ID 原文 译文
23605 反馈环路是模拟电路中有效容错的电路结构。反馈电路也因其存储性能而被广泛使用于数字电路的时序电路中，但是反馈电路在数字电路的组合电路的稳定特性鲜少被人研究，尤其是低功耗应用。 Feedback structure is an efficient topology for noise-reducing in analog circuit while the cyclic circuit is widely used in digital circuit only for sequential circuit design due to its data-keeping property. However, few works study the reliability of the feedback structure for combinational circuits especially for the low power application.
23606 以马氏随机场为理论的MRF 电路以其低功耗下的高稳定性得到研究和关注，但其电路的反馈结构缺乏理论支持和依据，因此马氏随机场电路的容错特性未被清晰得以解释。 Many researchers pay their attentions to Markov Random Field (MRF) theory based circuits, which can operate in ultra-low supply voltage with high noise-immune. However, the MRF based circuit design methodology has a lack of the proof of the final feedback structures. Thus the reliability of MRF based feedback structures is not explained clearly.
23607 该文以利用概率 CMOS 建模概率门来分析 MRF 核心反馈环 NAND-NAND，从理论上证明了反馈电路输出的计算正确概率具有递增且上有界的特点，并数学证明了 MRF 的核心反馈环电路具有优于传统 CMOS 电路的容错性能。 This paper uses the probabilistic CMOS model to analysis the NAND-NAND based feedback structure. The probability boundedness and increasing monotonicity properties of feedback structure are proved. Besides, it is proved that the feedback structure of MRF can achieve higher probability than the traditional design.
23608 其理论推导结果与测试结果呈现一致性。 In measurement, the result can support of proof and analysis.
23609 为了提高脉冲噪声下盲均衡器的性能，该文提出一种基于概率密度函数匹配与分数低阶矩的并行盲均衡算法。 In order to improve the performance of the blind equalizer under impulsive noise environments, a novel concurrent blind equalization algorithm based on probability density function matching and fractional lower order moments is presented.
23610 首先采用概率密度函数匹配的思想进行盲均衡，充分利用其收敛速度快的优势。 This algorithm uses the idea of probability density function matching at the beginning, and makes full use of the advantage of its fast convergence speed.
23611 为了解决此均衡过程中引起的相位信息损失以及抑制脉冲噪声能力差的问题，又以并行的方式结合判决信号的分数低阶矩，并以此作为代价函数来共同更新盲均衡器的权向量，进一步提高了算法在脉冲噪声下的收敛速度与收敛精度。 In order to solve the problems of the phase information loss and incapability of suppressing impulse noise, this paper combines the fractional lower order moments of the decision signal in parallel as the cost function to update the weight coefficients of the blind equalizer. The convergence speed and convergence precision is further improved.
23612 仿真实验表明，所提算法在有效解决相位旋转问题的同时较好地抑制了脉冲噪声，此外还具有较快的收敛速度和较小的稳态误差，稳健性较强。 The simulation experiments results show that the algorithm can effectively solve the problem of phase rotation and better suppress the impulse noise. Moreover, the algorithm has fast convergence speed, small steady-state error and strong robustness.
23613 在利用相对属性学习实现零样本图像分类中，现有的方法并没有考虑属性与类别之间的关系，为此该文提出一种基于共享特征相对属性的零样本图像分类方法。 Most algorithms of the zero-shot image classification with relative attributes do not consider the relationship between attributes and classes, therefore a new relative attributes method based on shared features is proposed for zero-shot image classification.
23614 该方法采用多任务学习的思想来共同学习类别分类器和属性分类器，获得一个低维的共享特征子空间，挖掘属性与类别之间的关系。 In analogy to the multi-task learning, the object classifier and attribute classifier are simultaneously learned in this method, from which a shared sub-space of lower dimensional features is obtained to mine the relationship between attributes and classes.