RESUMO
A Bayesian deep restricted Boltzmann-Kohonen architecture for data clustering termed deep restricted Boltzmann machine (DRBM)-ClustNet is proposed. This core-clustering engine consists of a DRBM for processing unlabeled data by creating new features that are uncorrelated and have large variance with each other. Next, the number of clusters is predicted using the Bayesian information criterion (BIC), followed by a Kohonen network (KN)-based clustering layer. The processing of unlabeled data is done in three stages for efficient clustering of the nonlinearly separable datasets. In the first stage, DRBM performs nonlinear feature extraction by capturing the highly complex data representation by projecting the feature vectors of d dimensions into n dimensions. Most clustering algorithms require the number of clusters to be decided a priori; hence, here, to automate the number of clusters in the second stage, we use BIC. In the third stage, the number of clusters derived from BIC forms the input for the KN, which performs clustering of the feature-extracted data obtained from the DRBM. This method overcomes the general disadvantages of clustering algorithms, such as the prior specification of the number of clusters, convergence to local optima, and poor clustering accuracy on nonlinear datasets. In this research, we use two synthetic datasets, 15 benchmark datasets from the UCI Machine Learning repository, and four image datasets to analyze the DRBM-ClustNet. The proposed framework is evaluated based on clustering accuracy and ranked against other state-of-the-art clustering methods. The obtained results demonstrate that the DRBM-ClustNet outperforms state-of-the-art clustering algorithms.
RESUMO
Endurance and muscle strengthening are required in a number of applications including fitness training, sports and prosthetics. Optimal performance can be achieved by understanding the relationship between muscle activity and angular variation of the elbow joint. Biceps brachii is an important muscle of the upper arm that helps in providing stability during flexion and extension. In this work, an attempt is made to analyze isometric contraction at various elbow angles using surface electromyography and rainflow counting algorithm. The signals are recorded from biceps brachii muscles of twenty healthy subjects while performing well defined protocol using a 6kg dumbbell at three different angles: 30°, 60° and 90°. The recorded signals are preprocessed and subjected to rainflow counting algorithm to obtain rainflow cycles. A new feature, cycle crossing intensity (CCI) is calculated from the rainflow cycles. The intensity peak parameter is computed from each CCI. The results show that the CCIs are distinct for all the three angles of elbow flexion and shifts towards higher magnitude values with increase in the angle of flexion. The average peak intensity is 425, 1073 and 2336 for 30°, 60° and 90° respectively. This feature is found to have high statistical significance for 30°-90° flexion and 60°-90° elbow flexion. The results shows that this method is useful in understanding muscle contractions at various angles of elbow joint. This technique can be extended to analyze muscles in neuromuscular conditions such as fatigue.
