Rajdeep Chatterjee

EEG Based Motor Imagery Classification Using SVM and MLP

This paper focuses on the classification of motor imagery of the left-right hand movements from a healthy subject. Elliptic Bandpass filters are used to discard the unwanted signals. Our study was on C3 and C4 electrodes particularly for the left-right limb movements. We deployed various feature extraction techniques on the EEG data. Statistical-based, wavelet-based energy-entropy & RMS, PSD based average power and bad power were performed to form the desired feature vectors. Variants of Support Vector Machines (SVM) were employed for classification and the results were also compared with Multi-layered Perceptron (MLP). Empirical results show that both SVM and MLP were suitable for such motor imagery classifications with the accuracy of 85% and 85.71% respectively. Among all employed feature extraction techniques wavelet-based methods specifically the energy-entropy feature set, gave promising results for both the classifiers.

A Survey on Fuzzy Association Rule Mining Methodologies

Abstract Fuzzy association rule mining (Fuzzy ARM) uses fuzzy logic to generate interesting association : rules. These association relationships can help in decision making for the solution of a given problem. Fuzzy ARM is a variant of classical association rule mining. Classical association rule mining uses the concept of crisp sets. Because of this reason classical association rule mining has several drawbacks. To overcome those drawbacks the concept of fuzzy association rule mining came. Today there is a huge number of different types of fuzzy association rule mining algorithms are present in research works and day by day these algorithms are getting better. But as the problem domain is also becoming more complex in nature, continuous research work is still going on. In this paper, we have studied several well-known methodologies and algorithms for fuzzy association rule mining. Four important methodologies are briefly discussed in this paper which will show the recent trends and future scope of research in the field of fuzzy association rule mining.

Discernibility matrix based dimensionality reduction for EEG signal

Classification of EEG signals is an important task in Brain Computer Interface (BCI) research. However, the large number of attributes of EEG data is regarded as a curse for classifiers. This paper aims at dimensionality reduction of EEG signals. We use rough set theory to reduce the dimensions of EEG data. In particular, we use discernibility matrix (DM) to compute an indispensable set of attributes of the data so that the attribute set is reduced before classification. We then use Naive Bayes (NB) classifier, Support Vector Machine (SVM) and Extreme Learning Machine (ELM) for classification of the EEG data containing only these attributes. We compare our method with the more popular Principal Component Analysis (PCA). We have used EEG dataset from BCI competition-II to perform the experiments. Accuracy, recall and precision are used as metrics to measure the performance of the classifiers with original dataset, PCA-reduced dataset and DM-reduced dataset. The classification results we obtained for the DM-reduced dataset are found to be as good as for the whole dataset without reduction and generally better than PCA-reduced dataset.

Estimation of Incomplete Data in Mixed Dataset

This paper puts forward a fresh approach which is a modification of original fuzzy kNN for dealing with categorical missing values in categorical and mixed attribute datasets. We have removed the irrelevant missing samples through list-wise deletion. Then, rest of the missing samples is estimated using kernel-based fuzzy kNN technique and partial distance strategy. We have calculated the errors at different percentage of missing values. Results highlight that mixture kernel gives minimum average of MAE, MAPE and RMSE at different missing percentage when implemented on lenses, SPECT heart and abalone dataset.

Realizing New Hybrid Rough Fuzzy Association Rule Mining Algorithm (RFA) Over Apriori Algorithm

Association rules shows us interesting associations among data items. And the procedure by which these rules are extracted and managed is known as association rule mining. Classical association rule mining had many limitations. Fuzzy association rule mining (Fuzzy ARM) is a better alternative of classical association rule mining. But fuzzy ARM also has its limitations like redundant rule generation and inefficiency in large mining tasks. Rough association rule mining (Rough ARM) seemed to be a better approach than fuzzy ARM. Mining task is becoming huge now days. Performing mining task efficiently and accurately over a large dataset is still a big challenge to us. This paper presents the realization of new hybrid mining method which has incorporated the concepts of both rough set theory and fuzzy set theory for association rule generation and shows comparative analysis with Apriori algorithm based on test results of the algorithm over popular datasets.

Image Compression Using VQ for Lossy Compression

The process to minimize the total number of bits required to depict an image is known as image compression. The main goal of image compression is to minimize the transmission cost and to reduce the storage space. Vector quantization is a most popular technique for lossy compression due to its high compression rate and simple decoding algorithm. The key technique of VQ is the codebook design. In this paper, we have compared and found out the compression ratios of jpg and tiff images using VQ for lossy compression and k-means clustering.

Comparative Study of Different Ensemble Compositions in EEG Signal Classification Problem

The leading perspective of this paper is an introduction to three \(\left( Type-I, Type-II,\,and\,Type-III\right) \) types of ensemble architectures in Electroencephalogram (EEG) signal classification problem. Motor imagery EEG signal is filtered and subsequently used for three different types of feature extraction techniques: Wavelet-based Energy and Entropy \(\left( EngEnt\right) \), Bandpower \(\left( BP\right) \), and Adaptive Autoregressive (AAR). Ensemble architectures have been used in various compositions with different classifiers as base learners along with majority voting as the combined method. This standard procedure is also compared with the mean accuracy method obtained from multiple base classifiers. The Type-I ensemble architecture with EngEnt and BP feature sets provides most consistent performance for both majority voting and mean accuracy combining techniques. Similarly, Type-II architecture with EngEnt and AAR feature sets provides most consistent performance for both majority voting and mean accuracy combining techniques. However, the Type-III ensemble architecture contributes highest result \(82.86\%\) with K-Nearest Neighbor \(\left( KNN\right) \) classifier among all three types.

Tourism Recommendation Using Machine Learning Approach

Puri tourism has always remained as the best tourist spot in Odisha. Researchers and town planners have always taken steps in finding out for proper tourism recommendation. But always they have preferred the method of machine learning approach for the tour recommendation models. Some methods give good simulation data but sometimes artificial neural network (ANN) and regression analysis techniques give better results. In this paper, Puri tourism recommendation method has been modelled based on the SOM architect, and by revenue management system. Here, a complete comparison has been described between supervised and unsupervised machine learning technique for tourism recommendation in Puri.

Comparative Analysis of Feature Extraction Techniques in Motor Imagery EEG Signal Classification

Hand movement (both physical and imaginary) is linked to the motor cortex region of human brain. This paper aims to compare the left–right hand movement classification performance of different classifiers with respect to different feature extraction techniques. We have deployed four types of feature extraction techniques—wavelet-based energy–entropy, wavelet-based root mean square, power spectral density-based average power, and power spectral density-based band power. Elliptic bandpass filters are used to discard noise and to extract alpha and beta rhythm which corresponds to limb movement. The classifiers used are Bayesian logistic regression, naive Bayes, logistic, variants of support vector machine, and variants of multilayered perceptron. Classifier performance is evaluated using area under ROC curve, recall, precision, and accuracy.

Dimensionality reduction of EEG signal using Fuzzy Discernibility Matrix

High dimensionality of feature space is a problem in supervised machine learning. Redundant or superfluous features either slow down the training process or dilute the quality of classification. Many methods are available in literature for dimensionality reduction. Earlier studies explored a discernibility matrix (DM) based reduct calculation for dimensionality reduction. Discernibility matrix works only on discrete values. But most real-world datasets are continuous in nature. Use of traditional discernibility matrix approach inevitably incurs information loss due to discretization. In this paper, we propose a fuzzified adaptation of discernibility matrix with four variants of dissimilarity measure to deal with continuous data. The proposed algorithm has been applied on EEG dataset-III from BCI competition-II. The reduced dataset is then classified using Support Vector Machine (SVM). The performance of the proposed Fuzzy Discernibility Matrix (FDM) variants are compared with original discernibility matrix based method and Principal Component Analysis (PCA). In our empirical study, the proposed method outperforms the other two methods, thus suggesting that it is competitive with them.