arXiv subjects

Chao Chen

Publications and source records attributed to Chao Chen.

At least 757 records · Page 42Linked to original sources

arXiv2022

Solving Linear Systems on a GPU with Hierarchically Off-Diagonal Low-Rank Approximations

We are interested in solving linear systems arising from three applications: (1) kernel methods in machine learning, (2) discretization of boundary integral equations from mathematical physics, and (3) Schur complements formed in the factorization of many large sparse matrices. The coefficient matrices are often data-sparse in the sense that their off-diagonal blocks have low numerical ranks; specifically, we focus on "hierarchically off-diagonal low-rank (HODLR)" matrices. We introduce algorithms for factorizing HODLR matrices and for applying the factorizations on a GPU. The algorithms leverage the efficiency of batched dense linear algebra, and they scale nearly linearly with the matrix size when the numerical ranks are fixed. The accuracy of the HODLR-matrix approximation is a tunable parameter, so we can construct high-accuracy fast direct solvers or low-accuracy robust preconditioners. Numerical results show that we can solve problems with several millions of unknowns in a couple of seconds on a single GPU.

math.NA↗

arXiv2022

Attention Hijacking in Trojan Transformers

Trojan attacks pose a severe threat to AI systems. Recent works on Transformer models received explosive popularity and the self-attentions are now indisputable. This raises a central question: Can we reveal the Trojans through attention mechanisms in BERTs and ViTs? In this paper, we investigate the attention hijacking pattern in Trojan AIs, \ie, the trigger token ``kidnaps'' the attention weights when a specific trigger is present. We observe the consistent attention hijacking pattern in Trojan Transformers from both Natural Language Processing (NLP) and Computer Vision (CV) domains. This intriguing property helps us to understand the Trojan mechanism in BERTs and ViTs. We also propose an Attention-Hijacking Trojan Detector (AHTD) to discriminate the Trojan AIs from the clean ones.

cs.LG↗

arXiv2022

Measurement of $e^{+}e^{-}\rightarrowπ^{+}π^{-}D^{+}D^{-}$ cross sections at center-of-mass energies from 4.190 to 4.946 GeV

Using data samples collected with the BESIII detector operating at the BEPCII storage ring, we measure the cross sections of the $e^{+}e^{-}\rightarrowπ^{+}π^{-}D^{+}D^{-}$ process at center-of-mass energies from 4.190 to 4.946 GeV with a partial reconstruction method. Two resonance structures are seen and the resonance parameters are determined from a fit to the cross section line shape. The first resonance we observe has a mass of (4373.1 $\pm$ 4.0 $\pm$ 2.2) MeV/$c^2$ and a width of (146.5 $\pm$ 7.4 $\pm$ 1.3) MeV, in agreement with those of the $Y(4390)$ state; the other resonance has a mass of (4706 $\pm$ 11 $\pm$ 4) MeV/$c^2$, a width of (45 $\pm$ 28 $\pm$ 9) MeV, and a statistical significance of $4.1$ standard deviations ($σ$). This is the first evidence for a vector state at this mass value. The spin-$3$ $D$-wave charmonium state $X(3842)$ is searched for through the $e^{+}e^{-}\rightarrowπ^{+}π^{-}X(3842)\rightarrowπ^{+}π^{-}D^{+}D^{-}$ process, and evidence with a significance of $4.2σ$ is found in the data samples with center-of-mass energies from 4.600 to 4.700 GeV.

hep-ex↗

arXiv2022

HOB-CNN: Hallucination of Occluded Branches with a Convolutional Neural Network for 2D Fruit Trees

Orchard automation has attracted the attention of researchers recently due to the shortage of global labor force. To automate tasks in orchards such as pruning, thinning, and harvesting, a detailed understanding of the tree structure is required. However, occlusions from foliage and fruits can make it challenging to predict the position of occluded trunks and branches. This work proposes a regression-based deep learning model, Hallucination of Occluded Branch Convolutional Neural Network (HOB-CNN), for tree branch position prediction in varying occluded conditions. We formulate tree branch position prediction as a regression problem towards the horizontal locations of the branch along the vertical direction or vice versa. We present comparative experiments on Y-shaped trees with two state-of-the-art baselines, representing common approaches to the problem. Experiments show that HOB-CNN outperform the baselines at predicting branch position and shows robustness against varying levels of occlusion. We further validated HOB-CNN against two different types of 2D trees, and HOB-CNN shows generalization across different trees and robustness under different occluded conditions.

cs.CV↗

arXiv2022

PixelFolder: An Efficient Progressive Pixel Synthesis Network for Image Generation

Pixel synthesis is a promising research paradigm for image generation, which can well exploit pixel-wise prior knowledge for generation. However, existing methods still suffer from excessive memory footprint and computation overhead. In this paper, we propose a progressive pixel synthesis network towards efficient image generation, coined as PixelFolder. Specifically, PixelFolder formulates image generation as a progressive pixel regression problem and synthesizes images via a multi-stage structure, which can greatly reduce the overhead caused by large tensor transformations. In addition, we introduce novel pixel folding operations to further improve model efficiency while maintaining pixel-wise prior knowledge for end-to-end regression. With these innovative designs, we greatly reduce the expenditure of pixel synthesis, e.g., reducing 89% computation and 53% parameters compared with the latest pixel synthesis method CIPS. To validate our approach, we conduct extensive experiments on two benchmark datasets, namely FFHQ and LSUN Church. The experimental results show that with much less expenditure, PixelFolder obtains new state-of-the-art (SOTA) performance on two benchmark datasets, i.e., 3.77 FID and 2.45 FID on FFHQ and LSUN Church, respectively.Meanwhile, PixelFolder is also more efficient than the SOTA methods like StyleGAN2, reducing about 72% computation and 31% parameters, respectively. These results greatly validate the effectiveness of the proposed PixelFolder.

cs.CV↗

arXiv2022

Topological spin texture of chiral edge states in photonic two-dimensional quantum walks

Topological insulators host topology-linked boundary states, whose spin and charge degrees of freedom could be exploited to design topological devices with enhanced functionality. We experimentally observe that dissipationless chiral edge states in a spin-orbit coupled anomalous Floquet topological phase exhibit topological spin texture on boundaries, realized via a two-dimensional quantum walk. Our experiment shows that, for a walker traveling around a closed loop along the boundary in real space, its spin evolves and winds through a great circle on the Bloch sphere, which implies that edge-spin texture has nontrivial winding. This winding is linked to the bulk Dirac Hamiltonian around the energy-gap opening point. Our experiment confirms that two-dimensional anomalous Floquet topological systems exhibit topological spin texture on the boundary, which could inspire novel topology-based spintronic phenomena and devices.

quant-ph↗

arXiv2022

SeqTR: A Simple yet Universal Network for Visual Grounding

In this paper, we propose a simple yet universal network termed SeqTR for visual grounding tasks, e.g., phrase localization, referring expression comprehension (REC) and segmentation (RES). The canonical paradigms for visual grounding often require substantial expertise in designing network architectures and loss functions, making them hard to generalize across tasks. To simplify and unify the modeling, we cast visual grounding as a point prediction problem conditioned on image and text inputs, where either the bounding box or binary mask is represented as a sequence of discrete coordinate tokens. Under this paradigm, visual grounding tasks are unified in our SeqTR network without task-specific branches or heads, e.g., the convolutional mask decoder for RES, which greatly reduces the complexity of multi-task modeling. In addition, SeqTR also shares the same optimization objective for all tasks with a simple cross-entropy loss, further reducing the complexity of deploying hand-crafted loss functions. Experiments on five benchmark datasets demonstrate that the proposed SeqTR outperforms (or is on par with) the existing state-of-the-arts, proving that a simple yet universal approach for visual grounding is indeed feasible. Source code is available at https://github.com/sean-zhuh/SeqTR.

cs.CV↗

arXiv2022

Latent Partition Implicit with Surface Codes for 3D Representation

Deep implicit functions have shown remarkable shape modeling ability in various 3D computer vision tasks. One drawback is that it is hard for them to represent a 3D shape as multiple parts. Current solutions learn various primitives and blend the primitives directly in the spatial space, which still struggle to approximate the 3D shape accurately. To resolve this problem, we introduce a novel implicit representation to represent a single 3D shape as a set of parts in the latent space, towards both highly accurate and plausibly interpretable shape modeling. Our insight here is that both the part learning and the part blending can be conducted much easier in the latent space than in the spatial space. We name our method Latent Partition Implicit (LPI), because of its ability of casting the global shape modeling into multiple local part modeling, which partitions the global shape unity. LPI represents a shape as Signed Distance Functions (SDFs) using surface codes. Each surface code is a latent code representing a part whose center is on the surface, which enables us to flexibly employ intrinsic attributes of shapes or additional surface properties. Eventually, LPI can reconstruct both the shape and the parts on the shape, both of which are plausible meshes. LPI is a multi-level representation, which can partition a shape into different numbers of parts after training. LPI can be learned without ground truth signed distances, point normals or any supervision for part partition. LPI outperforms the latest methods under the widely used benchmarks in terms of reconstruction accuracy and modeling interpretability. Our code, data and models are available at https://github.com/chenchao15/LPI.

cs.CV↗

arXiv2022

Learning Topological Interactions for Multi-Class Medical Image Segmentation

Deep learning methods have achieved impressive performance for multi-class medical image segmentation. However, they are limited in their ability to encode topological interactions among different classes (e.g., containment and exclusion). These constraints naturally arise in biomedical images and can be crucial in improving segmentation quality. In this paper, we introduce a novel topological interaction module to encode the topological interactions into a deep neural network. The implementation is completely convolution-based and thus can be very efficient. This empowers us to incorporate the constraints into end-to-end training and enrich the feature representation of neural networks. The efficacy of the proposed method is validated on different types of interactions. We also demonstrate the generalizability of the method on both proprietary and public challenge datasets, in both 2D and 3D settings, as well as across different modalities such as CT and Ultrasound. Code is available at: https://github.com/TopoXLab/TopoInteraction

cs.CV↗

arXiv2022

Synthesizing MR Image Contrast Enhancement Using 3D High-resolution ConvNets

\textit{Objective:} Gadolinium-based contrast agents (GBCAs) have been widely used to better visualize disease in brain magnetic resonance imaging (MRI). However, gadolinium deposition within the brain and body has raised safety concerns about the use of GBCAs. Therefore, the development of novel approaches that can decrease or even eliminate GBCA exposure while providing similar contrast information would be of significant use clinically. \textit{Methods:} In this work, we present a deep learning based approach for contrast-enhanced T1 synthesis on brain tumor patients. A 3D high-resolution fully convolutional network (FCN), which maintains high resolution information through processing and aggregates multi-scale information in parallel, is designed to map pre-contrast MRI sequences to contrast-enhanced MRI sequences. Specifically, three pre-contrast MRI sequences, T1, T2 and apparent diffusion coefficient map (ADC), are utilized as inputs and the post-contrast T1 sequences are utilized as target output. To alleviate the data imbalance problem between normal tissues and the tumor regions, we introduce a local loss to improve the contribution of the tumor regions, which leads to better enhancement results on tumors. \textit{Results:} Extensive quantitative and visual assessments are performed, with our proposed model achieving a PSNR of 28.24dB in the brain and 21.2dB in tumor regions. \textit{Conclusion and Significance:} Our results suggest the potential of substituting GBCAs with synthetic contrast images generated via deep learning. Code is available at \url{https://github.com/chenchao666/Contrast-enhanced-MRI-Synthesis

eess.IV↗

arXiv2022

Hiding Your Signals: A Security Analysis of PPG-based Biometric Authentication

Recently, physiological signal-based biometric systems have received wide attention. Unlike traditional biometric features, physiological signals can not be easily compromised (usually unobservable to human eyes). Photoplethysmography (PPG) signal is easy to measure, making it more attractive than many other physiological signals for biometric authentication. However, with the advent of remote PPG (rPPG), unobservability has been challenged when the attacker can remotely steal the rPPG signals by monitoring the victim's face, subsequently posing a threat to PPG-based biometrics. In PPG-based biometric authentication, current attack approaches mandate the victim's PPG signal, making rPPG-based attacks neglected. In this paper, we firstly analyze the security of PPG-based biometrics, including user authentication and communication protocols. We evaluate the signal waveforms, heart rate and inter-pulse-interval information extracted by five rPPG methods, including four traditional optical computing methods (CHROM, POS, LGI, PCA) and one deep learning method (CL_rPPG). We conducted experiments on five datasets (PURE, UBFC_rPPG, UBFC_Phys, LGI_PPGI, and COHFACE) to collect a comprehensive set of results. Our empirical studies show that rPPG poses a serious threat to the authentication system. The success rate of the rPPG signal spoofing attack in the user authentication system reached 0.35. The bit hit rate is 0.6 in inter-pulse-interval-based security protocols. Further, we propose an active defence strategy to hide the physiological signals of the face to resist the attack. It reduces the success rate of rPPG spoofing attacks in user authentication to 0.05. The bit hit rate was reduced to 0.5, which is at the level of a random guess. Our strategy effectively prevents the exposure of PPG signals to protect users' sensitive physiological data.

cs.CR↗

arXiv2022

Evidence for a neutral near-threshold structure in the $K^{0}_{\rm S}$ recoil-mass spectra in $e^+e^-\rightarrow K^{0}_{\rm S}D_s^+D^{-}$ and $e^+e^-\rightarrow K^{0}_{\rm S}D_s^{+}D^{-}$

We study the processes $e^+e^-\rightarrow K^{0}_{\rm S}D_s^+D^{*-}$ and $e^+e^-\rightarrow K^{0}_{\rm S}D_s^{*+}D^{-}$, as well as their charge conjugated processes, at five center-of-mass energies between 4.628~GeV and 4.699~GeV, using data samples corresponding to an integrated luminosity of $3.8~\rm{fb^{-1}}$ collected by the BESIII detector at the BEPCII storage ring. Based on a partial reconstruction technique, we find evidence of a structure near the thresholds for $D_s^+D^{*-}$ and $D_s^{*+}D^{-}$ production in the $K^{0}_{\rm S}$ recoil-mass spectrum, which we refer to as the $ Z_{cs}(3985)^0$. Fitting with a Breit-Wigner line shape, we find the mass of the structure to be $(3992.2\pm{1.7}\pm{1.6})~\rm{MeV}/c^2$ and the width to be $(7.7_{-3.8}^{+4.1}\pm{4.3})~\rm{MeV}$, where the first uncertainties are statistical and the second are systematic. The significance of the $ Z_{cs}(3985)^0$ signal is found to be $4.6σ$ including both the statistical and systematic uncertainty. We report the Born cross section multiplied by the branching fraction at different energy points. The mass of the $Z_{cs}(3985)^0$ is close to that of the $Z_{cs}(3985)^+$. Assuming SU(3) symmetry, the cross section of the neutral channel is consistent with that of the charged one. Hence, we conclude that the $Z_{cs}(3985)^0$ is the isospin partner of the $Z_{cs}(3985)^+$.

hep-ex↗

arXiv2022

A Study of the Attention Abnormality in Trojaned BERTs

Trojan attacks raise serious security concerns. In this paper, we investigate the underlying mechanism of Trojaned BERT models. We observe the attention focus drifting behavior of Trojaned models, i.e., when encountering an poisoned input, the trigger token hijacks the attention focus regardless of the context. We provide a thorough qualitative and quantitative analysis of this phenomenon, revealing insights into the Trojan mechanism. Based on the observation, we propose an attention-based Trojan detector to distinguish Trojaned models from clean ones. To the best of our knowledge, this is the first paper to analyze the Trojan mechanism and to develop a Trojan detector based on the transformer's attention.

cs.CR↗

arXiv2022

Measurement of the total and leptonic decay widths of the $J/ψ$ resonance with an energy scan method at BESIII

Using $e^+e^-$ annihilation data sets collected with the BESIII detector, we measure the cross sections of the processes $e^+e^- \to e^+e^-$ and $e^+e^- \to μ^+μ^-$ at fifteen center-of-mass energy points in the vicinity of the $J/ψ$ resonance. By a simultaneous fit to the measured, center-of-mass energy dependent cross sections of the two processes, the combined quantities $Γ_{ee} Γ_{ee} / Γ_{\rm tot}$ and $Γ_{ee} Γ_{μμ} / Γ_{\rm tot}$ are determined to be ($0.346 \pm 0.009$) and ($0.335 \pm 0.006$) keV, respectively, where $Γ_{ee}$, $Γ_{μμ}$, and $Γ_{\rm tot}$ are the electronic, muonic, and total decay widths of the $J/ψ$ resonance, respectively. Using the resultant $Γ_{ee} Γ_{μμ} / Γ_{\rm tot}$ and $Γ_{ee} Γ_{ee} / Γ_{\rm tot}$, the ratio $Γ_{ee} / Γ_{μμ}$ is calculated to be $1.031 \pm 0.015$, which is consistent with the expectation of lepton universality within about two standard deviations. Assuming lepton universality and using the branching fraction of the $J/ψ$ leptonic decay measured by BESIII in 2013, $Γ_{\rm tot}$ and $Γ_{ll}$ are determined to be ($93.0 \pm 2.1$) and ($5.56 \pm 0.11$) keV, respectively, where $Γ_{ll}$ is the average leptonic decay width of the $J/ψ$ resonance.

hep-ex↗

arXiv2022

Search for the semileptonic decay $D_s^+\to π^0e^+ν_e$

We present the first search for the semileptonic decay $D_s^+\to π^0 e^+ν_e$ using a data sample of electron-positron collisions recorded with the BESIII detector at center-of-mass energies between 4.178 and 4.226~GeV, corresponding to an integrated luminosity of 6.32~fb$^{-1}$. This decay is expected to be sensitive to $π^0$--$η$ mixing. No significant signal is observed. We set an upper limit of $6.4 \times 10^{-5}$ on the branching fraction at the $90\%$ confidence level.

hep-ex↗

arXiv2022

Observation of the double Dalitz decay $η'\to e^+e^-e^+e^-$

Based on $(10087 \pm 44)\times10^6$ $J/ψ$ events collected with the BESIII detector at BEPCII, the double Dalitz decay $η'\to e^+e^-e^+e^-$ is observed for the first time via the $J/ψ\toγη'$ decay process. The significance is found to be 5.7$σ$ with systematic uncertainties taken into consideration. Its branching fraction is determined to be $\mathcal{B}(η'\to e^+ e^- e^+ e^-) =(4.5\pm1.0(\mathrm{stat.})\pm0.5(\mathrm{sys.})) \times 10^{-6}$.

hep-ex↗

arXiv2022

Measurement of the branching fraction and decay asymmetry of $Λ\to nγ$

The radiative hyperon decay $Λ\to nγ$ is studied using $(10087\pm44)\times 10^6$ $J/ψ$ events collected with the BESIII detector operating at BEPCII. The absolute branching fraction of the decay $Λ\to nγ$ is determined with a significance of 5.6$σ$ to be $[0.832\pm0.038(\rm stat.)\pm0.054(\rm syst.)]\times10^{-3}$, which lies significantly below the current PDG value. By analyzing the joint angular distribution of the decay products, the first determination of the decay asymmetry $α_γ$ is reported with a value of $-0.16\pm0.10(\rm stat.)\pm0.05(\rm syst.)$.

hep-ex↗

arXiv2022

Cycle Representation Learning for Inductive Relation Prediction

In recent years, algebraic topology and its modern development, the theory of persistent homology, has shown great potential in graph representation learning. In this paper, based on the mathematics of algebraic topology, we propose a novel solution for inductive relation prediction, an important learning task for knowledge graph completion. To predict the relation between two entities, one can use the existence of rules, namely a sequence of relations. Previous works view rules as paths and primarily focus on the searching of paths between entities. The space of rules is huge, and one has to sacrifice either efficiency or accuracy. In this paper, we consider rules as cycles and show that the space of cycles has a unique structure based on the mathematics of algebraic topology. By exploring the linear structure of the cycle space, we can improve the searching efficiency of rules. We propose to collect cycle bases that span the space of cycles. We build a novel GNN framework on the collected cycles to learn the representations of cycles, and to predict the existence/non-existence of a relation. Our method achieves state-of-the-art performance on benchmarks.

cs.LG↗