Detection and characterization of breast tumours by time-domain scanning optical mammography

The paper gives a short overview of various methods of optical mammography, emphasizing scanning time-domain mammography. The results of a clinical study on time-domain optical mammography are reviewed, comprising 154 patients carrying a total of 102 carcinomas validated by histology. A visibility score attributed to each carcinoma as qualitative measure of tumour detectability indicates acceptable sensitivity but poor specificity for discrimination between malignant and benign lesions. Likewise, a multi-variate statistical analysis yields sensitivity and specificity between 80% and 85% for tumour detection and discrimination with respect to normal (healthy) breast tissue, but values less than 70% for discrimination between malignant and benign breast lesions, being too low to be of clinical relevance. For 87 of the 88 tumours detected retrospectively in both projection optical mammograms, optical properties and tissue parameters were derived based on the diffraction of photon density waves by a spherical inhomogeneity as forward model. Following injection of a bolus of indocyanine green as non-targeted absorbing contrast agent, dynamic contrast-enhanced time-domain optical mammography was carried out on a small number of patients, but no differences in wash-out kinetics of indocyanine green between tumours and healthy breast tissue were observed.     

Adam and the Ants: On the Influence of the Optimization Algorithm on the Detectability of DNN Watermarks

As training Deep Neural Networks (DNNs) becomes more expensive, the interest in protecting the ownership of the models with watermarking techniques increases. Uchida et al. proposed a digital watermarking algorithm that embeds the secret message into the model coefficients. However, despite its appeal, in this paper, we show that its efficacy can be compromised by the optimization algorithm being used. In particular, we found through a theoretical analysis that, as opposed to Stochastic Gradient Descent (SGD), the update direction given by Adam optimization strongly depends on the sign of a combination of columns of the projection matrix used for watermarking. Consequently, as observed in the empirical results, this makes the coefficients move in unison giving rise to heavily spiked weight distributions that can be easily detected by adversaries. As a way to solve this problem, we propose a new method called Block-Orthonormal Projections (BOP) that allows one to combine watermarking with Adam optimization with a minor impact on the detectability of the watermark and an increased robustness.     

Stability,stabilizability and detectability for Markov jump discrete-time linear systems with multiplicative noise in Hilbert spaces

In this article we discuss stability, stabilizability and detectability problems for Markov-jump discrete-time linear systems (MJDLSs) with multiplicative noise (MN) and countably infinite state space of the Markov chain. On the basis of a new solution representation formula, we give new deterministic characterizations of the stability and the detectability properties of MJDLSs with MN. These results are obtained using an operatorial approach and the properties of certain positive evolution operators defined on ordered Banach spaces of sequences of nuclear operators. Assuming detectability conditions and avoiding stochastic proofs, we prove that any global, nonnegative and bounded solution of the Riccati equation of control is stabilizing for the MJDLSs with MN and control. Finally, we apply our results to solve a linear quadratic optimal control problem. The theory is illustrated by an example.     

蛋白质组学质谱平台肽段可检测性预测研究进展

生物质谱是蛋白质组研究中的核心技术之一,可以实现大规模、高通量的蛋白质定性和定量分析。由于样品和实验过程自身的复杂性,质谱实验的重复性还存在一些问题,肽段鉴定和定量结果有很大的随机性,肽段的质谱检测概率问题在蛋白质组研究中,特别是定量蛋白质组研究中备受关注。本文总结了影响肽段可检测性的重要因素,分析了已经提出的计算预测方法,并对其在实验研究中的应用进行了综述。     

飞艇红外探测系统探测高超声速目标性能研究

以X-51A为例,研究了飞艇红外探测系统对临近空间高超声速目标的探测性能。首先,根据飞行器的飞行状态和飞行高度建立了临近空间高超声速目标不同波段的红外辐射特性模型,以及随高度变化的目标背景红外辐射强度模型;其次,综合考虑飞行器与飞艇高度、地球曲率及红外辐射在大气中传播的波段选择性等因素,建立了红外辐射在临近空间大气中传播的透过率模型;在此基础上,建立了飞艇红外探测系统对高超声速目标的探测距离模型。通过仿真得到了临近空间高超声速目标在不同飞行状态下3个波段的红外辐射强度随目标飞行高度变化的曲线,以及飞艇红外探测系统对飞行器在不同飞行状态下3个红外辐射波段的探测能力。研究结果表明:飞艇红外探测系统对高超声速目标的有效探测距离可以达到百公里量级;当飞行器飞行状态一定时,随着飞行器飞行高度的增加,系统对目标的探测距离先增大后减小;与长波波段相比,中短波波段的探测距离更大,并给出了临近空间飞艇应尽量布置在海拔高度大于18 km的高空中的部署建议。     

Numerical Solution for Linear-Quadratic Control Problems of Markov Jump Linear Systems and Weak Detectability Concept

A method for solving the linear-quadratic problem of Markov jump linear systems is developed in this paper, relying on the assumption of weak detectability. The concept of weak detectability generalizes previous concepts relevant to this class of systems, and most importantly, it allows us to revisit the quadratic control problem. In the main result of the paper, we show that, for weakly detectable systems, the solution obtained with the new method converges to the solution of the coupled algebraic Riccati equation that arises in the control problem if and only if the system is mean-square stabilizable. The paper shows how the concepts and the method involved are applied by means of numerical examples and comparisons.