Entropy Regularized Likelihood Learning on Gaussian Mixture: Two Gradient Implementations for Automatic Model Selection

In Gaussian mixture modeling, it is crucial to select the number of Gaussians or mixture model for a sample data set. Under regularization theory, we aim to solve this kind of model selection problem through implementing entropy regularized likelihood (ERL) learning on Gaussian mixture via a batch gradient learning algorithm. It is demonstrated by the simulation experiments that this gradient ERL learning algorithm can select an appropriate number of Gaussians automatically during the parameter learning on a sample data set and lead to a good estimation of the parameters in the actual Gaussian mixture, even in the cases of two or more actual Gaussians overlapped strongly. We further give an adaptive gradient implementation of the ERL learning on Gaussian mixture followed with theoretic analysis, and find a mechanism of generalized competitive learning implied in the ERL learning.     

基于Bayesian学习的适应性优化协商模型

在复杂的自动协商环境中,设计能够处理不完全信息和动态情形的协商agent有效学习机制正成为具有挑战性的议题.提出了一种基于Bayesian学习的时间依赖的双边多议题协商优化模型(BLMSEAN).通过只观察对手的历史报价,将Bayesian学习和基于混合策略的演化算法相结合,所提模型使得协商agent能够对于对手协商参数的概率分布有更精确的估计(如期限、保留报价和议题权重等),能够适应性地调整让步策略使协商双方都受益,提高了协商的成功率和效用.通过实验可以显示所提的模型学习对手私有信息和适应性调整让步策略的有效性.     

A unified perspective and new results on RHT computing, mixture based learning, and multi-learner based problem solving

On one hand, multiple object detection approaches of Hough transform (HT) type and randomized HT type have been extended into an evidence accumulation featured general framework for problem solving, with five key mechanisms elaborated and several extensions of HT and RHT presented. On the other hand, another framework is proposed to integrate typical multi-learner based approaches for problem solving, particularly on Gaussian mixture based data clustering and local subspace learning, multi-sets mixture based object detection and motion estimation, and multi-agent coordinated problem solving. Typical learning algorithms, especially those that base on rival penalized competitive learning (RPCL) and Bayesian Ying-Yang (BYY) learning, are summarized from a unified perspective with new extensions. Furthermore, the two different frameworks are not only examined with one viewed crossly from a perspective of the other, with new insights and extensions, but also further unified into a general problem solving paradigm that consists of five basic mechanisms in terms of acquisition, allocation, amalgamation, admission, and affirmation, or shortly A5 paradigm.     

A Necessary Condition about the Optimum Partition on a Finite Set of Samples and Its Application to Clustering Analysis

1IntroductionAsusedforclusteringanalysis,K-meansclusteringalgorithmhasbeenwidelyaPpliedindatacompression,characterextractionofpatternandcommunication-ConsideringthatK-meansclusteringalgorithmisadatacompressionmethodwithloss,wehopetominimizethelossundersomeconstraints,whichmaybringforththeoptimumdesigningofK-partition(orclusteringcenters).ThesequentialhardK-means(SHKM)isanoldclusteringapproach.TheMacQueen'sSHKMl1]algorithmwaspresentedin1967.Ref.[2]proposesamodifiedformtofindtheinitialclu…     

基于GLMB滤波和Gibbs采样的多扩展目标有限混合建模与跟踪算法

本文针对杂波条件下多扩展目标的状态估计, 目标个数估计, 扩展目标形状估计问题, 提出了一种基于标签随机有限集(Labelled random finite sets, L-RFS)框架下多扩展目标跟踪学习算法, 该学习算法主要包括两方面:多扩展目标动态建模和多扩展目标的跟踪估计.首先, 结合广义标签多伯努利滤波器(Generalized labelled multi-Bernoulli, GLMB)建立了扩展目标的量测有限混合模型(Finite mixture models, FMM), 利用Gibbs采样和贝叶斯信息准则(Bayesian information criterion, BIC)准则推导出有限混合模型的参数来对多扩展目标形状进行学习, 然后采用等效量测方法来替代扩展目标产生的量测, 对扩展目标形状采用椭圆逼近建模, 实现扩展目标形状与状态的估计.仿真实验表明本文所给的方法能够有效跟踪多扩展目标, 并且在目标个数估计方面优于CBMeMBer算法.此外, 与标签多伯努利滤波(LMB)计算比较表明: GLMB和LMB算法滤波估计精度接近, 二者精度高于CBMeMBer算法.     

A Necessary Condition about the Optimum Partition on a Finite Set of Samples and Its Application to Clustering Analysis

This paper presents another necessary condition about the optimum partition on a finite set of samples.From this condition,a corresponding generalized sequential hard k-means (GSHKM) clustering algorithm is built and many well-known clustering algorithms are found to be included in it.Under some assumptions the well-known MacQueen;s SHKM (Sequential Hard K-Means) algorithm,FSCL(Frequency Sensitive Competitive Learning) algorithm and RPCL (Rival Penalized Competitive Learning) algorithm are derived.It is shown that FSCL in fact still belongs to the kind of GSHKM clustering algorithm and is more suitable for producing means of K-partition of sample data,which is illustrated by numerical experiment.Meanwhile,some improvements on these algorithms are also given.     

基于深度学习和H&E染色病理图像的肿瘤相关指标预测研究综述

肿瘤的精确诊断对患者的治疗方案选择和预后预测都非常重要.病理学诊断被认为是肿瘤诊断的"金标准",但是,病理学发展目前仍然面临着巨大的挑战,如病理医生的缺乏,特别是在欠发达地区和小医院,这将导致病理医生长期超负荷工作,同时,病理诊断严重依赖于病理医生的专业知识和诊断经验,病理医生的主观性导致了诊断不一致性的激增.全切片扫...