Word Spotting Based on a posterior Measur of Keyword Confidence

In this paper,an approach of keyword confidence estimation is developed that ewll combines acoustic layer scores and syllable-based statistical language model(LM)scores.An a posterioir(AP)confidence measure and its forward-backward calculating algorithm are deduced.A zero false alarm(ZFA) assumption is proposed for evaluating relative confidence measures by word spotting task.In a word spotting experiment with a vocabulary of 240 keywords,the keyword accuracy under the AP measure is above 94%,which well approaches its theoretical upper limit.In addition,a syllable lattice Hidden Markov Model(SLHMM) is formulated and a unified view of confidence estimation.word spotting,optimal path search,and N-best syllable re-scoring is presented ,The proposed AP measure can be easily applied to various speech recognition systems as well.     

置信度判别嵌入式隐马尔可夫模型人脸识别

为了提高人脸识别率,提出了一种优化置信度的判别嵌入式隐马尔可夫（EHMM）人脸识别方法。提出的方法基于假设检验,通过最小化检验错误率得到优化置信度判别式训练准则。在优化置信度判别式训练准则的前提下,通过参数估计求解判别式转换矩阵,提取出具有判别性、低维度的图像特征,确保观察样本能正确地分配到其对应的模型状态,以提高所训练出的EHMM模型的正确识别率。理论分析证明了优化置信度判别式训练准则的有效性,详细的实验及与现有方法的比较结果表明,提出的识别方法具有更好的识别性能。     

基于图像差分的关键帧检测技术

在实时的视频环境中,摄像头采集到的是人脸表情图像序列.人脸识别时,要对采集到的每一帧图像都进行表情识别,计算量相当大,无法满足系统的实时性要求.为此提出了一种基于图像差分的关键帧检测技术,运用小波变换人脸识别技术,将中性表情图像帧和检测图像帧序列分别构造中性表情弹性图和表情弹性图,接着逐帧计算表情弹性图和中性表情弹性图问的欧氏距离,根据欧氏距离变化趋势,使得系统能够从视频序列中检测到表情处于极大状态的关键帧图像,再用于后续的特征检测.实验结果表明,将关键帧检测算法和基于弹性模板匹配识别算法结合起来,可以满足系统的实时性要求,并可获得理想的识别率.     

一种改进的基于音节循环的置信度判决方法

为获得较为鲁棒的识别性能,一般的语音识别系统中都会在后端加入一个置信度判决模块,以实现识别错误检测和集外词拒识等功能。针对命令词语音识别系统,传统的基于Filler模型的置信度方法由于自身模型结构的限制,性能相对有限,尤其是对集外词的检测效果不好。为此,使用了一种基于音节循环的置信度判决方法,并对该方法的解码网络进行精简,以满足实用化的效率要求。在中文命令词测试集上的实验结果表明,该方法相对于基于Filler模型的置信度方法对识别效果与识别效率都有了较大的提升。     

基于互信息置信度的网格连续汉语语音检索*

针对目前生活中涌现的海量语音数据,人们对语音检索技术准确度的要求越来越高。主要研究了汉语连续语音检索任务中,基于转换音节网格的研究方法。针对语音检索系统中置信度计算的问题,提出了一种基于音节间互信息的置信度计算方法,并将其用于网格结构的语音检索系统中。该方法能够有效地利用上下文之间的互信息量,从而更准确、合理地描述汉语语言模型。实验结果表明,用提出的方法建立转换音节网格来进行语音检索,其检出率（FOM）比后验概率法和N-best法有较大幅度的提高。得到的汉语语音检索系统其FOM最高可以达到83.7%。     

A dialog switching strategy for information retrieval in noisy environments based on a confidence score

This paper presents a dialog switching strategy for information retrieval in human–robot dialog in order to cope with different intensities of background noise. The strategy dynamically switches between a more open and a closed dialog scheme based on a continuously adapted confidence score evaluating resulting numbers of cases of recognition and non-recognition of user speech, which are dependent on the level of background noise. Thereby, more natural and more robust dialog components are balanced, respectively. Experimental results are presented, illustrating the effectiveness of the approach.     

Fast unsupervised adaptation based on efficient statistics accumulation using frame independent confidence within monophone states

This paper proposes a fast unsupervised acoustic model adaptation technique with efficient statistics accumulation for speech recognition. Conventional adaptation techniques accumulate the acoustic statistics based on a forward–backward algorithm or a Viterbi algorithm. Since both algorithms require a state sequence prior to statistic accumulation, the conventional techniques need time to determine the state sequence by transcribing the target speech in advance. Instead of pre-determining the state sequence, the proposed technique reduces the computation time by accumulating the statistics with state confidence within monophone per frame. It also rapidly selects the appropriate gender acoustic model before adaptation, and further increases the accuracy by employing a power term after adaptation. Recognition experiments using spontaneous speech show that the proposed technique reduces computation time by 57.3% while providing the same accuracy as the conventional adaptation technique.     

基于上下文词语同现向量的词语相似度计算

词语的语义相似度是词语间语义相似紧密的一种数量化表示。提出一种词语的语义相似度计算方法 ,利用上下文词语同现向量来描述词语的语义知识 ,在此基础上 ,使用 min/ max的方法计算词语之间的语义相似度。实验结果表明 ,该方法能够比较准确地反映词语之间的语义关系 ,为词语间的语义关系提供一种有效度量。     

基于框架表示学习的汉语框架排歧

为了改善框架排歧模型的性能,区别于传统分类算法人工提取特征的做法,直接从语料中的例句出发,使用神经网络模型给出了一种框架表示学习的算法,并将学习到的框架表示向量用于框架排歧任务,显著提升了框架排歧的性能。该算法充分利用CFN中例句库、词元库,基于hinge-loss的神经网络,学习到能最大区别正确框架与错误框架的框架表示向量。此外,还使用WSABIE算法学习到目标词及其上下文的表示向量,排歧时以上下文表示向量与框架表示向量做余弦夹角来判决。在CFN中88个有歧义的词元上进行3组2折交叉验证（3×2 BCV）实验,框架排歧精度最好达到72.52%,◢t◣-检验结果表明该方法性能显著高于其他框架排歧方法。     

Classifier combination based on confidence transformation

This paper investigates the effects of confidence transformation in combining multiple classifiers using various combination rules. The combination methods were tested in handwritten digit recognition by combining varying classifier sets. The classifier outputs are transformed to confidence measures by combining three scaling functions (global normalization, Gaussian density modeling, and logistic regression) and three confidence types (linear, sigmoid, and evidence). The combination rules include fixed rules (sum-rule, product-rule, median-rule, etc.) and trained rules (linear discriminants and weighted combination with various parameter estimation techniques). The experimental results justify that confidence transformation benefits the combination performance of either fixed rules or trained rules. Trained rules mostly outperform fixed rules, especially when the classifier set contains weak classifiers. Among the trained rules, the support vector machine with linear kernel (linear SVM) performs best while the weighted combination with optimized weights performs comparably well. I have also attempted the joint optimization of confidence parameters and combination weights but its performance was inferior to that of cascaded confidence transformation-combination. This justifies that the cascaded strategy is a right way of multiple classifier combination.     

Approaches to single-valued neutrosophic MADM based on MABAC,TOPSIS and new similarity measure with score function

In this paper, we initiate a new axiomatic definition of single-valued neutrosophic distance measure and similarity measure, which is expressed by single-valued neutrosophic number that will reduce the information loss and remain more original information. Meanwhile, a novel score function is proposed. Then, the objective weights of various attributes are determined via gray system theory. Moreover, we present the combined weights, which can show both the subjective information and the objective information. Later, we present three algorithms to deal with multi-attribute decision-making problem based on revised Technique for Order Preference by Similarity to an Ideal Solution, Multi-Attributive Border Approximation area Comparison and similarity measure. Finally, the effectiveness and feasibility of approaches are demonstrated by two numerical examples.

     

Face detection based on skin color likelihood

We propose a face detection method based on skin color likelihood via a boosting algorithm which emphasizes skin color information while deemphasizing non-skin color information. A stochastic model is adapted to compute the similarity between a color region and the skin color. Both Haar-like features and Local Binary Pattern (LBP) features are utilized to build a cascaded classifier. The boosted classifier is implemented based on skin color emphasis to localize the face region from a color image. Based on our experiments, the proposed method shows good tolerance to face pose variation and complex background with significant improvements over classical boosting-based classifiers in terms of total error rate performance.     

Distance to second cluster as a measure of classification confidence

Most image classification algorithms rely on computing the distance between the unique spectral signature of a given pixel and a set of possible clusters within an n-dimensional feature space that represents discrete land cover categories. Each scrutinized pixel will ultimately be closest to one of the predefined clusters; different classification algorithms differ in the details of which cluster is considered as closest or most likely, but in general the selected algorithm will label each pixel with the label of the closest cluster. However, pixels expressing virtually identical distances to two or more clusters identify a limitation of this typical classification approach. Conditions for limitations to distance based classification algorithms include when distances are long and the pixel may not clearly belong to any single category, may represent mixed land cover, or can be easily confused spectrally between two or more categories. We propose that retention of the distance to the second closest cluster can shed light on the confidence with which label assignment proceeds and present several examples of how such additional information might enhance accuracy assessments and improve classification confidence. The method was developed with simplicity as a goal, assuming the classification has already been performed, and standard clustering reports are available. Over a test site in central British Columbia, Canada, we illustrate the described technique using classified image data from a nation-wide land cover mapping project. Calculation of multi-spectral Euclidean distances to cluster centroids, standardized by cluster variance, allows comparison of all potential class assignments within a unified framework. The variable distances provide a measure of relative confidence in the actual classification at the level of individual pixels.     

一种基于HNC理论的文本相似度算法

计算文本相似度常用基于向量空间计算夹角余弦的方法,该方法忽视了同一文本中词与词之间的语义相似度,因而造成了文本表示模型的高维性以及计算的高复杂性。为此,提出了一种文本相似度算法,利用HNC理论先计算特征词之间的语义相似度,进行必要的降维,进一步计算每个文本向量中的TF*IDF值,最后计算两个向量的空间夹角余弦值并将其作为两个文本之间的相似度。将实验结果与直接计算余弦值的结果比较发现,改进后的算法中VSM的维数明显比改进前小得多,改进后的算法提高了召回率和准确率。因此,改进后的算法是切实有效的。     

基于用户评分和共同评分项的协同过滤算法研究

针对协同过滤算法存在的问题进行改进,以提高评分预测和推荐结果的准确性。传统的相似度度量方法只考虑用户评分,过于简单,在皮尔逊相似度的基础上引入用户评分时间和商品流行度对用户评分进行加权处理,并与基于共同评分项规模的相似度计算进行加权组合,使得计算结果更加准确,也更符合现实意义。实验结果表明,新算法评分预测的平均绝对误差明显低于皮尔逊相似度,将MAE降低了10%以上,并提高了推荐的召回率和覆盖率。只在电影评分数据集上进行实验验证有一定的局限,该算法能够提高协同过滤算法的准确性,具有一定的现实意义。     

A dissimilarity measure based on Jensen Shannon divergence measure

The need of suitable measures to find the distance between two probability distributions arises as they play an eminent role in problems based on discrimination and inferences. In this communication, we have introduced one such divergence measure based on well-known Shannon entropy and established its existence. In addition to this, a new dissimilarity measure for intuitionistic fuzzy sets corresponding to proposed divergence measure is also introduced and validated. Some major properties of the proposed dissimilarity measure are also discussed. Further, a new multiple attribute decision-making (MADM) method based on the proposed dissimilarity measure is introduced by using the concept of TOPSIS and is thoroughly explained with the help of an illustrated example on supplier selection problem. Finally, the application of proposed dissimilarity measure is given in pattern recognition and the performance is compared with some existing divergence measures in the literature.     

基于回归置信度的多尺度精确人脸检测

在大多数受限情况下人脸检测已经有了许多有效方案,但对于人脸尺度变化极大、小人脸,以及模糊、遮挡、光照等非受限环境的人脸检测问题,仍面临更多挑战。针对以上问题,提出一种多尺度卷积神经网络模型。在R-FCN网络的基础上进行改进,以多尺度特征替代单一特征,使网络对多尺度信息更加敏感,在预测阶段同时输出分类置信度与回归置信度,改进非极大值抑制(non-maximum suppression,NMS)算法,提出基于回归置信度的NMS算法。在WIDER FACE数据集上训练模型,在FDDB与WIDER FACE人脸评测库进行实验,实验结果表明,召回率、准确率等指标均优于其它人脸检测算法。     

Wi-Fi fingerprinting based on collaborative confidence level training

Wi-Fi fingerprinting has been a popular indoor positioning technique with the advantage that infrastructures are readily available in most urban areas. However wireless signals are prone to fluctuation and noise, introducing errors in the final positioning result. This paper proposes a new fingerprint training method where a number of users train collaboratively and a confidence factor is generated for each fingerprint. Fingerprinting is carried out where potential fingerprints are extracted based on the confidence factor. Positioning accuracy improves by 40% when the new fingerprinting method is implemented and maximum error is reduced by 35%.