Searcharxiv⌕ Search

arXiv subjects

Jiansheng Yang

Publications and source records attributed to Jiansheng Yang.

27 records · Page 2Linked to original sources

Decoder-free Robustness Disentanglement without (Additional) Supervision

Adversarial Training (AT) is proposed to alleviate the adversarial vulnerability of machine learning models by extracting only robust features from the input, which, however, inevitably leads to severe accuracy reduction as it discards the non-robust yet useful features. This motivates us to preserve both robust and non-robust features and separate them with disentangled representation learning. Our proposed Adversarial Asymmetric Training (AAT) algorithm can reliably disentangle robust and non-robust representations without additional supervision on robustness. Empirical results show our method does not only successfully preserve accuracy by combining two representations, but also achieve much better disentanglement than previous work.

stat.ML↗

Imaging of buried obstacles in a two-layered medium with phaseless far-field data

The inverse problem we consider is to reconstruct the location and shape of buried obstacles in the lower half-space of an unbounded two-layered medium in two dimensions from phaseless far-field data. A main difficulty of this problem is that the translation invariance property of the modulus of the far field pattern is unavoidable, which is similar to the homogenous background medium case. Based on the idea of using superpositions of two plane waves with different directions as the incident fields, we first develop a direct imaging method to locate the position of small anomalies and give a theoretical analysis of the algorithm. Then a recursive Newton-type iteration algorithm in frequencies is proposed to reconstruct extended obstacles. Finally, numerical experiments are presented to illustrate the feasibility of our algorithms.

math.NA↗

Weakly singular corners always scatter

Assume that a bounded scatterer is embedded into an infinite homogeneous isotropic background medium in two dimensions. The refractive index function is supposed to be piecewise constant. If the scattering interface contains a weakly or strongly singular point, we prove that the scattered field cannot vanish identically. This particularly leads to the absence of non-scattering energies for piecewise analytic interfaces with a weakly singular point. Local uniqueness is obtained for shape identification problems in inverse medium scattering with a single far-field pattern.

math.AP↗

Bibliometric Index for Academic Leadership

Academic leadership is essential for research innovation and impact. Until now, there has been no dedicated measure of leadership by bibliometrics. Popular bibliometric indices are mainly based on academic output, such as the journal impact factor and the number of citations. Here we develop an academic leadership index based on readily available bibliometric data that is sensitive to not only academic output but also research efficiency. Our leadership index was tested in two studies on peer-reviewed journal papers by extramurally-funded principal investigators in the field of life sciences from China and the USA, respectively. The leadership performance of these principal investigators was quantified and compared relative to university rank and other factors. As a validation measure, we show that the highest average leadership index was achieved by principal investigators at top national universities in both countries. More interestingly, our results also indicate that on an individual basis, strong leadership and high efficiency are not necessarily associated with those at top-tier universities nor with the most funding. This leadership index may become the basis of a comprehensive merit system, facilitating academic evaluation and resource management.

cs.DL↗

Moment-based cosh-Hilbert Inversion and Its Applications in Single-photon Emission Computed Tomography

The inversion of cosh-Hilbert transform (CHT) is one of the most crucial steps for single-photon emission computed tomography with uniform attenuation from truncated projection data. Although the uniqueness of the CHT inversion had been proved \cite{Noo2007}, there is no exact and analytic inverse formula so far. Several approximated inversion algorithms of the CHT had been developed \cite{Noo2007}\cite{You2007}. In this paper, we proposed a new numerical moment-based inversion algorithm.

math.NA↗

Academic Ranking with Web Mining and Axiomatic Analysis

Academic ranking is a public topic, such as for universities, colleges, or departments, which has significant educational, administrative and social effects. Popular ranking systems include the US News & World Report (USNWR), the Academic Ranking of World Universities (ARWU), and others. The most popular observables for such ranking are academic publications and their citations. However, a rigorous, quantitative and thorough methodology has been missing for this purpose. With modern web technology and axiomatic bibliometric analysis, here we perform a feasibility study on Microsoft Academic Search metadata and obtain the first-of-its-kind ranking results for American departments of computer science. This approach can be extended for fully automatic intuitional and college ranking based on comprehensive data on Internet.

cs.DL↗

Axiomatic Quantification of Co-authors' Relative Contributions

Over the past decades, the competition for academic resources has gradually intensified, and worsened with the current financial crisis. To optimize the resource allocation, individualized assessment of research results is being actively studied but the current indices, such as the number of papers, the number of citations, the h-factor and its variants have limitations, especially their inability of determining co-authors' credit shares fairly. Here we establish an axiomatic system and quantify co-authors' relative contributions. Our methodology avoids subjective assignment of co-authors' credits using the inflated, fractional or harmonic methods, and provides a quantitative tool for scientific management such as funding and tenure decisions.

stat.AP↗

On the Upper Bounds of MDS Codes

Let $M_{q}(k)$ be the maximum length of MDS codes with parameters $q,k$. In this paper, the properties of $M_{q}(k)$ are studied, and some new upper bounds of $M_{q}(k)$ are obtained. Especially we obtain that $M_{q}(q-1)\leq q+2(q\equiv4(mod 6)), M_{q}(q-2)\leq q+1(q\equiv4(mod 6)), M_{q}(k)\leq q+k-3 (q=36(5s+1), s\in N$ and $ k=6,7).

math.CO↗

The Improvement of the Bound on Hash Family

In this paper, we study the bound on three kinds of hash family using the Singleton bound. To $ε-U(N; n, m)$ hash family, in the caes of $n>m^2>1$ and $1\geqε\geq ε_1(n, m)$, we get that the new bound is better. To $ε-\bigtriangleup U(N; n, m)$ hash family, in the case of $n>m>1$ and $1\geqε\geqε_3(n,m)$, the new bound is better. To $ε-SU(N; n, m)$ hash family, in the case of $n>2^m>2$ and $1\geqε\geq ε_4(n, m)$, we get that the new bound is better.

cs.IT↗