Anatomization of EMG signals is one of the building blocks of modern prostheses. As the goal is to build robotic arms whose functions are identical to the natural ones, EMG signals produced from various hand gestures ...
详细信息
ISBN:
(数字)9781728130606
ISBN:
(纸本)9781728130613
Anatomization of EMG signals is one of the building blocks of modern prostheses. As the goal is to build robotic arms whose functions are identical to the natural ones, EMG signals produced from various hand gestures and finger movements have received much attention in recent times. Surface EMG signals collected from the upper hand muscles show specific patterns for a particular finger movement, which is also true for combined (more than one) finger movements. Utilizing Digital Signal Processing (DSP), and Machine Learning (ML) techniques this paper proposes a novel method to distinguish among various EMG signals generated from ten different hand gestures. To reduce complexity and make the signals more understandable to the algorithm statistical and frequency features were extracted from the raw EMG signals and used for classification. In order to prove the effectiveness of the method, it was tested on a practical EMG dataset and the results of the experiments are presented.
The Poincare plot is a method to reflect heart rate variability, which has its unique advantages in the diagnosis of arrhythmias. At present, the research on automatic classification of Poincare plots is still in its ...
详细信息
The Poincare plot is a method to reflect heart rate variability, which has its unique advantages in the diagnosis of arrhythmias. At present, the research on automatic classification of Poincare plots is still in its early stage, which greatly affects the clinical application of Poincare plots. Therefore, a new algorithm is proposed in this paper, which makes comprehensive use of the Modified Hausdorff Distance and support vector machine method to do multiple classification of Poincare plots. Then the arrhythmia data in MIT-BIH ECG database were used to test the algorithm. The results showed that the classification accuracy was 90.2%, indicating that the algorithm was basically reliable and could play a certain auxiliary role in clinical Poincare plots based diagnosis classification.
Nowadays a multi-label classification problem arises in different areas for which the significant amount of data has been gained. This problem can be viewed as the one comprising two steps: training some ranking funct...
详细信息
ISBN:
(数字)9789526924441
ISBN:
(纸本)9781728191072
Nowadays a multi-label classification problem arises in different areas for which the significant amount of data has been gained. This problem can be viewed as the one comprising two steps: training some ranking function sorting instances in each class and defining the optimal number of predictions for it. This paper is devoted to the second step of the optimal threshold selection while maximizing the F-macro measure. To do so, we reduce the multi-dimensional problem to the two-dimensional problem of finding a fixed point of a specifically introduced transformation defined on a unit square. We suggest the algorithm of finding the vector of optimal thresholds based on the domain analysis of the introduced transformation. Moreover, we provide the complexity estimations of the proposed algorithm. We evaluate the algorithm on the extreme classification benchmark WikiLSHTC-325K comparing its performance with some baseline results.
In the evolving technology of big data, high velocity data streams play a vital role since pattern of data is being changed over time. The temporal pattern change in data stream leads to a concept evolution called con...
详细信息
ISBN:
(纸本)9781509004331
In the evolving technology of big data, high velocity data streams play a vital role since pattern of data is being changed over time. The temporal pattern change in data stream leads to a concept evolution called concept drift where statistical properties of data differs from time to time and the drift is taken into account in order to update old and outdated classifier and make it adaptable to new data arrival and pattern change over. In order to classify the stream data, a scalable efficient classification algorithm is to be designed which perfectly classifies the data with minimizing misclassification rate in presence of concept drift due to high velocity data. Training time of the classifier must be reduced in order to reduce computational complexity. In this work, a novel algorithm has been implemented using Random Forest with stratified random sampling and Bloom filtering in order to reduce the training time and to handle high velocity data. Experimental results are shown by performing classification with sampling, classification with filtering and classification with sampling and filtering. This enhances the performance of the algorithm by decreasing the training time and testing time of the classifier with negligible compromise in accuracy of classification.
The goal of Non-Rigid Structure from Motion (NRSfM) is to recover 3D shapes of a deformable object from a monocular video sequence. Procrustean Normal Distribution (PND) is one of the best algorithms for NRSfM. It use...
详细信息
ISBN:
(纸本)9781538604953
The goal of Non-Rigid Structure from Motion (NRSfM) is to recover 3D shapes of a deformable object from a monocular video sequence. Procrustean Normal Distribution (PND) is one of the best algorithms for NRSfM. It uses Generalized Procrustes Analysis (GPA) model to accomplish this task. But the biggest problem of this method is that just a few non-rigid points in 2D observations can largely affect the reconstruction performance. We believe that PND can achieve better reconstruction performance by eliminating the affection of these points. In this paper, we present a novel reconstruction method to solve this problem. We present two solutions to simply classify the points into non-rigid and nearly rigid points. After that, we use EM algorithm of PND to recover 3D structure again for nearly rigid points. Experimental results show that the proposed method outperforms the existing state-of-the-art algorithms.
Resume block classification is the most significant step in resume information extraction. However, the existing algorithms applied to resume block classification are all the general text classification algorithms, wh...
详细信息
ISBN:
(数字)9781728162515
ISBN:
(纸本)9781728162522
Resume block classification is the most significant step in resume information extraction. However, the existing algorithms applied to resume block classification are all the general text classification algorithms, which failed to consider the contextual order of each block within a resume. In order to improve the performance of resume block classification, we propose in this paper a block-level bidirectional recurrent neural network model that makes full use of the contextual order relationship among different resume blocks. The experimental results show that the average F1-score value of our model on three 1,400 real resume datasets is 6% to 9% higher than the existing methods.
Compressive Sensing has garnered lot of attention in the fields of Information technology & Signal Processing as it offers an alternate, redundancy-free approach to signal compression and reconstruction. It exploi...
详细信息
Compressive Sensing has garnered lot of attention in the fields of Information technology & Signal Processing as it offers an alternate, redundancy-free approach to signal compression and reconstruction. It exploits the `sparsity' of the signal to under-sample the sequence and reconstruct it without adding aliasing noise as established by Shannon-Nyquist sampling theorem. However, the methods of reconstruct are expensive due to the usage of non-linear reconstruction polynomials. This paper draws comparison between existing sparse signal recovery method, Compressed Sampling Matching Pursuit (CoSaMP), to Iterative Soft Thresholding algorithm (IST), and utilizes IST for classification of Vehicles with Acoustic signal source. The signals are sampled at half the Nyquist rate, followed by reconstruction using IST. Various features like mean, variance, skewness, and kurtosis of signal is extracted from multiple transform domains. The extracted features from the reconstructed signal is fed to K-NN classifier which classifies target signal as a bike, car, tractor, or a bus.
The remote sensing shows a widest perspective for land reclamation in mining areas. Based on how to improve the classification accuracy of mine image, we did some classification researchs with LVQ2 neural network. The...
详细信息
The remote sensing shows a widest perspective for land reclamation in mining areas. Based on how to improve the classification accuracy of mine image, we did some classification researchs with LVQ2 neural network. The proposed method had been applied to the aerial image of Heng country, Guangxi Province. The total classification accuracy was 72%, comparing with the minimum distance method increased by 9%.
In this paper, a local SRC algorithm is presented and four fusion algorithms are obtained. This work is intended to get a better result in both recognition rate and speed. Experimental results based on AR Face Databas...
详细信息
In this paper, a local SRC algorithm is presented and four fusion algorithms are obtained. This work is intended to get a better result in both recognition rate and speed. Experimental results based on AR Face Database are given the comprehensive review of the four fusion algorithms.
With the increasing proportion of renewable energy in the power system, the load, photovoltaic and wind power characteristics show complex dynamic changes, and the traditional single-element scenario classification me...
详细信息
ISBN:
(数字)9798331505905
ISBN:
(纸本)9798331505912
With the increasing proportion of renewable energy in the power system, the load, photovoltaic and wind power characteristics show complex dynamic changes, and the traditional single-element scenario classification method is difficult to cope with this challenge. In order to solve this problem, this paper proposes a two-stage typical joint scene classification method for power systems based on improved iterative selforganizing data analysis technique algorithm (ISODATA) and elemental hierarchical clustering. In the first stage, for load, photovoltaic (PV) and wind power time-series data, the improved ISODATA algorithm is used to reconstruct the time-series data by variance and extreme variance thresholds to consider local differences, and combined with the composite Euclidean dynamic time warping (DTW) metrics to measure the overall differences, which optimizes the classification accuracy of the single-element time-series. In the second stage, on the basis of the elemental typical scenes obtained in the first stage, the interrelationships and dynamic change characteristics of load, PV and wind power are analyzed, the relevant feature quantities are set, and the elements are jointly analyzed by using the hierarchical clustering method, and a typical joint scene of multiple elements is finally obtained. The joint scene not only integrates the characteristics of multiple elements, but also is more complex and comprehensive than the single-element scene. The experimental results show that the method proposed in this paper can effectively improve the accuracy of typical scene classification, accurately capture the complex change characteristics of various elements in the power system, and provide strong data support for intelligent scheduling and optimization.
暂无评论