arXiv subjects

Xiaolin Chen

Publications and source records attributed to Xiaolin Chen.

At least 37 records · Page 2Linked to original sources

arXiv2023

Dual Semantic Knowledge Composed Multimodal Dialog Systems

Textual response generation is an essential task for multimodal task-oriented dialog systems.Although existing studies have achieved fruitful progress, they still suffer from two critical limitations: 1) focusing on the attribute knowledge but ignoring the relation knowledge that can reveal the correlations between different entities and hence promote the response generation}, and 2) only conducting the cross-entropy loss based output-level supervision but lacking the representation-level regularization. To address these limitations, we devise a novel multimodal task-oriented dialog system (named MDS-S2). Specifically, MDS-S2 first simultaneously acquires the context related attribute and relation knowledge from the knowledge base, whereby the non-intuitive relation knowledge is extracted by the n-hop graph walk. Thereafter, considering that the attribute knowledge and relation knowledge can benefit the responding to different levels of questions, we design a multi-level knowledge composition module in MDS-S2 to obtain the latent composed response representation. Moreover, we devise a set of latent query variables to distill the semantic information from the composed response representation and the ground truth response representation, respectively, and thus conduct the representation-level semantic regularization. Extensive experiments on a public dataset have verified the superiority of our proposed MDS-S2. We have released the codes and parameters to facilitate the research community.

cs.CL↗

arXiv2021

Fed-EINI: An Efficient and Interpretable Inference Framework for Decision Tree Ensembles in Federated Learning

The increasing concerns about data privacy and security drive an emerging field of studying privacy-preserving machine learning from isolated data sources, i.e., federated learning. A class of federated learning, vertical federated learning, where different parties hold different features for common users, has a great potential of driving a great variety of business cooperation among enterprises in many fields. In machine learning, decision tree ensembles such as gradient boosting decision trees (GBDT) and random forest are widely applied powerful models with high interpretability and modeling efficiency. However, stateof-art vertical federated learning frameworks adapt anonymous features to avoid possible data breaches, makes the interpretability of the model compromised. To address this issue in the inference process, in this paper, we firstly make a problem analysis about the necessity of disclosure meanings of feature to Guest Party in vertical federated learning. Then we find the prediction result of a tree could be expressed as the intersection of results of sub-models of the tree held by all parties. With this key observation, we protect data privacy and allow the disclosure of feature meaning by concealing decision paths and adapt a communication-efficient secure computation method for inference outputs. The advantages of Fed-EINI will be demonstrated through both theoretical analysis and extensive numerical results. We improve the interpretability of the model by disclosing the meaning of features while ensuring efficiency and accuracy.

cs.LG↗

arXiv2016

Entry effects of droplet in a micro confinement: implications for deformation-based CTC microfiltration

Deformation based circulating tumor cell (CTC) microchips are a representative diagnostic device for early cancer detection. This type of device usually involves a process of CTC trapping in a confined microgeometry. Further understanding of the CTC flow regime, as well as the threshold passing-through pressure is key to the design of deformation based CTC filtration devices. In the present numerical study, we investigate the transitional deformation and pressure signature from surface tension dominated flow to viscous shear stress dominated flow using a droplet model. Regarding whether CTC fully blocks the channel inlet, we observe two flow regimes: CTC squeezing and shearing regime. By studying the relation of CTC deformation at the exact critical pressure point for increasing inlet velocity, three different types of cell deformation are observed: 1) hemispherical front, 2) parabolic front, and 3) elongated CTC co-flowing with carrier media. Focusing on the circular channel, we observe a first increasing and then decreasing critical pressure change with increasing flow rate. By pressure analysis, the concept of optimum velocity is proposed to explain the behavior of CTC filtration and design optimization of CTC filter. Similar behavior is also observed in channels with symmetrical cross sessions like square and triangular but not in rectangular channels which only results in decreasing critical pressure.

cond-mat.soft↗

arXiv2016

Deformability-based circulating tumor cell separation with conical-shaped microfilters: concept, optimization and design criteria

The ability of detecting and separating CTCs can play a key role in early cancer detection and treatment. In recent years, there has been growing interest in using deformability-based CTC separation microfilters due to their simplicity and low cost. Most of previous studies in this area are mainly based on experimental work. Although experimental research provides useful insights in designing CTC separation devices, there is still a lack of design guidelines based on fundamental understandings of the cell separation process in the filers. While experimental efforts face challenges especially microfabrication difficulties, we adopt numerical simulation here to study conical-shaped microfilters using deformability difference between CTCs and blood cells for separation process. We use liquid drop model for modeling a CTC passing through such microfilters. The accuracy of the model in predicting the pressure signature of the system is validated by comparing with previous experiments. Pressure-deformability analysis of the cell going through the channel is then carried out in detail in order to better understand how a CTC behaves throughout the filtration process. Different system design criteria such as system throughput and unclogging of the system are discussed. Specifically, pressure behavior under different system throughput is analyzed. Regarding the unclogging issue, we define pressure ratio as a key parameter representing the ability to overcome clogging in such CTC separation devices and investigate the effect of conical angle on the optimum pressure ratio. Finally, the effect of unclogging applied pressure on the system performance is examined. Our study provides detailed understandings of the cell separation process and its characteristics, which can be used for developing more efficient CTC separation devices.

q-bio.CB↗

arXiv2015

$3$-Regular mixed graphs with optimum Hermitian energy

Let $G$ be a simple undirected graph, and $G^ϕ$ be a mixed graph of $G$ with the generalized orientation $ϕ$ and Hermitian-adjacency matrix $H(G^ϕ)$. Then $G$ is called the underlying graph of $G^ϕ$. The Hermitian energy of the mixed graph $G^ϕ$, denoted by $\mathcal{E}_H(G^ϕ)$, is defined as the sum of all the singular values of $H(G^ϕ)$. A $k$-regular mixed graph on $n$ vertices having Hermitian energy $n\sqrt{k}$ is called a $k$-regular optimum Hermitian energy mixed graph. In this paper, we first focus on the problem proposed by Liu and Li [J. Liu, X. Li, Hermitian-adjacency matrices and Hermitian energies of mixed graphs, Linear Algebra Appl. 466(2015), 182--207] of determining all the $3$-regular connected optimum Hermitian energy mixed graphs. We then prove that optimum Hermitian energy oriented graphs with underlying graph hypercube are unique (up to switching equivalence).

math.CO↗

arXiv2014

The matching energy of random graphs

The matching energy of a graph was introduced by Gutman and Wagner, which is defined as the sum of the absolute values of the roots of the matching polynomial of the graph. For the random graph $G_{n,p}$ of order $n$ with fixed probability $p\in (0,1)$, Gutman and Wagner [I. Gutman, S. Wagner, The matching energy of a graph, Discrete Appl. Math. 160(2012), 2177--2187] proposed a conjecture that the matching energy of $G_{n,p}$ converges to $\frac{8\sqrt{p}}{3π}n^{\frac{3}{2}}$ almost surely. In this paper, using analysis method, we prove that the conjecture is true.

math.CO↗

arXiv2014

Solution to a conjecture on the maximum skew-spectral radius of odd-cycle graphs

Let $G$ be a simple graph with no even cycle, called an odd-cycle graph. Cavers et al. [Cavers et al. Skew-adjacency matrices of graphs, Linear Algebra Appl. 436(2012), 4512--1829] showed that the spectral radius of $G^σ$ is the same for every orientation $σ$ of $G$, and equals the maximum matching root of $G$. They proposed a conjecture that the graphs which attain the maximum skew spectral radius among the odd-cycle graphs $G$ of order $n$ are isomorphic to the odd-cycle graph with one vertex degree $n-1$ and size $m=\lfloor 3(n-1)/2\rfloor$. This paper, by using the Kelmans transformation, gives a proof of the conjecture. Moreover, sharp upper bounds of the maximum matching roots of the odd-cycle graphs with given order $n$ and size $m$ are given and extremal graphs are characterized.

math.CO↗

arXiv2014

Lower bounds of the skew spectral radii and skew energy of oriented graphs

Let $G$ be a graph with maximum degree $Δ$, and let $G^σ$ be an oriented graph of $G$ with skew adjacency matrix $S(G^σ)$. The skew spectral radius $ρ_s(G^σ)$ of $G^σ$ is defined as the spectral radius of $S(G^σ)$. The skew spectral radius has been studied, but only few results about its lower bound are known. This paper determines some lower bounds of the skew spectral radius, and then studies the oriented graphs whose skew spectral radii attain the lower bound $\sqrtΔ$. Moreover, we apply the skew spectral radius to the skew energy of oriented graphs, which is defined as the sum of the norms of all the eigenvalues of $S(G^σ)$, and denoted by $\mathcal{E}_s(G^σ)$. As results, we obtain some lower bounds of the skew energy, which improve the known lower bound obtained by Adiga et al.

math.CO↗

arXiv2013

More on the skew-spectra of bipartite graphs and Cartesian products of graphs

Given a graph $G$, let $G^σ$ be an oriented graph of $G$ with the orientation $σ$ and skew-adjacency matrix $S(G^σ)$. Then the spectrum of $S(G^σ)$ is called the skew-spectrum of $G^σ$, denoted by $Sp_S(G^σ)$. It is known that a graph $G$ is bipartite if and only if there is an orientation $σ$ of $G$ such that $Sp_S(G^σ)=iSp(G)$. In [D. Cui, Y. Hou, On the skew spectra of Cartesian products of graphs, Electron. J. Combin. 20(2013), #P19], Cui and Hou conjectured that such orientation of a bipartite graph is unique under switching-equivalence. In this paper, we prove that the conjecture is true. Moreover, we give an orientation of the Cartesian product of a bipartite graph and a graph, and then determine the skew-spectrum of the resulting oriented product graph, which generalizes Cui and Hou's result, and can be used to construct more oriented graphs with maximum skew energy.

math.CO↗

arXiv2013

4-Regular oriented graphs with optimum skew energy

Let $G$ be a simple undirected graph, and $G^σ$ be an oriented graph of $G$ with the orientation $σ$ and skew-adjacency matrix $S(G^σ)$. The skew energy of the oriented graph $G^σ$, denoted by $\mathcal{E}_S(G^σ)$, is defined as the sum of the absolute values of all the eigenvalues of $S(G^σ)$. In this paper, we characterize the underlying graphs of all 4-regular oriented graphs with optimum skew energy and give orientations of these underlying graphs such that the skew energy of the resultant oriented graphs indeed attain optimum. It should be pointed out that there are infinitely many 4-regular connected optimum skew energy oriented graphs, while the 3-regular case only has two graphs: $K_4$ the complete graph on 4 vertices and $Q_3$ the hypercube.

math.CO↗

arXiv2013

The skew energy of random oriented graphs

Given a graph $G$, let $G^σ$ be an oriented graph of $G$ with the orientation $σ$ and skew-adjacency matrix $S(G^σ)$. The skew energy of the oriented graph $G^σ$, denoted by $\mathcal{E}_S(G^σ)$, is defined as the sum of the absolute values of all the eigenvalues of $S(G^σ)$. In this paper, we study the skew energy of random oriented graphs and formulate an exact estimate of the skew energy for almost all oriented graphs by generalizing Wigner's semicircle law. Moreover, we consider the skew energy of random regular oriented graphs $G_{n,d}^σ$, and get an exact estimate of the skew energy for almost all regular oriented graphs.

math.CO↗

arXiv2013

Note on packing of edge-disjoint spanning trees in sparse random graphs

The \emph{spanning tree packing number} of a graph $G$ is the maximum number of edge-disjoint spanning trees contained in $G$. Let $k\geq 1$ be a fixed integer. Palmer and Spencer proved that in almost every random graph process, the hitting time for having $k$ edge-disjoint spanning trees equals the hitting time for having minimum degree $k$. In this paper, we prove that for any $p$ such that $(\log n+ω(1))/n\leq p\leq (1.1\log n)/n$, almost surely the random graph $G(n,p)$ satisfies that the spanning tree packing number is equal to the minimum degree. Note that this bound for $p$ will allow the minimum degree to be a function of $n$, and in this sense we improve the result of Palmer and Spencer. Moreover, we also obtain that for any $p$ such that $p\geq (51\log n)/n$, almost surely the random graph $G(n,p)$ satisfies that the spanning tree packing number is less than the minimum degree.

math.CO↗

arXiv2012

Rainbow $k$-connectivity of random bipartite graphs

A path in an edge-colored graph $G$ is called a rainbow path if no two edges of the path are colored the same. The minimum number of colors required to color the edges of $G$ such that every pair of vertices are connected by at least $k$ internally vertex-disjoint rainbow paths is called the rainbow $k$-connectivity of the graph $G$, denoted by $rc_k(G)$. For the random graph $G(n,p)$, He and Liang got a sharp threshold function for the property $rc_k(G(n,p))\leq d$. In this paper, we extend this result to the case of random bipartite graph $G(m,n,p)$.

math.CO↗

arXiv2012

Acoustophoresis in Variously Shaped Liquid Droplets

The ability to precisely trap, transport and manipulate micrometer-sized objects, including biological cells, DNA-coated microspheres and microorganisms, is very important in life science studies and biomedical applications. In this study, acoustic radiation force in an ultrasonic standing wave field is used for micro-objects manipulation, a technique termed as acoustophoresis. Free surfaces of liquid droplets are used as sound reflectors to confine sound waves inside the droplets. Two techniques were developed for precise control of droplet shapes: edge pinning and hydrophilic/hydrophobic interface pinning. For all tested droplet shapes, including circular, annular and rectangular, our experiments show that polymer micro particles can be manipulated by ultrasound and form into a variety of patterns, for example, concentric rings and radial lines in an annular droplet. The complexity of the pattern increases with increasing frequency, and the observations are in line with simulation results. The acoustic manipulation technique developed here has the potential to be integrated into a more complex on-chip microfluidic circuit. Especially because our method is well compatible with electrowetting technology, which is a powerful tool for manipulating droplets with free surfaces, the combination of the two methods can provide more versatile manipulation abilities and may bring a wealth of novel applications. In the end, we demonstrate for the first time that acoustophoresis can be used for manipulating Caenorhabditis elegans.

physics.flu-dyn↗

arXiv2010

A solution to a conjecture on the rainbow connection number

For a graph $G$, Chartrand et al. defined the rainbow connection number $rc(G)$ and the strong rainbow connection number $src(G)$ in "G. Charand, G.L. John, K.A. Mckeon, P. Zhang, Rainbow connection in graphs, Mathematica Bohemica, 133(1)(2008) 85-98". They raised the following conjecture: for two given positive $a$ and $b$, there exists a connected graph $G$ such that $rc(G)=a$ and $src(G)=b$ if and only if $a=b\in\{1,2\}$ or $ 3\leq a\leq b$". In this short note, we will show that the conjecture is true.

math.CO↗

arXiv2004

$SU(3)_F$ Meson Mass Formula from Random Phase Approximation

We present a $SU(3)_F$ meson mass formula from random phase approximation (RPA). Both the mesons of ground-state pseudoscalar octet and the ground-state vector octet are described quite well in this mass formula. We also estimate the current and constituent quark masses from the na\"ıve quark model with the PCAC relation.

hep-ph↗

arXiv2003

Chiral Symmetry in an Extended Constituent Quark Potential Model

The chiral symmetry is applied to an extended constituent quark potential model. With random phase approximation (RPA), the model comprises sea quark excitations. To obtain the pseudoscalar $π$ meson as a Goldstone boson, the quark effective potentials are modified in the model to account for the dynamical breaking of chiral symmetry. Also the vector $ρ$ meson is calculated and the KSRF relation about $π$ and $ρ$ meson decay constants is derived in the model.

hep-ph↗

arXiv2002

Extended Quark Potential Model from Random Phase Approximation

The quark potential model is extended to include the sea quark excitation using the random phase approximation (RPA). The effective quark interaction preserves the important Quantum Chromodynamics (QCD) properties -- chiral symmetry and confinement simultaneously. A primary qualitive analysis shows that the $π$ meson as a well-known typical Goldstone boson and the other mesons made up of valence $q\bar{q}$ quark pair such as the $ρ$ meson can also be described in this extended quark potential model.

hep-ph↗