Modelling vocabulary acquisition, adaptation and generalization in infants using adaptive Bayesian PLSA

During the early stages of language acquisition, young infants face the task of learning a basic vocabulary without the aid of prior linguistic knowledge. Attempts have been made to model this complex behaviour computationally, using a variety of machine learning algorithms, a.o. non-negative matrix factorization (NMF). In this paper, we replace NMF in a vocabulary learning setting with a conceptually similar algorithm, probabilistic latent semantic analysis (PLSA), which can learn word representations incrementally by Bayesian updating. We further show that this learning framework is capable of modelling certain cognitive behaviours, e.g. forgetting, in a simple way.     

Using the genetic algorithm to enhance nonnegative matrix factorization initialization

Nonnegative matrix factorization (NMF) algorithms have been utilized in a wide range of real applications; however, the performance of NMF is highly dependent on three factors including: (1) choosing a problem dependent cost function; (2) using an effective initialization method to start the updating procedure from a near‐optimal point; and (3) determining the rank of factorized matrices prior to decomposition. Due to the nonconvex nature of the NMF cost function, finding an analytical‐based optimal solution is impossible. This paper is aimed at proposing an efficient initialization method to modify the NMF performance. To widely explore the search space for initializing the factorized matrices in NMF, the island genetic algorithm (IGA) is employed as a diverse multiagent search scheme. To adapt IGA for NMF initialization, we present a specific mutation operator. To assess how the proposed IGA initialization method efficiently enhances NMF performance, we have implemented state‐of‐the‐art initialization methods and applied to the Japanese Female Facial Expression dataset to recognize the facial expression states. Experimental results demonstrate the superiority of the proposed approach to the compared methods in terms of relative error and fast convergence.     

基于流形正则化非负矩阵分解的高光谱数据降维

采用基于流形正则化非负矩阵分解(MR-NMF)的高 光谱数据降维方法。新方 法通过构建样本的近邻图描述数据几何结构,然后将其作为正则项加入NMF的目标函 数中进行组合优化。在真实的高光谱数据集HYDICE上进行的实验结果表明,新方法能 够提高高光谱图像分类的精度。     

以非负矩阵分解提取局部特征的SAR目标稀疏表示分类

合成孔径雷达( SAR)目标分类是自动目标识别系统的核心功能之一,对于战场监视等应用具有重要意义。利用SAR图像局部散射明显的特点,提出了通过训练样本的非负矩阵分解获得低维数局部特征编码,并以该编码作为字典进行稀疏表示分类的方法。采用Gotcha项目民用车辆目标的实测数据进行了验证,结果显示在不同信噪比条件下该方法的分类正确率均优于广泛采用的由降采样、随机投影、主成分分析提取低维数特征的稀疏表示分类方法,表明了该方法的性能优势。另外,还通过实验对比分析了非负约束的稀疏表示与标准稀疏表示在分类性能上的差别,结果显示非负约束的稀疏表示导致分类正确率下降,故针对分类问题不宜在稀疏表示时进行非负约束。     

基于小波变换和NMF的人脸识别方法的研究

为了克服PCA、ICA等传统方法在人脸图像特征抽取时存在速度慢、识别率低的缺点，该文提出了一种将非负矩分解思想应用于人脸特征提取的算法。利用小波变换对人脸图像进行分解，对其中包含主要信息的低频子带运用NMF构造特征子空间，在子空间内实现识别。实验结果表明，该方法实用、有效，减少了计算量，提高了系统的识别率，使识别率达到90％以上，有着广泛的研究价值和应用 前景。     

非负矩阵因子法分析多元混合酸的pKa

提出了多元混合酸pKa的计算的新方法。主要原理基于混合酸体系滴定的数学模型,对于pKa相近的混合酸出现的重叠峰,首次把非负矩阵因子分析(Non-negative Matrix Factorization(NMF))法应用到滴定谱图解析,从而清楚识别出各种混合酸的pKa,总的平均相对误差小于3％,结果令人满意。     

三维非负矩阵因子分解代谢组学数据解析

多维数据解析方法越来越引起人们的重视,非负矩阵因子分解算法已较广泛地用于图像分析。基于PARAFAC模型,将非负矩阵因子分解算法拓展为三维非负矩阵因子分解算法(three dimension non-negative matrix factorization,NMF3)。其原理简明,算法易于执行。与基于向量计算的其他三维化学计量学算法不同,NMF3基于矩阵计算单个元素,所以不必将三维数据平铺处理,就可直接解析,为三维数据解析研究提供了一种全新的思路和方法。应用NMF3解析模拟三维数据和代谢组学数据,结果令人满意。     

基于Hessian图正则稀疏NMF的高光谱解混

基于非负矩阵分解(Nonnegative Matrix Factorization, NMF)的高光谱解混(Hyperspectral Unmixing,HU)方法引起了大家的关注,因为可以将一个非负高光谱图像(Hyperspectral Imagery, HSI)数据矩阵分解为两个非负矩阵的乘积,分别对应于端元矩阵和丰度系数矩阵。目前,图约束的NMF算法已经被证明对高光谱解混是有效的,因为它们可以捕获HSI的几何特性。为了挖掘数据在混合过程中的几何结构和稀疏性,提出了一种稀疏的Hessian图正则化NMF(SHGNMF)算法。SHGNMF算法是将丰度矩阵的L1/2正则化器和Hessian图正则化项都添加到每个NMF模型中,同时采用乘法更新规则。最后用模拟数据和真实数据进行实验,验证了所提出的SHGNMF算法相对于其他NMF算法的优越性。     

Speech Separation Methodology for Hearing Aid

In the design of hearing aids (HA), the real-time speech-enhancement is done. The digital hearing aids should provide high signal-to-noise ratio, gain improvement and should eliminate feedback. In generic hearing aids the performance towards different frequencies varies and non uniform. Existing noise cancellation and speech separation methods drops the voice magnitude under the noise environment. The performance of the HA for frequency response is non uniform. Existing noise suppression methods reduce the required signal strength also. So, the performance of uniform sub band analysis is poor when hearing aid is concern. In this paper, a speech separation method using Non-negative Matrix Factorization (NMF) algorithm is proposed for wavelet decomposition. The Proposed non-uniform filter-bank was validated by parameters like band power, Signal-to-noise ratio (SNR), Mean Square Error (MSE), Signal to Noise and Distortion Ratio (SINAD), Spurious-free dynamic range (SFDR), error and time. The speech recordings before and after separation was evaluated for quality using objective speech quality measures International Telecommunication Union -Telecommunication standard ITU-T P.862.     

二维投影非负矩阵分解算法及其在人脸识别中的应用

建立在最小化非负矩阵分解损失函数上的人脸识别算法需同时计算基矩阵和系数矩阵, 导致求解这类问题十分耗时. 本文把非负属性引入二维主成分分析(2-dimensional principal component analysis, 2DPCA)中, 提出了一种新的二维投影非负矩阵分解(2-dimensional projective non-negative matrix factorization, 2DPNMF)人脸识别算法. 该算法在保持人脸图像的局部结构情况下, 突破了最小化非负矩阵分解损失函数的约束, 仅需计算投影矩阵(基矩阵), 从而降低了计算复杂度. 本文从理论上证明了所提出算法的收敛性, 同时, 使用了YALE、FERET和AR三个人脸库进行实验, 结果表明2DPNMF不仅识别率高, 而且速度优于非负矩阵分解和二维主成分分析.