Knowledge-based Features Research Articles

Perceiving Boundaries in Unfamiliar Turkish Makam Music

Music segmentation is a widely researched topic within music perception. Even though there is extensive data on the role of surface structure features and music training (e.g., Deliège, 1987) in segmentation, not much is known yet about the influence of implicit knowledge-based features acquired through musical enculturation. The goal of our study was to fill this gap. Makam music-trained musicians, nonmusicians, and Western listeners marked their segmentations online as they listened to mostly 19th century, unfamiliar Turkish makam tunes, all recorded in a Qānūn timbre on MIDI with retained microtonal structure. In addition, two experts segmented the tunes in a free-time setting. We found considerable within- and across-group agreement, as well as good agreement with the expert segmentations. After transforming each participant’s segmentations into “hits” and “false alarms” based on their match or mismatch with expert segmentations, we observed that musicians overlapped significantly more with expert segmentations than do the other two groups. Segmentations in all three groups were strongly driven by mostly local surface features. Overall, our results are more supportive of a universality claim as proposed by the Gestalt school of psychology than an enculturation claim.

Knowledge-based open Internet of Things service provisioning architecture on beacon-enabled Web of Objects

Web of Objects uses a semantic ontology to create and provide knowledge-based Internet of Things services in terms of virtual objects and composite virtual objects. The objects are created through virtualization with the use of semantic ontology to form the virtual objects, where multiple virtual objects are combined to form the composite virtual objects to offer the services. Beacon-enabled Web of Objects is the extension of the existing web that allows the beacon to broadcast the uniform resource identifier of the real-world object to the nearer mobile devices. Based on the user selection and request, the service is handled and offered by the Web of Objects platform; thus, an architecture on the beacon-enabled Web of Objects has been proposed in this article. To offer the knowledge-based services, a knowledge creation model has been presented. To realize the knowledge-based service features on the proposed architecture, a use case scenario has been presented and a conceptual semantic ontology model has been designed. Finally, to assess the features, a prototype has been implemented and demonstrated on the use case scenario.

Large scale biomedical texts classification: a kNN and an ESA-based approaches

BackgroundWith the large and increasing volume of textual data, automated methods for identifying significant topics to classify textual documents have received a growing interest. While many efforts have been made in this direction, it still remains a real challenge. Moreover, the issue is even more complex as full texts are not always freely available. Then, using only partial information to annotate these documents is promising but remains a very ambitious issue.MethodsWe propose two classification methods: a k-nearest neighbours (kNN)-based approach and an explicit semantic analysis (ESA)-based approach. Although the kNN-based approach is widely used in text classification, it needs to be improved to perform well in this specific classification problem which deals with partial information. Compared to existing kNN-based methods, our method uses classical Machine Learning (ML) algorithms for ranking the labels. Additional features are also investigated in order to improve the classifiers’ performance. In addition, the combination of several learning algorithms with various techniques for fixing the number of relevant topics is performed. On the other hand, ESA seems promising for this classification task as it yielded interesting results in related issues, such as semantic relatedness computation between texts and text classification. Unlike existing works, which use ESA for enriching the bag-of-words approach with additional knowledge-based features, our ESA-based method builds a standalone classifier. Furthermore, we investigate if the results of this method could be useful as a complementary feature of our kNN-based approach.ResultsExperimental evaluations performed on large standard annotated datasets, provided by the BioASQ organizers, show that the kNN-based method with the Random Forest learning algorithm achieves good performances compared with the current state-of-the-art methods, reaching a competitive f-measure of 0.55 % while the ESA-based approach surprisingly yielded unsatisfactory results.ConclusionsWe have proposed simple classification methods suitable to annotate textual documents using only partial information. They are therefore adequate for large multi-label classification and particularly in the biomedical domain. Thus, our work contributes to the extraction of relevant information from unstructured documents in order to facilitate their automated processing. Consequently, it could be used for various purposes, including document indexing, information retrieval, etc.

PlantMirP: an efficient computational program for the prediction of plant pre-miRNA by incorporating knowledge-based energy features.

MicroRNAs are a predominant type of small non-coding RNAs approximately 21 nucleotides in length that play an essential role at the post-transcriptional level by either RNA degradation, translational repression or both through an RNA-induced silencing complex. Identification of these molecules can aid the dissecting of their regulatory functions. The secondary structures of plant pre-miRNAs are much more complex than those of animal pre-miRNAs. In contrast to prediction tools for animal pre-miRNAs, much less effort has been contributed to plant pre-miRNAs. In this study, a set of novel knowledge-based energy features that has very high discriminatory power is proposed and incorporated with the existing features for specifically distinguishing the hairpins of real/pseudo plant pre-miRNAs. A promising performance area under a receiver operating characteristic curve of 0.9444 indicates that 5 knowledge-based energy features have very high discriminatory power. The 10-fold cross-validation result demonstrates that plantMirP with full features has a promising sensitivity of 92.61% and a specificity of 98.88%. Based on various different datasets, it was found that plantMirP has a higher prediction performance by comparison with miPlantPreMat, PlantMiRNAPred, triplet-SVM, and microPred. Meanwhile, plantMirP can greatly balance sensitivity and specificity for real/pseudo plant pre-miRNAs. Taken together, we developed a promising SVM-based program, plantMirP, for predicting plant pre-miRNAs by incorporating knowledge-based energy features. This study shows it to be a valuable tool for miRNA-related studies.

Automated annual cropland mapping using knowledge-based temporal features

Global, timely, accurate and cost-effective cropland mapping is a prerequisite for reliable crop condition monitoring. This article presented a simple and comprehensive methodology capable to meet the requirements of operational cropland mapping by proposing (1) five knowledge-based temporal features that remain stable over time, (2) a cleaning method that discards misleading pixels from a baseline land cover map and (3) a classifier that delivers high accuracy cropland maps (>80%). This was demonstrated over four contrasted agrosystems in Argentina, Belgium, China and Ukraine. It was found that the quality and accuracy of the baseline impact more the certainty of the classification rather than the classification output itself. In addition, it was shown that interpolation of the knowledge-based features increases the stability of the classifier allowing for its re-use from year to year without recalibration. Hence, the method shows potential for application at larger scale as well as for delivering cropland map in near real time.

Cross-domain polarity classification using a knowledge-enhanced meta-classifier

Current approaches to single and cross-domain polarity classification usually use bag of words, n-grams or lexical resource-based classifiers. In this paper, we propose the use of meta-learning to combine and enrich those approaches by adding also other knowledge-based features. In addition to the aforementioned classical approaches, our system uses the BabelNet multilingual semantic network to generate features derived from word sense disambiguation and vocabulary expansion. Experimental results show state-of-the-art performance on single and cross-domain polarity classification. Contrary to other approaches, ours is generic. These results were obtained without any domain adaptation technique. Moreover, the use of meta-learning allows our approach to obtain the most stable results across domains. Finally, our empirical analysis provides interesting insights on the use of semantic network-based features.

On-board Robust Vehicle Detection Using Knowledge-based Features and Motion Trajectory

This paper presents a robust and efficient method for vehicle detection in dynamic traffic environments. First, two adaptive vehicle hypothesis generation methods based on shadow and vehicle wave are presented, and then we assemble these two features into vehicle hypothesis. A hypothesis verification algorithm based on vehicle motion trajectory is proposed, the on-line hypothesis verification algorithm based on vehicle motion trajectory can not only reduce the false positive alarm caused by interferences, but also handle the problem that the classifiers generated in the off-line training phase is closely related to the diversity of positive and negative samples. Quantitative analysis on both public vehicle image datasets and real-time video presents a result of 85.58% detection rate with 4.13% false positive rate. And our algorithm could run as fast as 40ms/frame on PC platform.

The Impact of Technological Regimes on the Establishment of Technological Standards

According to the framework of Technological regimes theory, the Technological regimes can be decomposed into four-dimensional elements: technological opportunities, appropriation, cumulative nature of technological progress and knowledge-based features. We selected 25 groups of data from manufacturing industries in a certain province in 2006 and study the impact of four dimensional elements of technological regimes on the establishment of industry technological standards by the method of Logistic regression. The results show that each dimension of technological regimes affects establishment of the industry technological standards to a certain extent in different directions. The size of a firm and its market needs of new products should also be considered at the same time.

Knowledge-Based Features for Place Classification of Unvoiced Stops

AbstractThe classification of unvoiced stops in consonant–vowel (CV) syllables, segmented from continuous speech, is investigated by features related to speech production. As burst and vocalic transitions contribute to identification of stops in the CV context, features are computed from both regions. Although formants are the truly discriminating articulatory features, their estimation from the speech signal is a challenge especially in unvoiced regions like the release burst of stops. This may be compensated partially by sub-band energy-based features. In this work, formant features from the vocalic region are combined with features from the burst region comprising sub-band energies, as well as features from a formant tracking method developed for unvoiced regions. The overall combination of features at the classifier level obtains an accuracy of 84.4%, which is significantly better than that obtained with solely sub-band features on unvoiced stops in CV syllables of TIMIT.

Feature Recognition System for Rotational Features

This paper presents the development of a knowledge-based features recognition system for mechanical parts. The main aim of the research is to concentrate on recognition of rotational features, namely, blind hole, through hole and slanting hole to be used in automatic process planning system. The rule-based technique applied to recognise rotational features. The recognition criteria are based upon geometrical data from STandard for Exchange of Product data file (STEP) of computer-aided design (CAD) system. The system consists of several modules, namely, pre-processing module, inference engine module and user interface module. The developed system succeed the design activities in manufacturing industries.

Discovering Fine-grained Sentiment in Suicide Notes

This paper presents our solution for the i2b2 sentiment classification challenge. Our hybrid system consists of machine learning and rule-based classifiers. For the machine learning classifier, we investigate a variety of lexical, syntactic and knowledge-based features, and show how much these features contribute to the performance of the classifier through experiments. For the rule-based classifier, we propose an algorithm to automatically extract effective syntactic and lexical patterns from training examples. The experimental results show that the rule-based classifier outperforms the baseline machine learning classifier using unigram features. By combining the machine learning classifier and the rule-based classifier, the hybrid system gains a better trade-off between precision and recall, and yields the highest micro-averaged F-measure (0.5038), which is better than the mean (0.4875) and median (0.5027) micro-average F-measures among all participating teams.

Direct multi-objective optimization of parametric geometrical models stored in PLM systems to improve functional product design

This paper presents a functional analysis and design method, including an optimization loop, which has been integrated into our self-developed, web-based Product Lifecycle Management (PLM) platform. Our design methodology, which includes advanced and parametric CAD modelling and direct multi-objective optimization is described here. Functional design and knowledge-based engineering features, such as functional parameters, expert rule definitions and design experience feedback are all developed in the product design process in order to reduce costs, lead time and also to improve product quality and value. To confirm our research hypotheses, an experimental case-study is chosen: the ground-link system of a racing car design and manufacturing project, including conceptual, embodiment, detailed design and manufacturing phases.

Approach for Automated Product Modeling Using Knowledge-Based Design Features

AbstractThe globalization of the markets, increasing complexity and diversification of products as well as product and process innovations as central factors of business competition have forced companies to create new strategies of product development. This paper points out the possibility of standardization in the modeling phase in consideration of calculation results and conceptual design activities. The introduced approach is based on user defined features (UDFs) as knowledge carrier, whereas these UDFs have full control over the model setup, initialized by a few input parameters. In consideration of company specific model properties, the augmentation of UDFs with knowledge provides all information required regarding the shape, function and manufacturing of the product for all phases of the product life cycle. By means of an exemplary realization of the illustrated concept this paper makes the approach accessible to the reader. Finally the paper will be concluded with prospects in future challenges of ...

Putting the pathway back into protein folding

The article by Liwo et al. in this issue of PNAS (1) on ab initio simulations of the folding pathway of a number of representative small proteins marks a renaissance in efforts to simulate the mechanism of protein folding without prior knowledge of the native structure. While there has been recent progress in predicting the three-dimensional native structure of a protein (2, 3), the most successful approaches incorporate many knowledge-based features (e.g., use of already solved protein structures and predicted secondary structure) that render the assembly mechanism provided by such simulations physically meaningless. On the other hand, the first principles of simulation of the folding process at complete atomic detail, including water that starts from the random state and finishes with the native structure, is computationally intractable for all but the simplest systems (4). For the near future, as in Liwo et al. (1), reduced models that describe the protein by a subset of its constituent atoms and implicitly treat solvent and that search conformational space by using Langevin or molecular dynamics (5) offer the most promising way of exploring how proteins fold. Remarkably, although Liwo et al. (1) examined the folding of only seven proteins, representing all secondary structural classes, their simulations demonstrate a richness that addresses key questions in protein folding. Is there native-like secondary (6) or even tertiary structure in the denatured state (7)? Does a protein collapse first and then the secondary structure (helices and β-sheets) appears, or does the secondary structure appear first, and then the native tertiary assembles (8)? Alternatively, perhaps secondary and at least loose, tertiary structures simultaneously assemble (9)? Partial evidence supporting the first and third scenarios is provided in the Liwo et al. (1) article. Furthermore, a molten globule state (10) is believed to form quite rapidly, with more or …

AUTOMATIC 3D PROSTATE SURFACE DETECTION FROM TRUS WITH LEVEL SETS

Prostate boundary detection from ultrasound images plays an important role in prostate disease diagnoses and treatments. However, due to the low contrast, speckle noise and shadowing in ultrasound images, this still remains a difficult task. Currently, prostate boundary detection is performed manually, which is arduous and heavily user dependent. A possible solution is to improve the efficiency by automating the boundary detection process with minimal manual involvement. This paper presents a new approach based on the level set method to automatically detect the prostate surface from 3D transrectal ultrasound images. The user interaction in the initialization procedure is relieved by automatically putting the centroid of the initial zero level sets close to the image center. Region information, instead of the image gradient, is integrated into the level set method to remedy the "boundary leaking" problem caused by gaps or weak boundaries. Moreover, to increase the accuracy and robustness, knowledge-based features, such as expected shape (kidney-like) and ultrasound appearance of the prostate (looking from within the gland, the intensities are transitions from dark to light), are also incorporated into the model. The proposed method is applied to eight 3D TRUS images and the results have shown its effectiveness.

Information fusion approach to microcalcification characterization

The paper presents an information fusion-based approach to one of the most challenging problems in mammogram interpretation: the problem of characterizing mammographic microcalcifications as benign or malignant. There are two categories of methods typically used for designing decision aids for diagnosis of microcalcifications: computer vision methods employing intensity-based features automatically extracted from images and methods using mammogram characteristics considered by human experts. The achieved recognition accuracy of both types of methods is not yet sufficient for them to be utilized in clinical practice. The paper introduces a hybrid system combining decisions of classifiers utilizing both domain knowledge-based and intensity-based features within the framework of the Evidence theory. The system comprises a hierarchical evidential classifier employing a combination of texture features of individual microcalcifications and a neural network employing cluster features observed and described by a radiologist. The results of a pilot study have shown that a false alarm rate of the hybrid system is lower than the false alarm rate of each single classifier used in the combination as well as that of the radiologists participated in the study.