Our work focuses on the interdisciplinary field of detailed analysis of behaviors exhibited by individuals during sessions of distributed collaboration. With a particular focus on ergonomics, we propose new mechanisms...
详细信息
ISBN:
(纸本)9728865600
Our work focuses on the interdisciplinary field of detailed analysis of behaviors exhibited by individuals during sessions of distributed collaboration. With a particular focus on ergonomics, we propose new mechanisms to be integrated into existing tools to enable increased productivity in distributed learning and working. Our technique is to record ocular movements (eye tracking) to analyze various scenarios of distributed collaboration in the context of computer-based training. In this article, we present a low-cost oculometric device that is capable of making ocular measurements without interfering with the natural behavior of the subject. We expect that this device could be employed anywhere that a natural, non-intrusive method of observation is required, and its low-cost permits it to be readily integrated into existing popular tools, particularly E-learning campus.
Visual Motor Integration tests, which involve a subject copying geometric shapes, are often used as one of a battery of tests to assess the needs of a child who may have a specific learning difficulty (SpLD). As part ...
详细信息
ISBN:
(纸本)0889865477
Visual Motor Integration tests, which involve a subject copying geometric shapes, are often used as one of a battery of tests to assess the needs of a child who may have a specific learning difficulty (SpLD). As part of the Dyslexia Early Screening Test (DEST), the resulting free-hand sketches are assessed by an expert who analyses them visually and decides on the degree of similarity between the sketches and a shape copying scoring template. The assessment is time-consuming and rather subjective. In this paper, we investigate the use of Zernike moment descriptors as a feature extraction technique for training a k-nearest neighbour classifier to recognise and automatically assign scores to a set of hand-sketched shapes. A prototype shape copying assessment system DESCAR has been implemented using the Matlab programming environment. Scoring classification accuracy has been evaluated on a test corpus of 840 sketches comprising 120 different drawings of each of 7 different shapes used in the DEST study [16]. Experimental results show that machine score accuracy rates in the range 63.6-77.9% can be obtained in the correct assignment of scores when compared to a human expert assessor. Accuracy rates depend on geometric shape, order of Zernike moments and choice of classification test used.
A novel algorithm FFSPAN (Fast Frequent Sequential patternmining algorithm) is proposed in this paper. FFSPAN mines all the frequent sequential patterns in large datasets, and solves the problem of searching frequent...
详细信息
ISBN:
(纸本)3540459162
A novel algorithm FFSPAN (Fast Frequent Sequential patternmining algorithm) is proposed in this paper. FFSPAN mines all the frequent sequential patterns in large datasets, and solves the problem of searching frequent sequences in a sequence database by searching frequent items or frequent itemsets. Moreover, the databases that FFSPAN scans keep shrinking quickly, which makes the algorithm more efficient when the sequential patterns are longer. Experiments on standard test data show that FFSPAN is very effective.
This paper presents IPD (intelligent pervasive middleware) that provides automatic home services (consumer electronics: TV, DVD, audio, light, and air-conditioner) for human through analysis of the biometrics and envi...
详细信息
ISBN:
(纸本)3540380914
This paper presents IPD (intelligent pervasive middleware) that provides automatic home services (consumer electronics: TV, DVD, audio, light, and air-conditioner) for human through analysis of the biometrics and environment contexts. The IPD receives the biometrics context (pulse, facial expression and body temperature, human location in smart home and human motion) from sensor devices. We handled the context's pattern analysis in two steps. The first step selects consumer electronics (TV, DVD, audio, air-conditioner, light, project) from IPD's rules. In the second step, IPD predicts detailed home service (for example, a detailed home service of the TV includes news, sports, and drama), using the supervised algorithm-based pattern analyzer. We used the SVM (support vector machine) for detailed service pattern analysis. We experimented on the intelligent pervasive middleware in two directions, and it was shown to have an effective performance in practical application. We are currently studying the association technique of home service (by using datamining) that can happen when IPD predicts home service by the home service predictor.
A conceptual clustering program CLUSTER3 is described that, given a set of objects represented by attribute-value tuples, groups them into clusters described by generalized conjunctive descriptions in attributional ca...
详细信息
ISBN:
(纸本)1845641787
A conceptual clustering program CLUSTER3 is described that, given a set of objects represented by attribute-value tuples, groups them into clusters described by generalized conjunctive descriptions in attributional calculus. The descriptions are optimized according to a user-designed multi-criterion clustering quality measure. The clustering process in CLUSTER3 depends on a viewpoint underlying the clustering goal, and employs the view-relevant attribute subsetting method (VAS) that selects for clustering only attributes relevant to this viewpoint. The program is illustrated by a simple designed problem and by its application to clustering of US Congressional voting records. The ongoing research concerns application of CLUSTER3 to large and complex datasets such as collections of web pages.
Today, the performance of even the best state-of-the-art Automatic Speech recognition (ASR) tends to deteriorate obviously when speech is transmitted over telephone lines. How to improve ASR robustness in noisy channe...
详细信息
ISBN:
(纸本)0889865477
Today, the performance of even the best state-of-the-art Automatic Speech recognition (ASR) tends to deteriorate obviously when speech is transmitted over telephone lines. How to improve ASR robustness in noisy channel environments becomes a life and death problem for many real applications. The challenge in addressing such network environments is that they change every moment and show quite different characteristics in terms of signal-to-noise ratio (SNR), stationarity and spectral structure. Previous adaptation methods with complex parameterization could not follow these channel-related variations reliably during the process of a single utterance. So an online adaptation especially designed for noisy channel environments is necessary. In this paper, a prototype library is established to describe acoustic similarities by exploring large amount of channel-contaminated data. The pre-calculated statistics of this library makes it possible to implement a fast channel selection reliably. Furthermore, a Bayesian learning scheme is developed to compensate channel distortion dynamically through a linear interpolation across the library. In our experiments, the new method leads to 10% relative reduction in Word Error Rate (WER) with respect to conventional Maximum Likelihood Linear Regression (MLLR).
We describe the systems submitted to the NIST RT06s evaluation for the Speech Activity Detection (SAD) and Speaker Diarization (SPKR) tasks. For speech activity detection, a new analysis methodology is presented that ...
详细信息
ISBN:
(纸本)9783540692676
We describe the systems submitted to the NIST RT06s evaluation for the Speech Activity Detection (SAD) and Speaker Diarization (SPKR) tasks. For speech activity detection, a new analysis methodology is presented that generalizes the Detection Erorr Tradeoff analysis commonly used in speaker detection tasks. The speaker diarization systems are based on the TNO and ICSI system submitted for RT05s. For the conference room evaluation Single Distant Microphone condition, the SAD results perform well at 4.23% error rate, and the 'HMM-BIC' SPKR results perform competatively at an error rate of 37.2% including overlapping speech.
Various approaches are presented to solve the growing spam problem. However, most of these approaches are inflexible to adapt to spam dynamically. This paper proposes a novel approach to counter spam based on. spam be...
详细信息
ISBN:
(纸本)3540459162
Various approaches are presented to solve the growing spam problem. However, most of these approaches are inflexible to adapt to spam dynamically. This paper proposes a novel approach to counter spam based on. spam behavior recognition using Decision Tree learned from data maintained during transfer sessions. A classification is set up according to email transfer patterns enabling normal servers to detect malicious connections before mail body delivered, which contributes much to save network bandwidth wasted by spams. An integrated Anti-Spam framework is founded combining the Behavior Classification with a Bayesian classification. Experiments show that the Behavior Classification has high precision rate with acceptable recall rate considering its bandwidth saving feature. The integrated filter has a higher recall rate than either of the sub-modules, and the precision rate remains quite close to the Bayesian Classification.
The performance of individual classifiers applied to complex data sets has for predictive toxicology a significant importance. An investigation was conducted to improve classification performance of combinations of cl...
详细信息
ISBN:
(纸本)3540459162
The performance of individual classifiers applied to complex data sets has for predictive toxicology a significant importance. An investigation was conducted to improve classification performance of combinations of classifiers. For this purpose some representative classification methods for individual classifier development have been used to assure a good range for model diversity. The paper proposes a new effective multi-classifier system based on Dempster's rule of combination of individual classifiers. The performance of the new method has been evaluated on seven toxicity data sets. The classification accuracy of the proposed combination models achieved, according to our initial experiments, 2.97% better average than that of the best individual classifier among five classification methods (Instance-based learning algorithm, Decision Tree, Repeated Incremental Pruning to Produce Error Reduction, Multi-Layer Perceptrons and Support Vector machine) studied.
Prostate cancer remains one of the leading causes of cancer death worldwide, with a reported incidence rate of 650,000 cases per annum worldwide. The causal factors of prostate cancer still remain to be determined. In...
详细信息
ISBN:
(纸本)1424401968
Prostate cancer remains one of the leading causes of cancer death worldwide, with a reported incidence rate of 650,000 cases per annum worldwide. The causal factors of prostate cancer still remain to be determined. In this paper, we investigate a medical dataset containing clinical information on 502 prostate cancer patients using the machinelearning technique of rough sets. Our preliminary results yield a classification accuracy of 90%, with high sensitivity and specificity (both at approximately 91%). Our results yield a predictive positive value (PPN) of 81% and a predictive negative value (PNV) of 95% In addition to the high classification accuracy of our system, the rough set approach also provides a rule-based inference mechanism for information extraction that is suitable for integration into a rule-based system. The generated rules relate directly to the attributes and their values and provide a direct mapping between them.
暂无评论