Word, thought, and deed: The role of object categories in children's inductive inferences and exploratory play.

Previous research (e.g., S. A. Gelman & E. M. Markman, 1986; A. Gopnik & D. M. Sobel, 2000) suggests that children can use category labels to make inductive inferences about nonobvious causal properties of objects. However, such inductive generalizations can fail to predict objects' causal properties when (a) the property being projected varies within the category, (b) the category is arbitrary (e.g., things smaller than a bread box), or (c) the property being projected is due to an exogenous intervention rather than intrinsic to the object kind. In 4 studies, the authors showed that preschoolers (M = 48 months; range = 42-57 months) were sensitive to these constraints on induction and selectively engaged in exploration when evidence about objects' causal properties conflicted with inductive generalizations from the objects' kind to their causal powers. This suggests that the exploratory actions children generate in free play could support causal learning. (PsycINFO Database Record (c) 2010 APA, all rights reserved)     

Grounding spatial language in perception: An empirical and computational investigation.

The present paper grounds the linguistic categorization of space in aspects of visual perception; specifically, the structure of projective spatial terms such as above are grounded in the process of attention and in vector-sum coding of overall direction. This is formalized in the attentional vector-sum (AVS) model. This computational model accurately predicts linguistic acceptability judgments for spatial terms, under a variety of spatial configurations. In 7 experiments, the predictions of the AVS model are tested against those of 3 competing models. The results support the AVS model and disconfirm its competitors. The authors conclude that the structure of linguistic spatial categories can be partially explained in terms of independently motivated perceptual processes. (PsycINFO Database Record (c) 2010 APA, all rights reserved)     

面向政策法规数据的分类方法

通过设计阀值、构造类别的基向量和待分类向量的复数指标,建立了面向政策法规数据的适于有效机器学习的实时动态可扩展的分类方法.改进了文档的自动分类多采用以类别为中心的分类模式,综合了以文档为中心和以类别为中心两种分类模式的优点,实现了一次性完成单标号分类和多标号分类问题,克服了以往分类中预先定义主题类别和聚类分类中预先指定类别的种数的缺陷.编制了相应的算法.     

Clinical psychologists' theory-based representations of mental disorders predict their diagnostic reasoning and memory.

The theory-based model of categorization posits that concepts are represented as theories, not feature lists. Thus, it is interesting that the Diagnostic and Statistical Manual of Mental Disorders (4th ed.; DSM-IV, American Psychiatric Association, 1994) established atheoretical guidelines for mental disorder diagnosis. Five experiments investigated how clinicians handled an atheoretical nosology. Clinicians' causal theories of disorders and their responses on diagnostic and memory tasks were measured. Participants were more likely to diagnose a hypothetical patient with a disorder if that patient had causally central rather than causally peripheral symptoms according to their theory of the disorder. Their memory for causally central symptoms was also biased. Clinicians are cognitively driven to use theories despite decades of practice with the atheoretical DSM. (PsycINFO Database Record (c) 2010 APA, all rights reserved)     

基于主动学习和否定选择的垃圾邮件分类算法

针对现在网络上泛滥的垃圾邮件问题,本文结合主动学习方法和否定选择算法提出了一种二类文本分类方法：主动否定学习算法.根据用户少量标注建立双向兴趣集,利用否定选择算法的自体异常检测机制改善主动学习中的采样策略,并将双向兴趣集作为检测器,新增样本集作为自体集,对两者进行异常匹配.本文算法与在线垃圾邮件快速识别方法、增强差异性的半监督协同分类算法、垃圾邮件过滤方法、基于人工高免疫的多层垃圾邮件过滤算法和在线主动多领域学习方法在六个常用邮件语料集上进行了分析比较,结果表明本文算法具有较高的准确率、召回率、分类精度,和较低的用户标注负担.使用用户个性喜好转换为双向兴趣特征的方式有助于提高算法的分类能力;利用异常检测匹配选取未知类别特征的方式,有效地降低了用户标注负担.     

基于朴素贝叶斯分类器的朝鲜语文本分类的研究

该文基于朴素贝叶斯分类器对朝鲜语文本分类进行了研究。首先,利用基于类别选择的特征选择方法对朝鲜语文本进行特征选择,并使用类TF-IDF估算方法计算权重;其次,构造朴素贝叶斯分类器;最后,利用分类器实现对朝鲜语文本的分类。实验表明,该方法在朝鲜语文本分类中具有较好的效果,为朝汉结合文本分类提供了一定的依据。     

基于特征辨别能力和元信息的特征选择

特征选择是文本分类的关键步骤之一,所选特征子集的优劣直接影响文本分类的结果。在分析词频方法和文档频方法不足的基础上提出了特征辨别能力,把元信息引入粗糙集并提出了一个基于元信息的属性约简算法,给出了一个综合性特征选择方法。该方法利用特征辨别能力进行特征初选以过滤掉一些词条来降低特征空间的稀疏性,使用所提属性约简算法消除冗余,从而获得较具代表性的特征子集。实验结果表明：所提特征选择方法在一定程度上具有一定的优势。     

基于多信息融合的视觉目标类识别算法研究

视觉目标类识别是计算机视觉研究领域中的最具挑战性的难题之一,目前仍有许多问题没有得到很好的解决.近年来提出的空域金字塔直方图特征表示,在描述特征点集分布属性方面取得了比较好的实验效果.但是由于其描述的信息不全面,在性能上仍有较大改进余地.从信息互补性角度出发,提出了基于多信息融合的集成策略,将空域金字塔直方图表示与费舍分数表示各自描述的优势相结合,用于视觉目标类识别.实验证明该策略是有效的,在所进行测试的所有类别上相比单信息识别的性能均取得了一致性提高.     

特征权对贝叶斯分类器文本分类性能的影响

在文本分类研究中,人们希望用特征权来改善文本分类效果。以最优分类器——贝叶斯分类器为基准分类器,研究了特征权对文本分类性能的可能影响。理论推导表明,就最优分类器而言,特征权不能有效提高文本分类效果。     

基于类别加权和方差统计的特征选择方法

为提高不均衡文本分类的准确率和稳定性, 提出了一种基于类别加权和方差统计的联合特征选择方法.首先, 基于类别文档数大小对特征选择的影响, 给出了一种类别加权策略以强化小类别的特征;其次, 在探究特征类别区分能力的基础上, 设计了类别方差统计策略来凸显含有丰富类别信息的特征;最后, 将2种策略相融合, 实现了一种联合特征选择的新算法.在Reuters-21578和复旦大学语料这2个不均衡语料上的实验都表明:该算法有效, 特别是在小类别的分类效果上远远好于IG、CHI和DFICF等流行的通用算法.