arXiv subjects

Siyu Chen

Publications and source records attributed to Siyu Chen.

At least 163 records · Page 9Linked to original sources

arXiv2021

Video Temporal Relationship Mining for Data-Efficient Person Re-identification

This paper is a technical report to our submission to the ICCV 2021 VIPriors Re-identification Challenge. In order to make full use of the visual inductive priors of the data, we treat the query and gallery images of the same identity as continuous frames in a video sequence. And we propose one novel post-processing strategy for video temporal relationship mining, which not only calculates the distance matrix between query and gallery images, but also the matrix between gallery images. The initial query image is used to retrieve the most similar image from the gallery, then the retrieved image is treated as a new query to retrieve its most similar image from the gallery. By iteratively searching for the closest image, we can achieve accurate image retrieval and finally obtain a robust retrieval sequence.

cs.CV↗

arXiv2021

Probing $μ$eV ALPs with future LHAASO observation of AGN $γ$-ray spectra

Axion-like particles (ALPs) are predicted in some well-motivated theories beyond the Standard Model. The TeV gamma-rays from active galactic nuclei (AGN) suffer attenuation by the pair production interactions with the cosmic background light (EBL/CMB) during its travel to the earth. The attenuation can be circumvented through photon-ALP conversions in the AGN and Galaxy magnetic-field, and a flux enhancement is expected to arise in the observed spectrum. In this work, we study the potential of the AGN gamma-ray spectrum for energy up to above 100\,TeV to probe ALP-parameter space at around $μ$eV, where the coupling $g_{aγ}$ is so far relatively weakly constrained. We find the nearby and bright sources, Mrk\,501, IC\,310 and M\,87, are suitable for our objective. Assuming an intrinsic spectrum exponential cutoff energy, we extrapolate the observed spectra of these sources up to above 100\,TeV by the models with/without ALPs. For $g_{aγ}\gtrsim 2\times$$10^{-11} \rm GeV^{-1}$ with $m_{a}\lesssim0.5\,μ$eV, the flux at around 100\,TeV predicted by the ALP model can be enhanced more than an order of magnitude than that from the standard absorption, and could be detected by LHAASO. Our result is subject to the uncertainty from the intrinsic cutoff energy and the AGN lobe (or plume) magnetic-field. For an optimistic estimation, the constraint can be improved to $g_{aγ}\gtrsim 2\times$$10^{-11} \rm GeV^{-1}$ with $m_{a}\lesssim1\,μ$eV. This require further observations on these sources by the forthcoming CTA, LHAASO, SWGO and so on.

astro-ph.HE↗

arXiv2021

Manipulation and braiding of Weyl nodes using symmetry-constrained phase transitions

Weyl semimetals are arguably the most paradigmatic form of a gapless topological phase. While the stability of Weyl nodes, as quantified by their topological charge, has been extensively investigated, recent interest has shifted to the manipulation of the location of these Weyl nodes for non-Abelian braiding. To accomplish this braiding it is necessary to drive significant Weyl node motion using realistic experimental parameter changes. We show that a family of phase transitions characterized by certain symmetry constraints impose that the Weyl nodes have to reorganise by a large amount, shifting from one high symmetry plane to another. Additionally, for a subset of pairs of nodes with nontrivial Euler class topology, this reorganization can only occur through a braiding process with adjacent nodes. As a result, the Weyl nodes are forced to move a large distance across the Brillouin zone and to braid, all driven by small temperature changes, a process we illustrate with Cd$_2$Re$_2$O$_7$. Our work opens up routes to readily manipulate Weyl nodes using only slight external parameter changes, paving the way for the practical realization of reciprocal space braiding.

cond-mat.mes-hall↗

arXiv2021

ForgeryNet: A Versatile Benchmark for Comprehensive Forgery Analysis

The rapid progress of photorealistic synthesis techniques has reached at a critical point where the boundary between real and manipulated images starts to blur. Thus, benchmarking and advancing digital forgery analysis have become a pressing issue. However, existing face forgery datasets either have limited diversity or only support coarse-grained analysis. To counter this emerging threat, we construct the ForgeryNet dataset, an extremely large face forgery dataset with unified annotations in image- and video-level data across four tasks: 1) Image Forgery Classification, including two-way (real / fake), three-way (real / fake with identity-replaced forgery approaches / fake with identity-remained forgery approaches), and n-way (real and 15 respective forgery approaches) classification. 2) Spatial Forgery Localization, which segments the manipulated area of fake images compared to their corresponding source real images. 3) Video Forgery Classification, which re-defines the video-level forgery classification with manipulated frames in random positions. This task is important because attackers in real world are free to manipulate any target frame. and 4) Temporal Forgery Localization, to localize the temporal segments which are manipulated. ForgeryNet is by far the largest publicly available deep face forgery dataset in terms of data-scale (2.9 million images, 221,247 videos), manipulations (7 image-level approaches, 8 video-level approaches), perturbations (36 independent and more mixed perturbations) and annotations (6.3 million classification labels, 2.9 million manipulated area annotations and 221,247 temporal forgery segment labels). We perform extensive benchmarking and studies of existing face forensics methods and obtain several valuable observations.

cs.CV↗

arXiv2021

Actor-Context-Actor Relation Network for Spatio-Temporal Action Localization

Localizing persons and recognizing their actions from videos is a challenging task towards high-level video understanding. Recent advances have been achieved by modeling direct pairwise relations between entities. In this paper, we take one step further, not only model direct relations between pairs but also take into account indirect higher-order relations established upon multiple elements. We propose to explicitly model the Actor-Context-Actor Relation, which is the relation between two actors based on their interactions with the context. To this end, we design an Actor-Context-Actor Relation Network (ACAR-Net) which builds upon a novel High-order Relation Reasoning Operator and an Actor-Context Feature Bank to enable indirect relation reasoning for spatio-temporal action localization. Experiments on AVA and UCF101-24 datasets show the advantages of modeling actor-context-actor relations, and visualization of attention maps further verifies that our model is capable of finding relevant higher-order relations to support action detection. Notably, our method ranks first in the AVA-Kineticsaction localization task of ActivityNet Challenge 2020, out-performing other entries by a significant margin (+6.71mAP). Training code and models will be available at https://github.com/Siyu-C/ACAR-Net.

cs.CV↗

arXiv2021

Managing network congestion with a tradable credit scheme: a trip-based MFD approach

This study investigates the efficiency and effectiveness of an area-based tradable credit scheme (TCS) using the trip-based Macroscopic Fundamental Diagram model for the morning commute problem. In the proposed TCS, the regulator distributes initial credits to all travelers and designs a time-varying and trip length specific credit tariff. Credits are traded between travelers and the regulator via a credit market, and the credit price is determined by the demand and supply of credits. The heterogeneity of travelers is considered in terms of desired arrival time, trip length and departure-time choice preferences. The TCS is incorporated into a day-to-day modelling framework to examine the travelers' learning process, the evolution of network, and the properties of the credit market. The existence of an equilibrium solution and the uniqueness of the credit price at the equilibrium state are established analytically. Furthermore, an open-source simulation framework is developed to validate the analytical properties of the proposed TCS and compare it with alternative control strategies in terms of mobility, network performance, and social welfare. Bayesian optimization is then adopted to optimize the credit toll scheme. The numerical results demonstrate that the proposed TCS outperforms the no-control case and matches the performance of the time-of-day pricing strategy, while maintaining revenue-neutral nature.

cs.GT↗

arXiv2020

Two-dimensional adaptive membranes with programmable water and ionic channels

Membranes are ubiquitous in nature with primary functions that include adaptive filtering and selective transport of chemical and molecular species. Being critical to cellular functions, they are also fundamental in many areas of science and technology. Of particular importance are the adaptive and programmable membranes that can change their permeability or selectivity depending on the environment. Here, we explore implementation of such biological functions in artificial membranes and demonstrate two dimensional self assembled heterostructures of graphene oxide and polyamine macromolecules, forming a network of ionic channels that exhibit regulated permeability of water and monovalent ions. This permeability can be tuned by a change of pH or the presence of certain ions. Unlike traditional membranes, the regulation mechanism reported here relies on specific interactions between the membranes internal components and ions. This allows fabrication of membranes with programmable, predetermined permeability and selectivity, governed by the choice of components, their conformation and their charging state.

cond-mat.mtrl-sci↗

arXiv2020

Maximum a posteriori signal recovery for optical coherence tomography angiography image generation and denoising

Optical coherence tomography angiography (OCTA) is a novel and clinically promising imaging modality to image retinal and sub-retinal vasculature. Based on repeated optical coherence tomography (OCT) scans, intensity changes are observed over time and used to compute OCTA image data. OCTA data are prone to noise and artifacts caused by variations in flow speed and patient movement. We propose a novel iterative maximum a posteriori signal recovery algorithm in order to generate OCTA volumes with reduced noise and increased image quality. This algorithm is based on previous work on probabilistic OCTA signal models and maximum likelihood estimates. Reconstruction results using total variation minimization and wavelet shrinkage for regularization were compared against an OCTA ground truth volume, merged from six co-registered single OCTA volumes. The results show a significant improvement in peak signal-to-noise ratio and structural similarity. The presented algorithm brings together OCTA image generation and Bayesian statistics and can be developed into new OCTA image generation and denoising algorithms.

eess.IV↗

arXiv2020

Parametrized classifiers for optimal EFT sensitivity

We study unbinned multivariate analysis techniques, based on Statistical Learning, for indirect new physics searches at the LHC in the Effective Field Theory framework. We focus in particular on high-energy $ZW$ production with fully leptonic decays, modeled at different degrees of refinement up to NLO in QCD. We show that a considerable gain in sensitivity is possible compared with current projections based on binned analyses. As expected, the gain is particularly significant for those operators that display a complex pattern of interference with the Standard Model amplitude. The most effective method is found to be the "Quadratic Classifier" approach, an improvement of the standard Statistical Learning classifier where the quadratic dependence of the differential cross section on the EFT Wilson coefficients is built-in and incorporated in the loss function. We argue that the Quadratic Classifier performances are nearly statistically optimal, based on a rigorous notion of optimality that we can establish for an approximate analytic description of the $ZW$ process.

hep-ph↗

arXiv2020

RANDOM MASK: Towards Robust Convolutional Neural Networks

Robustness of neural networks has recently been highlighted by the adversarial examples, i.e., inputs added with well-designed perturbations which are imperceptible to humans but can cause the network to give incorrect outputs. In this paper, we design a new CNN architecture that by itself has good robustness. We introduce a simple but powerful technique, Random Mask, to modify existing CNN structures. We show that CNN with Random Mask achieves state-of-the-art performance against black-box adversarial attacks without applying any adversarial training. We next investigate the adversarial examples which 'fool' a CNN with Random Mask. Surprisingly, we find that these adversarial examples often 'fool' humans as well. This raises fundamental questions on how to define adversarial examples and robustness properly.

cs.CV↗

arXiv2020

Thermal Regelation of a Single Particle

In this paper we study the process termed thermal regelation at single particle level. We experimentally extract the physical parameters that determine the interactions between ice and the silica particle for the regelation process. As the premelted layer around the particle plays a key role here, we use the extracted parameters to calculate the thickness of this thin premelted layer, at different given particle sizes.

cond-mat.soft↗

arXiv2020

1st place solution for AVA-Kinetics Crossover in AcitivityNet Challenge 2020

This technical report introduces our winning solution to the spatio-temporal action localization track, AVA-Kinetics Crossover, in ActivityNet Challenge 2020. Our entry is mainly based on Actor-Context-Actor Relation Network. We describe technical details for the new AVA-Kinetics dataset, together with some experimental results. Without any bells and whistles, we achieved 39.62 mAP on the test set of AVA-Kinetics, which outperforms other entries by a large margin. Code will be available at: https://github.com/Siyu-C/ACAR-Net.

cs.CV↗

arXiv2019

Gram-Gauss-Newton Method: Learning Overparameterized Neural Networks for Regression Problems

First-order methods such as stochastic gradient descent (SGD) are currently the standard algorithm for training deep neural networks. Second-order methods, despite their better convergence rate, are rarely used in practice due to the prohibitive computational cost in calculating the second-order information. In this paper, we propose a novel Gram-Gauss-Newton (GGN) algorithm to train deep neural networks for regression problems with square loss. Our method draws inspiration from the connection between neural network optimization and kernel regression of neural tangent kernel (NTK). Different from typical second-order methods that have heavy computational cost in each iteration, GGN only has minor overhead compared to first-order methods such as SGD. We also give theoretical results to show that for sufficiently wide neural networks, the convergence rate of GGN is \emph{quadratic}. Furthermore, we provide convergence guarantee for mini-batch GGN algorithm, which is, to our knowledge, the first convergence result for the mini-batch version of a second-order method on overparameterized neural networks. Preliminary experiments on regression tasks demonstrate that for training standard networks, our GGN algorithm converges much faster and achieves better performance than SGD.

cs.LG↗

arXiv2019

Deep Neural Network Based Hyperspectral Pixel Classification With Factorized Spectral-Spatial Feature Representation

Deep learning has been widely used for hyperspectral pixel classification due to its ability of generating deep feature representation. However, how to construct an efficient and powerful network suitable for hyperspectral data is still under exploration. In this paper, a novel neural network model is designed for taking full advantage of the spectral-spatial structure of hyperspectral data. Firstly, we extract pixel-based intrinsic features from rich yet redundant spectral bands by a subnetwork with supervised pre-training scheme. Secondly, in order to utilize the local spatial correlation among pixels, we share the previous subnetwork as a spectral feature extractor for each pixel in a patch of image, after which the spectral features of all pixels in a patch are combined and feeded into the subsequent classification subnetwork. Finally, the whole network is further fine-tuned to improve its classification performance. Specially, the spectral-spatial factorization scheme is applied in our model architecture, making the network size and the number of parameters great less than the existing spectral-spatial deep networks for hyperspectral image classification. Experiments on the hyperspectral data sets show that, compared with some state-of-art deep learning methods, our method achieves better classification results while having smaller network size and less parameters.

cs.CV↗

arXiv2019

A Deep Optimization Approach for Image Deconvolution

In blind image deconvolution, priors are often leveraged to constrain the solution space, so as to alleviate the under-determinacy. Priors which are trained separately from the task of deconvolution tend to be instable, or ineffective. We propose the Golf Optimizer, a novel but simple form of network that learns deep priors from data with better propagation behavior. Like playing golf, our method first estimates an aggressive propagation towards optimum using one network, and recurrently applies a residual CNN to learn the gradient of prior for delicate correction on restoration. Experiments show that our network achieves competitive performance on GoPro dataset, and our model is extremely lightweight compared with the state-of-art works.

cs.CV↗

arXiv2018

Construction D$^\prime$ Lattices from Quasi-Cyclic Low-Density Parity-Check Codes

Recently, Branco da Silva and Silva described an efficient encoding and decoding algorithm for Construction D$^\prime$ lattices. Using their algorithm, we propose a Construction D$^\prime$ lattice based on binary quasi-cyclic low-density parity-check (QC-LPDC) codes and single parity-check product codes. The underlying codes designed by the balanced-distances rule contribute in a balanced manner to the squared minimum distance of the constructed lattice, which results in a high lattice coding gain. The proposed lattice based on IEEE 802.16e QC-LDPC codes is shown to provide competitive error-rate performance on the power-unconstrained additive white Gaussian noise channel.

cs.IT↗

arXiv2018

Estimating the Frequency Coupling Matrix from Network Measurements

Power converters are present in increasing numbers in the electric power grid. They are a major source of harmonic currents and voltages, which can reduce power quality and trip protection devices. The frequency coupling matrix (FCM) is a general technique for modeling converter harmonics. It can be obtained through experimental characterization or, given a converter's internal parameters, direct calculation. In this paper, we estimate FCMs from network measurements. We give a novel harmonic reduction theorem for computing equivalent, virtual FCMs for unobservable portions of a network. We estimate FCMs and harmonic line admittances from PMU measurements, and give an efficient online version for the FCM problem. We test our approaches on a numerical example, and show that the estimation error is low under noisy observations.

eess.SY↗

arXiv2018

Spectral Image Visualization Using Generative Adversarial Networks

Spectral images captured by satellites and radio-telescopes are analyzed to obtain information about geological compositions distributions, distant asters as well as undersea terrain. Spectral images usually contain tens to hundreds of continuous narrow spectral bands and are widely used in various fields. But the vast majority of those image signals are beyond the visible range, which calls for special visualization technique. The visualizations of spectral images shall convey as much information as possible from the original signal and facilitate image interpretation. However, most of the existing visualizatio methods display spectral images in false colors, which contradict with human's experience and expectation. In this paper, we present a novel visualization generative adversarial network (GAN) to display spectral images in natural colors. To achieve our goal, we propose a loss function which consists of an adversarial loss and a structure loss. The adversarial loss pushes our solution to the natural image distribution using a discriminator network that is trained to differentiate between false-color images and natural-color images. We also use a cycle loss as the structure constraint to guarantee structure consistency. Experimental results show that our method is able to generate structure-preserved and natural-looking visualizations.

cs.CV↗