一种基于叶分量分析的带有监督信息的在线学习方法

在一些模式识别应用中,具有类属信息的样本数量较少,此时监督学习算法会遇到小样本问题,导致分类器的识别精度大幅低于预期水平.基于叶分量分析,提出一种带监督信息的在线学习方法.该方法在训练过程进行监督学习,而在模式识别阶段能够在对输入样本进行分类的同时基于这些样本进行非监督在线学习,因此实现了监督学习与非监督学习的结合.在小本量情况下,在线学习可以弥补训练阶段监督学习的不足,仍能保证获得较高的识别精度.实验证明,该方法能够有效克服小样本问题.     

连续型2维图形矩阵码格式设计及掩模方案

2维条码目前已广泛应用于各个领域,2维条码的编码过程中有一个重要步骤称为掩模调制,其目的是为了限制生成图像各区域的平均灰度,以提高码字的易印制、易识别能力。但现有的掩模方法大多基于感性,掩模模式有限并缺乏规范的方法支持及客观的评估手段,难以在各种应用条件下取得最佳效果。本文的研究目的是对能够对2维条码的掩模进行一种客观的数学上的评估。本文结合作者自行提出的一种可承载不确定容量信息的连续型2维条码,从统计分布模型角度对掩模技术进行研究,通过一种归于正态分布的数学评估模型,最终引导出一种比传统方法更为客观、规范、可最优化的掩模方法,该技术还可以应用于QR码等2维编码规则中。     

社团挖掘和话题监控的互动模型研究

社团挖掘是Web信息挖掘领域的重要应用,而话题监控是文本信息研究领域的重要应用,目前这两种技术是各自独立的。为更好地应用于互联网形成的复杂社会网络,将这两种技术结合起来研究,发现了社团和话题之间的关系,创建了社团挖掘和话题监控的静态和动态互动模型,设计了社团挖掘、话题识别以及社团跟踪算法。     

基于局部子图匹配的SLAM 方法

针对现有的SLAM 解决方法在机器人被“绑架”时失效的问题,提出了基于局部子图匹配的方法．该 方法对现有的SLAM 解决构架进行了改进,提出交点最优匹配的特征相关算法,并且将奇异值分解方法引入机器人 定位．最后,在结构化环境下将本方法和基于扩展卡尔曼滤波器的方法进行比较,讨论了基于局部子图匹配的方法 在结构化环境中解决机器人“绑架”问题的有效性和可行性．     

Progress of Phosphate-based Polyanion Cathodes for Aqueous Rechargeable Zinc Batteries

Aqueous rechargeable zinc batteries (ARZBs) are recently prevailing devices that utilize the abundant Zn resources and the merits of aqueous electrolytes to become a competitive alternative for large-scale energy storage. Benefiting from the unique inductive effect and flexible structure, the past five years have experienced a diversiform of phosphate-based polyanion materials that are used as cathodes in ARZBs. In this review, the most recent advances in the Zn²⁺ storage mechanisms and electrolyte optimization of the phosphate-based cathodes of ARZBs, which mainly focus on vanadium/iron-based phosphates and their derivatives are presented. Furthermore, in addition to significant progress on polyanion phosphate-based cathode materials, the design strategies both for electrode materials and compatible electrolytes are also elaborated to improve the energy density and extend the cycling life of aqueous Zn/polyanion batteries.     

Activatable NIR-II Lanthanides-Polymetallic Oxomolybdate Hybrid Nanosensors for Monitoring Chemotherapy Induced Enteritis

Chemotherapy-induced enteritis is one of the side effects associated with cancer therapy, which significantly affects the treatment effect, but there is no effective clinical detection method that can early diagnose its occurrence and progression. Here, a series of second near-infrared window (NIR-II) hybrid nanosensors are designed that consisted of lanthanide nanoparticles and β-Mo2C-derived polymetallic oxomolybdate nanoclusters (Ln@POM). Based on the high sensitivity of POM to reactive oxygen species (ROS) closely related to chemotherapy-induced enteritis, the NIR-II luminescence intensity and lifetime of Ln@POM (Ln: Yb³⁺, Nd³⁺, Ho³⁺, Tm³⁺, Er³⁺) show excellent responsiveness to H₂O₂ and HClO with the detection limit down to 0.15 and 0.14 µm , respectively. Utilizing Nd@POM as a ROS-activated NIR-II nanosensor, the chemotherapeutic enteritis is successfully detected within 7 h after induction of chemotherapy drugs, which is significantly earlier than the gold standard method (immunohistochemistry, 24 h). These results demonstrate that the designed hybrid nanosensors are promising optical tools for the early diagnosis of ROS-related diseases.     

Regulating the Water Molecular in the Solvation Structure for Stable Zinc Metal Batteries

Rechargeable aqueous zinc batteries are promising energy storage devices because of their low cost, high safety, and high energy density. However, their performance is plagued by the unsatisfied cyclability due to the dendrite growth and hydrogen evolution reaction (HER) at the Zn anode. Herein, it is demonstrated that the concentrated hybrid aqueous/non-aqueous ZnCl₂ electrolytes constitute a peculiar chemical environment for not only the Zn-ions but also water molecules. The high concentration of chloride ions substitutes the H₂O molecular in the solvation structure of Zn²⁺, while the acetonitrile further interacts with H₂O to decrease its activity. The hybrid electrolytes both inhibit the dendrite formation and HER, enabling an ultrahigh average Coulombic efficiency of 99.9% in the Zn||Cu half-cell and a highly reversible Zn plating/stripping with a low overpotential of 21 mV. Using this hybrid electrolyte, the Zn||polytriphenylamine (PTPAn) full cell deliveres a high discharge capacity of 110 mAh g⁻¹, a high power density of 9200 W kg⁻¹ at 100 °C and maintains 85% of the capacity for over 6000 cycles at 10 °C. This study provides a deep understanding between the solvation structure and columbic efficiency of Zn anode, thus inspiring the development for stable Zn batteries.     

隐蔽图像后门攻击

目的 图像后门攻击是一种经典的对抗性攻击形式,后门攻击使被攻击的深度模型在正常情况下表现良好,而当隐藏的后门被预设的触发器激活时就会出现恶性结果。现有的后门攻击开始转向为有毒样本分配干净标签或在有毒数据中隐蔽触发器以对抗人类检查,但这些方法在视觉监督下很难同时具备这两种安全特性,并且它们的触发器可以很容易地通过统计分析检测出来。因此,提出了一种隐蔽有效的图像后门攻击方法。方法 首先通过信息隐藏技术隐蔽图像后门触发,使标签正确的中毒图像样本(标签不可感知性)和相应的干净图像样本看起来几乎相同(图像不可感知性)。其次,设计了一种全新的后门攻击范式,其中毒的源图像类别同时也是目标类。提出的后门攻击方法不仅视觉上是隐蔽的,同时能抵御经典的后门防御方法(统计不可感知性)。结果 为了验证方法的有效性与隐蔽性,在ImageN et、MNIST、CIFAR-10数据集上与其他3种方法进行了对比实验。实验结果表明,在3个数据集上,原始干净样本分类准确率下降均不到1%,中毒样本分类准确率都超过94%,并具备最好的图像视觉效果。另外,验证了所提出的触发器临时注入的任意图像样本都可以发起有效的后门攻击。结论 ...     

二进制代码安全分析综述

近几十年来,计算机硬件性能和软件规模技术已不同以往,其承载了人类社会生活生产的方方面面.计算机技术的飞速发展,也带来了人们对程序安全问题的关注.由于市面上存在着较多的遗留软件,这些软件无人维护且缺乏源代码支持,其安全性令人担忧,而二进制分析技术被用来解决该类软件问题.二进制分析技术根据其检测方式不同可分为:基于静态的二进制代码分析技术、基于动态的二进制代码分析技术和动静态混合的二进制代码分析技术.本文调研了近年来的二进制代码安全分析领域上相关研究,分别详细阐述了这3类技术中的主要方法,并对其关键技术进行详细介绍.     

On the multi-level embedding of crypto-image reversible data hiding

Crypto-space reversible image steganography has attracted increasing attention, given its ability to embed authentication information without revealing the image content. This paper presents an efficient reversible data hiding scheme for crypto-images: a block predictor is applied to compute prediction errors, then an adaptive block mapping algorithm is utilized to compress them whose amplitudes are within a small threshold, finally, this strategy can be applied in a multi-level manner to achieve a higher embedding capacity. Due to the correlations among adjacent pixels in the block, images can be sufficiently compressed to reserve abundant space for additional data embedding. Different from the prior arts, the compression code of the image is fully encrypted. Experimental results verify that the embedded data and original image can be perfectly recovered, the security is higher compared with the state-of-the-arts, and a significant improvement in the average embedding rate is achieved on two large-scale image datasets.