关键词： PSO-SVM hybrid training   PSO algorithm   SVM multiclass   Speech patterns recognition  

摘要： This paper proposes the implementation of an Automatic Speech Recognition (ASR) process through extraction of Mel-Frequency Cepstral Coefficients (MFCCs) from voice signal commands, application of the Discrete Cosine Transform (DCT) in these coefficients, Support Vector Machine (SVM) training optimized by the Particle Swarm Optimization (PSO) technique in order to speed up the whole process and using One Against All (OAA) multiclass SVM classification. The main contribution is in training phase that it is the combination of SVM with PSO algorithm, resulting in computational load and processing time reduction. This novel algorithm is called here as PSO-SVM hybrid training application and its performance is shown as the experimental results of voice signal commands in Brazilian Portuguese language. Such commands comprise 10 isolated digits (from zero to nine) and 20 action commands such as "go ahead", "finish", "pause", etc.;that is, there are 30 different pattern types (classes) to be separated (recognized). The process is speaker independent type, that is, the voice bank used in training is different from the one used in tests. The obtained results presented success rates of 92% to 99% during the tests for the classifier using RBF kernel function. Besides, the comparison section shows that this technique is 25 times faster than the recognition without optimization and also, it presents 10% of improvement in recognition success rate when compared to the well-known technique, Gaussian Mixture Models (GMM) algorithm. In addition, the proposed algorithm can be applied in any data processing board for voice signals (DSP, FPGA, DSPIC, ...).

关键词： Supervise machine learning algorithm   Classification   Fall risk   Prospective   Older adults  

摘要： Introduction: Given their major health consequences in the elderly, identifying people at risk of fall is a major challenge faced by clinicians. A la of studies have confirmed the relationships between gait parameters and falls incidence. However, accurate tools to predict individual risk among independent older adults without a history of falls are lacking. Objective: This study aimed to apply a supervised learning algorithm to a data set recorded in a two-year longitudinal study, in order to build a classification tree that could discern subsequent fallers based on their gait patterns. Methods: A total of 105 adults aged > 65 years, living independently at home and without a recent fall history were included in a two-year longitudinal study. All underwent physical and functional assessment. Gait speed, stride length, frequency, symmetry and regularity, and minimum toe clearance were recorded in comfortable, fast and dual task walking conditions in a standardized laboratory environment. Fall events were recorded using personal falls diaries. A supervised machine learning algorithm (J48) has been applied to the data recorded at inclusion in order to obtain a classification tree able to identify future fallers. Results: Based on fall information from 96 volunteers, a classification tree correctly identifying 80% of future fallers based on gait patterns, gender, and stiffness, was obtained, with accuracy of 84%, sensitivity of 80%, specificity of 87%, a positive predictive value of 78%, and a negative predictive value of 88%. Discussion: While the performances of the classification tree warrant further confirmation, it is the first predictive tool based on gait parameters that are identified (not clustered) allowing its use by other research teams. Conclusion: This original longitudinal pilot study using a supervised machine learning algorithm, shows that gait parameters and clinical data can be used to identify future fallers among independent older adults.

关键词： histology education   microscopic anatomy   eye gaze   feedforward training   visual literacy   pattern recognition skills   digital histology   expert eye movement model  

摘要： Histology is a visually oriented, foundational anatomical sciences subject in professional health curricula that has seen a dramatic reduction in educational contact hours and an increase in content migration to a digital platform. While the digital migration of histology laboratories has transformed histology education, few studies have shown the impact of this change on visual literacy development, a critical competency in histology. The objective of this study was to assess whether providing a video clip of an expert's gaze while completing leukocyte identification tasks would increase the efficiency and performance of novices completing similar identification tasks. In a randomized study, one group of novices (n = 9) was provided with training materials that included expert eye gaze, while the other group (n = 12) was provided training materials with identical content, but without the expert eye gaze. Eye movement parameters including fixation rate and total scan path distance, and performance measures including time-to-task-completion and accuracy, were collected during an identification task assessment. Compared to the control group, the average fixation duration was 13.2% higher (P < 0.02) and scan path distance was 35.0% shorter in the experimental group (P = 0.14). Analysis of task performance measures revealed no significant difference between the groups. These preliminary results suggest a more efficient search performed by the experimental group, indicating the potential efficacy of training using an expert's gaze to enhance visual literacy development. With further investigation, such feedforward enhanced training methods could be utilized for histology and other visually oriented subjects.

关键词： Electromyography   Pattern recognition   Artificial neural network   Armband system   Hand prosthesis  

摘要： A robust algorithm to classify various hand postures using EMG signals is needed for the EMG-based electric hand prosthesis with the multiple degrees of freedom. In this study, an armband-type multi-channel EMG module was designed, and an algorithm for classifying seven different types of hand postures was developed using the artificial neural network (ANN). The classification accuracy was evaluated for ten normal volunteers, according to the independence of the EMG feature groups, donning and doffing training data size, and whether or not majority voting was applied. The results revealed an optimized accuracy of 97.49 +/- 3.87% when majority voting was applied after using high independence feature group (HIFG) to perform classification training for seven or more sessions. The algorithm was successfully applied to provide seven different hand postures in a 5-finger myoelectric hand prosthesis. Confusion matrices and separability indexes of ANN classifiers showed that the major misclassifications, in spite of a good accuracy, were found to be lateral pinch versus palmar pinch, and index versus thumb-up. However, with the classification training for seven or more sessions, the probability of misclassification significantly decreased.

关键词： chromosomal microdeletion   ocular malformation   PITX2 mutation   PITX2-related tooth agenesis pattern  

摘要： Objectives To investigate the mutations in patients with Axenfeld-Rieger syndrome (ARS) and the pattern of PITX2-related tooth agenesis. Methods Whole-exome sequencing (WES) and copy number variation (CNV) array were used to screen the mutations in four ARS probands. After Sanger sequencing and quantitative polymerase chain reaction (qPCR) validation, secondary structure prediction and dual-luciferase assay were employed to investigate the functional impact. Eighteen PITX2-mutated patients with definite dental records were retrieved from our database and literatures, and the pattern of PITX2-related tooth agenesis was analyzed. Results A novel de novo segmental deletion of chromosome 4q25 (GRCh37/hg19 chr4:111, 320, 052-111, 754, 236) encompassing PITX2 and three novel PITX2 mutations c.148C > T, c.257G > A, and c.630insCG were identified. Preliminary functional studies indicated the transactivation capacity of mutant PITX2 on Distal-less homeobox 2 (DLX2) promoter was compromised. The maxillary teeth showed significantly higher rate of agenesis (57.94%) than the mandibular teeth (44.05%). The most often missing teeth were upper lateral incisors (83.33%) and upper second premolars (69.44%). Teeth with the least agenesis rate were the lower second molars (19.44%) and lower first molars (8.33%). Conclusions We identified a novel 4q25 microdeletion including PITX2 and three novel PITX2 mutations, and statistically analyzed the PITX2-related tooth agenesis pattern.

关键词： Image pattern recognition   Machine vision   Tool scraping wear   Surface texture  

摘要： An image pattern recognition system, consisting of a CMOS camera, a set of illumination devices, and a lab-developed machine vision image analysis software, has been developed for evaluation of tool scraping wear conditions. The images of the scraped surface texture on the workpiece were generated by lab-developed automatic scraping machine under controlled conditions;the image was first transformed by Haar wavelet transform with Otsu's algorithm to discriminate the vertical details of surface patterns, and they were then emphasized by weighted calculations according to tool scraping wear conditions. The experimental results show that the average difference between the data obtained from the surface image analysis and the roughness measurement is about +/- 2%, proving the feasibility of the proposed system. Furthermore, a statistic analysis reveals that the standard deviation of the non-zero proportion of the medium frequency domain obtained by image processing might be considered a significant reference for tool replacement. For the field applications, the proposed system provides the capability of online scraping tool wear evaluation for scraping process in manufacturing site, compared with the traditional scraping tool wear evaluation method, which was time consuming;user needs to remove the workpiece to measure instrument.

关键词： Force myography   electromyography   control   regression   classification   usability  

摘要： Objective: Force myography (FMG), which measures the surface pressure profile exerted by contracting muscles, has been proposed as an alternative to electromyography (EMG) for human-machine interfaces. Although FMG pattern recognition-based control systems have yielded higher offline classification accuracy, comparatively few works have examined the usability of FMG for real-time control. In this work, we conduct a comprehensive comparison of EMG- and FMG-based schemes using both classification and regression controllers. Methods: A total of 20 participants performed a two-degree-of-freedom Fitts' Law-style virtual target acquisition task using both FMG- and EMG-based classification and regression control schemes. Performance was evaluated based on the standard Fitts' law testing metrics throughput, path efficiency, average speed, number of timeouts, overshoot, stopping distance, and simultaneity. Results: The FMG-based classification system significantly outperformed the EMG-based classification system in both throughput (0.902 +/- 0.270) versus (0.751 +/- 0.309), (rho < 0.001) and path efficiency (87.2 +/- 8.7) versus (83.2 +/- 7.8), (rho < 0.001). Similarly, FMG-based regression significantly outperformed EMG-based regression in throughput (0.871 +/- 0.2) versus (0.69 +/- 0.3), (rho < 0.001) and path efficiency (64.8 +/- 5.3) versus (58.8 +/- 7.1), (rho < 0.001). Conclusions: The FMG-based schemes outperformed the EMG-based schemes regardless of which controller was used. This provides further evidence for FMG as a viable alternative to EMG for human-machine interfaces. Significance: This work describes a comprehensive evaluation of the online usability of FMG- and EMG-based control using both sequential classification and simultaneous regression control.

关键词： Semiconductor device modeling   Feature extraction   Pattern recognition   Computational modeling   Databases   Clustering algorithms   Analytical models   Semiconductor defects   pattern recognition   neural networks   information retrieval  

摘要： Wafer map defect pattern contains the critical information about semiconductor manufacturing, effective defect analysis technology can improve the yield of products. This paper develops a wafer defect pattern recognition and analysis method based on convolutional neural network (CNN). We build an 8-layer CNN model to inspect wafer map defect and a 13-layer model to classify defect patterns. To analyze the defects, we first extract features based on the classification network, then perform PCA dimensionality reduction and similarity ranking, we infer the root causes of tested samples according to the retrieved wafer maps. In particular, we analyze the impacts of different network layers and feature dimensions on image retrieval performance, it turns out that the appropriate dimensionality reduction can increase the accuracy and speed of wafer map retrieval. The models are trained and tested on the WM-811K wafer map dataset, which are collected from real wafer maps. The experimental results show that the proposed method is able to identify common wafer defect patterns and analyze the root causes accurately and effectively.

关键词： Phenols  

摘要： In this paper, acoustic emission (AE) signals obtained during quasi-static crack propagation in adhesively bonded beech wood were classified using an unsupervised pattern recognition method. Two ductile one-component polyurethane (1C-PUR) adhesives with the same formulation except for one system being reinforced with short polyamide (similar to 1 mm long) fibers were compared to a relative brittle phenol-resorcinol-formaldehyde (PRF) adhesive. Using only localized AE signals, it was shown that the signals originating from the crack propagation could be classified into two different clusters. Comparing the AE signals with a new fractography method, it was estimated that different clusters are due to distinct failure mechanisms, with signals of cluster 1 being associated with wood failure and signals of cluster 2 with adhesive failure. The obtained results suggest that for the PRF adhesive the wood fibers help to slow down the crack propagation. A similar but lesser effect was noted for the polyamide fibers added to the 1C-PUR adhesive matrix.

关键词： nonparametric pattern recognition algorithm   maximum likelihood criterion   multidimensional kernel probability density estimate   quick choice of blur coefficients   asymptotic properties of probability density estimates of Rosenblatt   Parzen type  

摘要： A fast procedure is proposed for choosing the blur coefficients of kernel functions in a multidimensional nonparametric estimation of the equation of a decision surface for a two-alternative problem of pattern recognition. The decision classification rule meets the maximum likelihood criterion. The theoretical basis of the procedure under consideration is the result of a study of the asymptotic properties of multidimensional nonparametric estimates of the decision function in the problem of recognizing patterns and probability densities of the distribution of random variables in classes. The possibility of using fast procedures for choosing the blur coefficients of kernel estimates of probability densities in the synthesis of non-parametric estimates of the equation of the decision surface between classes is substantiated. The effectiveness of the proposed approach is confirmed by the results of computational experiments.