Accelerated image reconstruction using ordered subsets of projection data

The authors define ordered subset processing for standard algorithms (such as expectation maximization, EM) for image restoration from projections. Ordered subsets methods group projection data into an ordered sequence of subsets (or blocks). An iteration of ordered subsets EM is defined as a single pass through all the subsets, in each subset using the current estimate to initialize application of EM with that data subset. This approach is similar in concept to block-Kaczmarz methods introduced by Eggermont et al. (1981) for iterative reconstruction. Simultaneous iterative reconstruction (SIRT) and multiplicative algebraic reconstruction (MART) techniques are well known special cases. Ordered subsets EM (OS-EM) provides a restoration imposing a natural positivity condition and with close links to the EM algorithm. OS-EM is applicable in both single photon (SPECT) and positron emission tomography (PET). In simulation studies in SPECT, the OS-EM algorithm provides an order-of-magnitude acceleration over EM, with restoration quality maintained.     

A Method for Evaluating All the Minimal Cuts of a Graph

This paper presents a new technique to determine all minimal cuts for all the sink nodes in a non-planar network. The algorithm uses a subset method and an iterative process to achieve high efficiency. For an N-sink node network there are (2N - 1) possible combinations of nodes (non-empty subsets). These subsets are checked against several conditions to see if they can be transformed into minimal cutsets. An iterative process is used to generate these non-empty subsets efficiently so that the number of subsets to be checked is about (2N - 1)/2. Since this algorithm generates all the minimal cutsets for all nodes in one operation, it is faster compared to conventional methods which compute for one sink node at a time. Tests using random graphs showed a small cpu time per minimal cutset.     

三维锥束OSEM算法中子集序列的研究

有序子集算法的提出提高了迭代算法的图像重建速度,但是三维锥束OSEM算法中的子集个数和子集迭代的顺序都会影响图像重建的收敛速度和质量.文中提出了划分子集序列的方法,该方法通过应用距离加权方案改变子集的迭代顺序,使子集均匀分布,来减小相邻子集的相关性对收敛性的影响,并且采用每次迭代后逐渐减少子集个数,提高图像重建收敛速度...     

一种新的多类模式识别方法

提出了一种基于支撑向量机的多类分类器,用N-1个支撑向量机组合构成一个具有二叉树结构形式的N-多类分类器．讨论了该多类分类器的泛化推广能力,同时还提出了该多类分类器的基于特征空间的BTSVM学习算法,BTSVM算法使用核函数转换的方式计算特征空间的样本距离;采用类间最小距离最大化作为聚类准则,在每个决策结点产生两个最优子集;然后采用支撑向量机学习算法学习两个最优子集,确定决策结点的最优分类面．理论和实验结果表明,本文提出的基于支撑向量机的多类分类器在整体性能上要优于其它类似的分类器系统。     

Convergent block-iterative algorithms for image reconstruction frominconsistent data

It has been shown that convergence to a solution can be significantly accelerated for a number of iterative image reconstruction algorithms, including simultaneous Cimmino-type algorithms, the "expectation maximization" method for maximizing likelihood (EMML) and the simultaneous multiplicative algebraic reconstruction technique (SMART), through the use of rescaled block-iterative (BI) methods. These BI methods involve partitioning the data into disjoint subsets and using only one subset at each step of the iteration. One drawback of these methods is their failure to converge to an approximate solution in the inconsistent case, in which no image consistent with the data exists; they are always observed to produce limit cycles (LCs) of distinct images, through which the algorithm cycles. No one of these images provides a suitable solution, in general. The question that arises then is whether or not these LC vectors retain sufficient information to construct from them a suitable approximate solution; we show that they do. To demonstrate that, we employ a "feedback" technique in which the LC vectors are used to produce a new "data" vector, and the algorithm restarted. Convergence of this nested iterative scheme to an approximate solution is then proven. Preliminary work also suggests that this feedback method may be incorporated in a practical reconstruction method.     

LDPC编码调制系统中基于反馈LLR均值的迭代解调/译码算法

该文针对LDPC码编码的BICM系统,提出一种对LDPC码译码器输出外附信息的计算方法进行改进的迭代解调/译码算法。与传统的解调/译码算法不同在于,该算法对每次BP迭代中译码器输出的各编码比特的外附LLR分别求均值后,再将其作为先验信息反馈给软解调器开始下次的迭代解调/译码。采用该方法可有效地减轻LDPC码在BP迭代过程中某些比特LLR值的振荡现象,从而使得传递给软解调器的外附信息更准确。仿真结果表明,和传统的两种迭代解调/译码算法相比,该算法能进一步提高LDPC编码BICM迭代系统的译码性能,而复杂度并无明显增加。     

一种并行决策树学习方法研究

提出一种并行的决策树学习方法。该方法首先将数据库分为若干个子部分,针对每个子部分分别进行决策树学习,优选分裂属性,再对各个决策树学习的结果综合,生成最终的树。在树上剪枝以降低分类错误率。通过在变压器绝缘故障诊断中的应用表明该方法有很强的学习能力和诊断速度,是一种有效的决策树学习方法。     

混合蒙特卡罗搜索的特征选择算法的优化

特征选择是机器学习和数据挖掘中处理高维数据的初步步骤,通过消除冗余或不相关的特征来识别数据集中最重要和最相关的特征,从而提高分类精度和降低计算复杂度。文中提出混合蒙特卡罗树搜索特征选择算法(HMCTS),首先,根据蒙特卡罗树搜索方法迭代生成一个初始特征子集,利用ReliefF算法过滤选择前k个特征形成候选特征子集;然后,利用KNN分类器的分类精度评估候选特征,通过反向传播将模拟结果更新到迭代路径上所有选择的节点;最后,选择高精度的候选特征作为最佳特征子集。仿真结果表明,对比HPSO-LS和MOTiFS算法,HMCTS算法具有良好的可扩展性,且分类精度高。     

Efficient Detection Algorithms for MIMO Communication Systems

In this paper, two new efficient detection algorithms, Type 1 (T1) with better complexity-performance tradeoff and Type 2 (T2) with lower complexity, are derived from one generalized framework for multiple-input multiple-output (MIMO) communication systems. The proposed generalized detection framework constructed by parallel interference cancellation (PIC), group, and iteration techniques provides three parameters and three sub-algorithms to generate two efficient detection algorithms and conventional BLAST-ordered decision feedback (BODF), grouped, iterative, and B-Chase detection algorithms. Since the group interference suppression (GIS) technique is applied to the proposed detection algorithms, the complexities of the preprocessing (PP) and tree search (TS) can be reduced. In (8,8) system with uncoded 16-QAM inputs, one example of the T1 algorithm can save complexity by 21.2% at the penalty of 0.6 dB loss compared with the B-Chase detector. The T2 algorithm not only reduces complexity by 21.9% but also outperforms the BODF algorithm by 3.1 dB.     

基于Spark的分层子空间权重树随机森林算法

高维数据的很多特征与类别的相关性弱,影响了随机森林的分类正确率。针对原始随机森林算法在高维数据上的分类问题,提出了一种分层子空间权重树随机森林算法。同时,传统的单机模式无法满足高维数据计算效率的需求,因此利用开源集群计算框架Spark在内存缓存和迭代计算上的优势,将所提算法在Spark上实现。所提算法采用以决策树为单位的分层抽样来生成特征子空间,在提高单棵决策树性能的同时,保证决策树之间的多样性;并且采用权重树的集成策略,使分类能力强的树在集成过程中影响力更大。通过在Mnist和Gisette数据集上的实验结果表明,相比原始随机森林算法、TWRF算法以及分层子空间随机森林算法,所提算法具有更好的正确率,提高了泛化误差性能,可扩展性良好,能够有效分类高维数据。      

改进的SC-FDE块迭代判决反馈均衡器

在单载波频域传输(SC-FDE)系统中,块迭代判决反馈均衡器(IBDFE)明显提升了传统线性频域均衡器的性能.未知发送信号与迭代判决信号的相关因子估计是其关键技术,直接影响均衡器的性能.首先对IBDFE相关因子估计算法进行了改进,同时提出一种基于独特字(UW)帧结构的估计方法.该方法利用独特字的已知性和恒包络性进行判决,基于此判决方式使相关因子得到更精准的估计.实验结果表明,两种方法对IBDFE的性能有较为明显的提升.     

Extending a Fixed-Complexity Sphere Decoder to Obtain Likelihood Information for Turbo-MIMO Systems

A list extension for a fixed-complexity sphere decoder (FSD) to perform iterative detection and decoding in turbo-multiple input-multiple output (MIMO) systems is proposed in this paper. The algorithm obtains a list of candidates that can be used to calculate likelihood information about the transmitted bits required by the outer decoder. The list FSD (LFSD) overcomes the two main problems of the list sphere decoder (LSD), namely, its variable complexity and the sequential nature of its tree search. It combines a search through a very small subset of the complete transmit constellation and a specific channel matrix ordering to approximate the soft- quality of the list of candidates obtained by the LSD. A simple method is proposed to generate that subset, extending the subset searched by the original FSD. Simulation results show that the LFSD can be used to approach the performance of the LSD while having a lower and fixed complexity, making the algorithm suitable for hardware implementation.      

基于递归和SVM的特征选择方法研究

特征选择是模式识别领域的一个重要的研究方向,它可以提高分类的效率与效果。本文将递归特征排除算法与SVM决策树结合起来运用于特征选择,首先利用递归特征排除算法对所选择的特征进行初排序,然后依次将特征送入SVM决策树中进行优化评估,对数据中起显著作用的特征进行筛选,除去冗余和次要特征,得到特征子集。最后,通过对Linux主机和相关网络的27个入侵特征数据进行特征选择实验,实验结果表明,特征个数降至21个,而测试精度仍然能达到94%,从而证明本文所提出的递归和SVM相结合的方法是解决特征选择问题的一种有效方法。     

Top-down induction of decision trees classifiers - a survey

Decision trees are considered to be one of the most popular approaches for representing classifiers. Researchers from various disciplines such as statistics, machine learning, pattern recognition, and data mining considered the issue of growing a decision tree from available data. This paper presents an updated survey of current methods for constructing decision tree classifiers in a top-down manner. The paper suggests a unified algorithmic framework for presenting these algorithms and describes the various splitting criteria and pruning methodologies.     

The condensed nearest neighbor rule using the concept of mutual nearest neighborhood (Corresp.)

A two-stage iterative algorithm for selecting a subset of a training set of samples for use in a condensed nearest neighbor (CNN) decision rule is introduced. The proposed method uses the concept of mutual nearest neighborhood for selecting samples close to the decision line. The efficacy of the algorithm is brought out by means of an example.     

基于随机蕨的集成分类图像隐写分析算法

提出了一种基于随机蕨(random ferns)和集成学习的图像隐写分析算法。首先利 用图像高维特征构 建蕨特征,采用成对采样策略构造样本子集,生成若干个基分类器;然后计算出训练样本 在基分类器中各个蕨的 先验概率并集成各个基分类器,进行隐写检测判别。实验结果表明,本文算法复杂度低,能 有效降低隐写检测错误率。     

Matching From the First Iteration: An Iterative Switching Algorithm for an Input Queued Switch

An iterative switching algorithm for an input queued switch consists of a number of iterations in every time step, where each iteration computes a disjoint matching. If input is matched to output in a given iteration, a packet (if any) is forwarded from to in the corresponding time step. Most of the iterative switching algorithms use a request grant accept (RGA) arbitration type (e.g. iSLIP). Unfortunately, due to this particular type of arbitration, the matching computed in one iteration is not necessarily maximal (more input and output ports can still be matched). This is exactly why multiple iterations are needed. However, multiple iterations make the time step larger and reduce the speed of the switch. We present a new iterative switching algorithm (based on the RGA arbitration) called with the underlying assumption that the number of iterations is possibly limited to one, hence reducing the time step and allowing the switch to run at a higher speed. We prove that achieves throughput and delay guarantees with a speedup of 2 and one iteration under a constant burst traffic model, which makes as good as any maximal matching algorithm in the theoretical sense. We also show by simulation that achieves relatively high throughput in practice under uniform and non-uniform traffic patterns with one iteration and no speedup.     

高光谱图像波段子集模糊积分融合异常检测

针对高光谱图像中背景及目标先验知识未知条件下的异常目标检测问题,该文给出一种基于高相关性波段子集分割的模糊积分低概率目标检测融合算法。依据高光谱图像数据的波段相关性将原始高光谱数据分割为若干连续波段子集;利用非参核密度估计得到原假设下各波段子集数据RX检测器输出的概率密度函数,构造出非参隶属度映射函数;利用数据光谱维的特征值定义目标信号噪声能量比(TNER),衡量各波段子集信源检测结果的重要程度;最后,通过Sugeno模糊积分实现波段子集检测结果的决策级融合。使用可见光/近红外波段OMIS-I高光谱图像进行了实验,实验结果证明了算法的有效性。     

基于簇相似的多分类器目标跟踪算法

由于跟踪过程中目标和背景的变化,传统的单分类器跟踪算法学习到大量的非目标信息而导致跟踪精度降低.针对该问题,本文提出使用树形结构保存历史分类器.在每一帧,根据树中路径距离选择分类器集对测试样本分类.提出了一种基于簇相似性比较的分类算法.通过建立以方差为尺度的特征空间,比较测试样本到簇中心的距离计算相似度,快速计算出目标样本.实验表明本算法能够在复杂条件下实现对目标的鲁棒跟踪.