms

A simple approach is proposed to obtain complexity controls for neural networks with general activation functions. The approach is motivated by approximating the general activation functions with one-dimensional ReLU networks, which reduces the problem to the complexity controls of ReLU networks. Specifically, we consider two-layer networks and deep residual networks, for which path-based norms are derived to control complexities. We also provide preliminary analyses of the function spaces induced by these norms and a priori estimates of the corresponding regularized estimators.

0 citations0 reviewsNo code linkedOpen access
preprint2020arXiv

Cross-Modality 3D Object Detection

In this paper, we focus on exploring the fusion of images and point clouds for 3D object detection in view of the complementary nature of the two modalities, i.e., images possess more semantic information while point clouds specialize in distance sensing. To this end, we present a novel two-stage multi-modal fusion network for 3D object detection, taking both binocular images and raw point clouds as input. The whole architecture facilitates two-stage fusion. The first stage aims at producing 3D proposals through sparse point-wise feature fusion. Within the first stage, we further exploit a joint anchor mechanism that enables the network to utilize 2D-3D classification and regression simultaneously for better proposal generation. The second stage works on the 2D and 3D proposal regions and fuses their dense features. In addition, we propose to use pseudo LiDAR points from stereo matching as a data augmentation method to densify the LiDAR points, as we observe that objects missed by the detection network mostly have too few points especially for far-away objects. Our experiments on the KITTI dataset show that the proposed multi-stage fusion helps the network to learn better representations.

0 citations0 reviewsNo code linkedOpen access
preprint2020arXiv

Deep Graph Library: A Graph-Centric, Highly-Performant Package for Graph Neural Networks

Advancing research in the emerging field of deep graph learning requires new tools to support tensor computation over graphs. In this paper, we present the design principles and implementation of Deep Graph Library (DGL). DGL distills the computational patterns of GNNs into a few generalized sparse tensor operations suitable for extensive parallelization. By advocating graph as the central programming abstraction, DGL can perform optimizations transparently. By cautiously adopting a framework-neutral design, DGL allows users to easily port and leverage the existing components across multiple deep learning frameworks. Our evaluation shows that DGL significantly outperforms other popular GNN-oriented frameworks in both speed and memory consumption over a variety of benchmarks and has little overhead for small scale workloads.

0 citations0 reviewsNo code linkedOpen access
preprint2020arXiv

DGL-KE: Training Knowledge Graph Embeddings at Scale

Knowledge graphs have emerged as a key abstraction for organizing information in diverse domains and their embeddings are increasingly used to harness their information in various information retrieval and machine learning tasks. However, the ever growing size of knowledge graphs requires computationally efficient algorithms capable of scaling to graphs with millions of nodes and billions of edges. This paper presents DGL-KE, an open-source package to efficiently compute knowledge graph embeddings. DGL-KE introduces various novel optimizations that accelerate training on knowledge graphs with millions of nodes and billions of edges using multi-processing, multi-GPU, and distributed parallelism. These optimizations are designed to increase data locality, reduce communication overhead, overlap computations with memory accesses, and achieve high operation efficiency. Experiments on knowledge graphs consisting of over 86M nodes and 338M edges show that DGL-KE can compute embeddings in 100 minutes on an EC2 instance with 8 GPUs and 30 minutes on an EC2 cluster with 4 machines with 48 cores/machine. These results represent a 2x~5x speedup over the best competing approaches. DGL-KE is available on https://github.com/awslabs/dgl-ke.

0 citations0 reviewsNo code linkedOpen access
preprint2020arXiv

Long-lived and disorder-free charge transfer states enable endothermic charge separation in efficient non-fullerene organic solar cells

Organic solar cells (OSCs) based on non-fullerene acceptors can show high charge generation yields despite near-zero donor-acceptor energy offsets to drive charge separation and overcome the mutual Coulomb attraction between electron and hole. Here we use time-resolved optical spectroscopy to show that free charges in these systems are generated by thermally activated dissociation of interfacial charge-transfer excitons (CTEs) that occurs over hundreds of picoseconds at room temperature, three orders of magnitude slower than comparable fullerene-based systems. Upon free electron-hole encounters at later times, CTEs and emissive excitons are regenerated, thus setting up an equilibrium between excitons, CTEs and free charges. This endothermic charge separation process enables these systems to operate close to quasi-thermodynamic equilibrium conditions with no requirement for energy offsets to drive charge separation and achieve greatly suppressed non-radiative recombination.

0 citations0 reviewsNo code linkedOpen access
preprint2020arXiv

MR-Based PET Attenuation Correction using a Combined Ultrashort Echo Time/Multi-Echo Dixon Acquisition

We propose a magnetic resonance (MR)-based method for estimation of continuous linear attenuation coefficients (LAC) in positron emission tomography (PET) using a physical compartmental model and ultrashort echo time (UTE)/multi-echo Dixon (mUTE) acquisitions. Specifically, we propose a three-dimensional (3D) mUTE sequence to acquire signals from water, fat, and short-T2 components (e.g., bones) simultaneously in a single acquisition. The proposed mUTE sequence integrates 3D UTE with multi-echo Dixon acquisitions and uses sparse radial trajectories to accelerate imaging speed. Errors in the radial k-space trajectories are measured using a special k-space trajectory mapping sequence and corrected for image reconstruction. A physical compartmental model is used to fit the measured multi-echo MR signals to obtain fractions of water, fat and bone components for each voxel, which are then used to estimate the continuous LAC map for PET attenuation correction. The performance of the proposed method was evaluated via phantom and in vivo human studies, using LACs from Computed Tomography (CT) as reference. Compared to Dixon- and atlas-based MRAC methods, the proposed method yielded PET images with higher correlation and similarity in relation to the reference. The relative absolute errors of PET activity values reconstructed by the proposed method were below 5% in all of the four lobes (frontal, temporal, parietal, occipital), cerebellum, whole white matter and gray matter regions across all subjects (n=6). The proposed mUTE method can generate subject-specific, continuous LAC map for PET attenuation correction in PET/MR.

0 citations0 reviewsNo code linkedOpen access
preprint2020arXiv

Rethinking Image Deraining via Rain Streaks and Vapors

Single image deraining regards an input image as a fusion of a background image, a transmission map, rain streaks, and atmosphere light. While advanced models are proposed for image restoration (i.e., background image generation), they regard rain streaks with the same properties as background rather than transmission medium. As vapors (i.e., rain streaks accumulation or fog-like rain) are conveyed in the transmission map to model the veiling effect, the fusion of rain streaks and vapors do not naturally reflect the rain image formation. In this work, we reformulate rain streaks as transmission medium together with vapors to model rain imaging. We propose an encoder-decoder CNN named as SNet to learn the transmission map of rain streaks. As rain streaks appear with various shapes and directions, we use ShuffleNet units within SNet to capture their anisotropic representations. As vapors are brought by rain streaks, we propose a VNet containing spatial pyramid pooling (SSP) to predict the transmission map of vapors in multi-scales based on that of rain streaks. Meanwhile, we use an encoder CNN named ANet to estimate atmosphere light. The SNet, VNet, and ANet are jointly trained to predict transmission maps and atmosphere light for rain image restoration. Extensive experiments on the benchmark datasets demonstrate the effectiveness of the proposed visual model to predict rain streaks and vapors. The proposed deraining method performs favorably against state-of-the-art deraining approaches.

0 citations0 reviewsNo code linkedOpen access
preprint2020arXiv

Robust Tracking against Adversarial Attacks

While deep convolutional neural networks (CNNs) are vulnerable to adversarial attacks, considerably few efforts have been paid to construct robust deep tracking algorithms against adversarial attacks. Current studies on adversarial attack and defense mainly reside in a single image. In this work, we first attempt to generate adversarial examples on top of video sequences to improve the tracking robustness against adversarial attacks. To this end, we take temporal motion into consideration when generating lightweight perturbations over the estimated tracking results frame-by-frame. On one hand, we add the temporal perturbations into the original video sequences as adversarial examples to greatly degrade the tracking performance. On the other hand, we sequentially estimate the perturbations from input sequences and learn to eliminate their effect for performance restoration. We apply the proposed adversarial attack and defense approaches to state-of-the-art deep tracking algorithms. Extensive evaluations on the benchmark datasets demonstrate that our defense method not only eliminates the large performance drops caused by adversarial attacks, but also achieves additional performance gains when deep trackers are not under adversarial attacks.

0 citations0 reviewsNo code linkedOpen access
preprint2020arXiv

See More, Know More: Unsupervised Video Object Segmentation with Co-Attention Siamese Networks

We introduce a novel network, called CO-attention Siamese Network (COSNet), to address the unsupervised video object segmentation task from a holistic view. We emphasize the importance of inherent correlation among video frames and incorporate a global co-attention mechanism to improve further the state-of-the-art deep learning based solutions that primarily focus on learning discriminative foreground representations over appearance and motion in short-term temporal segments. The co-attention layers in our network provide efficient and competent stages for capturing global correlations and scene context by jointly computing and appending co-attention responses into a joint feature space. We train COSNet with pairs of video frames, which naturally augments training data and allows increased learning capacity. During the segmentation stage, the co-attention model encodes useful information by processing multiple reference frames together, which is leveraged to infer the frequently reappearing and salient foreground objects better. We propose a unified and end-to-end trainable framework where different co-attention variants can be derived for mining the rich context within videos. Our extensive experiments over three large benchmarks manifest that COSNet outperforms the current alternatives by a large margin.

0 citations0 reviewsNo code linkedOpen access
preprint2020arXiv

Semantic Equivalent Adversarial Data Augmentation for Visual Question Answering

Visual Question Answering (VQA) has achieved great success thanks to the fast development of deep neural networks (DNN). On the other hand, the data augmentation, as one of the major tricks for DNN, has been widely used in many computer vision tasks. However, there are few works studying the data augmentation problem for VQA and none of the existing image based augmentation schemes (such as rotation and flipping) can be directly applied to VQA due to its semantic structure -- an $\langle image, question, answer\rangle$ triplet needs to be maintained correctly. For example, a direction related Question-Answer (QA) pair may not be true if the associated image is rotated or flipped. In this paper, instead of directly manipulating images and questions, we use generated adversarial examples for both images and questions as the augmented data. The augmented examples do not change the visual properties presented in the image as well as the \textbf{semantic} meaning of the question, the correctness of the $\langle image, question, answer\rangle$ is thus still maintained. We then use adversarial learning to train a classic VQA model (BUTD) with our augmented data. We find that we not only improve the overall performance on VQAv2, but also can withstand adversarial attack effectively, compared to the baseline model. The source code is available at https://github.com/zaynmi/seada-vqa.

0 citations0 reviewsNo code linkedOpen access
preprint2020arXiv

The Quenching-Activation Behavior of the Gradient Descent Dynamics for Two-layer Neural Network Models

A numerical and phenomenological study of the gradient descent (GD) algorithm for training two-layer neural network models is carried out for different parameter regimes when the target function can be accurately approximated by a relatively small number of neurons. It is found that for Xavier-like initialization, there are two distinctive phases in the dynamic behavior of GD in the under-parametrized regime: An early phase in which the GD dynamics follows closely that of the corresponding random feature model and the neurons are effectively quenched, followed by a late phase in which the neurons are divided into two groups: a group of a few "activated" neurons that dominate the dynamics and a group of background (or "quenched") neurons that support the continued activation and deactivation process. This neural network-like behavior is continued into the mildly over-parametrized regime, where it undergoes a transition to a random feature-like behavior. The quenching-activation process seems to provide a clear mechanism for "implicit regularization". This is qualitatively different from the dynamics associated with the "mean-field" scaling where all neurons participate equally and there does not appear to be qualitative changes when the network parameters are changed.

0 citations0 reviewsNo code linkedOpen access
preprint2020arXiv

The Slow Deterioration of the Generalization Error of the Random Feature Model

The random feature model exhibits a kind of resonance behavior when the number of parameters is close to the training sample size. This behavior is characterized by the appearance of large generalization gap, and is due to the occurrence of very small eigenvalues for the associated Gram matrix. In this paper, we examine the dynamic behavior of the gradient descent algorithm in this regime. We show, both theoretically and experimentally, that there is a dynamic self-correction mechanism at work: The larger the eventual generalization gap, the slower it develops, both because of the small eigenvalues. This gives us ample time to stop the training process and obtain solutions with good generalization property.

0 citations0 reviewsNo code linkedOpen access
preprint2020arXiv

Two-stage model and optimal SI-SNR for monaural multi-speaker speech separation in noisy environment

In daily listening environments, speech is always distorted by background noise, room reverberation and interference speakers. With the developing of deep learning approaches, much progress has been performed on monaural multi-speaker speech separation. Nevertheless, most studies in this area focus on a simple problem setup of laboratory environment, which background noises and room reverberations are not considered. In this paper, we propose a two-stage model based on conv-TasNet to deal with the notable effects of noises and interference speakers separately, where enhancement and separation are conducted sequentially using deep dilated temporal convolutional networks (TCN). In addition, we develop a new objective function named optimal scale-invariant signal-noise ratio (OSI-SNR), which are better than original SI-SNR at any circumstances. By jointly training the two-stage model with OSI-SNR, our algorithm outperforms one-stage separation baselines substantially.

0 citations0 reviewsNo code linkedOpen access
preprint2020arXiv

Unsupervised Deep Representation Learning for Real-Time Tracking

The advancement of visual tracking has continuously been brought by deep learning models. Typically, supervised learning is employed to train these models with expensive labeled data. In order to reduce the workload of manual annotations and learn to track arbitrary objects, we propose an unsupervised learning method for visual tracking. The motivation of our unsupervised learning is that a robust tracker should be effective in bidirectional tracking. Specifically, the tracker is able to forward localize a target object in successive frames and backtrace to its initial position in the first frame. Based on such a motivation, in the training process, we measure the consistency between forward and backward trajectories to learn a robust tracker from scratch merely using unlabeled videos. We build our framework on a Siamese correlation filter network, and propose a multi-frame validation scheme and a cost-sensitive loss to facilitate unsupervised learning. Without bells and whistles, the proposed unsupervised tracker achieves the baseline accuracy as classic fully supervised trackers while achieving a real-time speed. Furthermore, our unsupervised framework exhibits a potential in leveraging more unlabeled or weakly labeled data to further improve the tracking accuracy.

0 citations0 reviewsNo code linkedOpen access
preprint2020arXiv

VAEM: a Deep Generative Model for Heterogeneous Mixed Type Data

Deep generative models often perform poorly in real-world applications due to the heterogeneity of natural data sets. Heterogeneity arises from data containing different types of features (categorical, ordinal, continuous, etc.) and features of the same type having different marginal distributions. We propose an extension of variational autoencoders (VAEs) called VAEM to handle such heterogeneous data. VAEM is a deep generative model that is trained in a two stage manner such that the first stage provides a more uniform representation of the data to the second stage, thereby sidestepping the problems caused by heterogeneous data. We provide extensions of VAEM to handle partially observed data, and demonstrate its performance in data generation, missing data prediction and sequential feature selection tasks. Our results show that VAEM broadens the range of real-world applications where deep generative models can be successfully deployed.

0 citations0 reviewsNo code linkedOpen access
preprint2019arXiv

Anomalous Interlayer Exciton Diffusion in Twist-Angle-Dependent Moiré Potentials of WS$_2$-WSe$_2$ Heterobilayers

The nanoscale periodic potentials introduced by moiré patterns in semiconducting van der Waals (vdW) heterostructures provide a new platform for designing exciton superlattices. To realize these applications, a thorough understanding of the localization and delocalization of interlayer excitons in the moiré potentials is necessary. Here, we investigated interlayer exciton dynamics and transport modulated by the moiré potentials in WS$_2$-WSe$_2$ heterobilayers in time, space, and momentum domains using transient absorption microscopy combined with first-principles calculations. Experimental results verified the theoretical prediction of energetically favorable K-Q interlayer excitons and unraveled exciton-population dynamics that was controlled by the twist-angle-dependent energy difference between the K-Q and K-K excitons. Spatially- and temporally-resolved exciton-population imaging directly visualizes exciton localization by twist-angle-dependent moiré potentials of ~100 meV. Exciton transport deviates significantly from normal diffusion due to the interplay between the moiré potentials and strong many-body interactions, leading to exciton-density- and twist-angle-dependent diffusion length. These results have important implications for designing vdW heterostructures for exciton and spin transport as well as for quantum communication applications.

0 citations0 reviewsNo code linkedOpen access
preprint2019arXiv

Real-Time Correlation Tracking via Joint Model Compression and Transfer

Correlation filters (CF) have received considerable attention in visual tracking because of their computational efficiency. Leveraging deep features via off-the-shelf CNN models (e.g., VGG), CF trackers achieve state-of-the-art performance while consuming a large number of computing resources. This limits deep CF trackers to be deployed to many mobile platforms on which only a single-core CPU is available. In this paper, we propose to jointly compress and transfer off-the-shelf CNN models within a knowledge distillation framework. We formulate a CNN model pretrained from the image classification task as a teacher network, and distill this teacher network into a lightweight student network as the feature extractor to speed up CF trackers. In the distillation process, we propose a fidelity loss to enable the student network to maintain the representation capability of the teacher network. Meanwhile, we design a tracking loss to adapt the objective of the student network from object recognition to visual tracking. The distillation process is performed offline on multiple layers and adaptively updates the student network using a background-aware online learning scheme. Extensive experiments on five challenging datasets demonstrate that the lightweight student network accelerates the speed of state-of-the-art deep CF trackers to real-time on a single-core CPU while maintaining almost the same tracking accuracy.

0 citations0 reviewsNo code linkedOpen access