Adaptive text mining: inferring structure from sequences

Text mining is about inferring structure from sequences representing natural language text, and may be defined as the process of analyzing text to extract information that is useful for particular purposes. Although hand-crafted heuristics are a common practical approach for extracting information from text, a general, and generalizable, approach requires adaptive techniques. This paper studies the way in which the adaptive techniques used in text compression can be applied to text mining. It develops several examples: extraction of hierarchical phrase structures from text, identification of keyphrases in documents, locating proper names and quantities of interest in a piece of text, text categorization, word segmentation, acronym extraction, and structure recognition. We conclude that compression forms a sound unifying principle that allows many text mining problems to be tacked adaptively.     

Dimensionality in Voice Quality

This study concerns speaking voice quality in a group of male teachers (n = 35) and male actors (n = 36), as the purpose was to investigate normal and supranormal voices. The goal was the development of a method of valid perceptual evaluation for normal to supranormal and resonant voices. The voices (text reading at two loudness levels) had been evaluated by 10 listeners, for 15 vocal characteristics using VA scales. In this investigation, the results of an exploratory factor analysis of the vocal characteristics used in this method are presented, reflecting four dimensions of major importance for normal and supranormal voices. Special emphasis is placed on the effects on voice quality of a change in the loudness variable, as two loudness levels are studied. Furthermore, the vocal characteristics Sonority and Ringing voice quality are paid special attention, as the essence of the term "resonant voice" was a basic issue throughout a doctoral dissertation where this study was included.     

Wave atom transform generated strong image hashing scheme

The rapid development of multimedia technology has resulted in a rising rate on digital unauthorized utilization and forgery, which makes the situation of image authentication increasingly severe. A novel strong image hashing scheme is proposed based on wave atom transform, which can better authenticate images by precisely distinguishing the malicious tampered images from the non-maliciously processed ones. Wave atom transform is employed since it has significantly sparser expansion and better characteristics of texture feature extraction than other traditional transforms. For better detection sensitivity, gray code is applied instead of natural binary code to optimize the hamming distance. Randomizations are also performed using Rényi chaotic map for the purposes of secure image hashing and key sensitivity. The experimental results show that the proposed strong scheme is robust to non-malicious content-preserving operations and also fragile to malicious content-altering operations. The scheme also outperforms DCT and DWT based schemes in terms of receiving operating characteristic (ROC) curves. Moreover, to provide an application-defined tradeoff, a security enhancement approach based on Rényi map is presented, which can further protect the integrity and secrecy of images.     

Surrogate docking: structure-based virtual screening at high throughput speed

Summary Structure-based screening using fully flexible docking is still too slow for large molecular libraries. High quality docking of a million molecule library can take days even on a cluster with hundreds of CPUs. This performance issue prohibits the use of fully flexible docking in the design of large combinatorial libraries. We have developed a fast structure-based screening method, which utilizes docking of a limited number of compounds to build a 2D QSAR model used to rapidly score the rest of the database. We compare here a model based on radial basis functions and a Bayesian categorization model. The number of compounds that need to be actually docked depends on the number of docking hits found. In our case studies reasonable quality models are built after docking of the number of molecules containing 50 docking hits. The rest of the library is screened by the QSAR model. Optionally a fraction of the QSAR-prioritized library can be docked in order to find the true docking hits. The quality of the model only depends on the training set size – not on the size of the library to be screened. Therefore, for larger libraries the method yields higher gain in speed no change in performance. Prioritizing a large library with these models provides a significant enrichment with docking hits: it attains the values of 13 and 35 at the beginning of the score-sorted libraries in our two case studies: screening of the NCI collection and a combinatorial libraries on CDK2 kinase structure. With such enrichments, only a fraction of the database must actually be docked to find many of the true hits. The throughput of the method allows its use in screening of large compound collections and in the design of large combinatorial libraries. The strategy proposed has an important effect on efficiency but does not affect retrieval of actives, the latter being determined by the quality of the docking method itself. Electronic supplementary material is available at http://dx.doi.org/10.1007/s10822-005-9002-6.     

可见光与红外彩色融合图像感知清晰度评价模型

提出了感知清晰度评价模型,来评价人眼对红外与可见光彩色融合图像细节和边缘的可辨识度。首先,利用人眼对比度敏感函数模型,抑制在特定观察条件下图像中人眼不敏感的频率成分。之后,在局部频带对比度模型基础上,结合人眼亮度掩模特性构造了感知对比度模型。最后,计算融合图像人眼兴趣区域(细节和边缘区域)的感知对比度,进而评价融合图像的感知清晰度。实验结果表明,与现有的五种彩色图像清晰(模糊)度的客观评价模型相比,考虑人眼视觉特性感知清晰度模型的计算结果与人眼主观感受具有较好的一致性,可以有效地对彩色融合图像清晰度进行客观评价。     

The Relationship Between Perceptual Evaluation and Objective Multiparametric Evaluation of Dysphonia Severity

SUMMARY: The purpose of this study was to investigate the usefulness of the Dysphonia Severity Index (DSI) as an objective multiparametric measurement in assessing dysphonia. The DSI was compared with the score on Grade of the GRBAS scale. Investigated was also whether the DSI is related to severity of dysphonia, which was represented by different diagnosis groups. Furthermore, it was investigated whether the DSI can differentiate between a group of patients and a control group. A total of 294 patients with different voice pathologies were included. A control group consisted of 118 volunteers without any voice complaints. The voices of all participants were perceptually evaluated on Grade, and the DSI was measured. The groups of patients with voice complaints have a lower DSI and higher scores on Grade than the control group. The DSI was significantly lower when the score on Grade was higher. The DSI discriminates between patients with nonorganic voice disorders, vocal fold mass lesions, and vocal fold paresis/paralysis. To determine whether the DSI discriminates between patients and controls, the sensitivity and specificity for different DSI cutoff points were calculated. With a DSI cutoff of 3.0, maximum sensitivity (0.72) and specificity (0.75) were found. We conclude that the DSI is a useful instrument to objectively measure the severity of dysphonia.     

言语知觉研究

言语知觉是心理学的一个领域，其发展与语音学、语音工程和人工智能等许多学科有关．本文简要介绍了在认知心理学和其它相关学科的推动下，言语知觉发展的主线、当前的主要问题和研究现状。