Lavneet Singh

A novel image watermarking scheme using Extreme Learning Machine

In this paper, a novel digital image watermarking algorithm based on a fast neural network known as Extreme Learning Machine (ELM) for two grayscale images is proposed. The ELM algorithm is very fast and completes its training in milliseconds unlike its other counterparts such as BPN. The proposed watermarking algorithm trains the ELM by using low frequency coefficients of the grayscale host image in transform domain. The trained ELM produces a sequence of 1024 real numbers, normalized as per N(0, 1) as an output. This sequence is used as watermark to be embedded within the host image using Coxs formula to obtain the signed image. The visual quality of the signed images is evaluated by PSNR. High PSNR values indicate that the quality of signed images is quite good. The computed high value of SIM (X, X*) establishes that the extraction process is quite successful and overall the algorithm finds good practical applications, especially in situations that warrant meeting time constraints.

A hybrid approach to increase the performance of protein folding recognition using support vector machines

In area of bioinformatics, large amount of data is being harvested with functional and genetic features of proteins. The data is being generated consists of thousands of features with least observations instances. In such case, we need computational tools to analyze and extract useful information from vast amount of raw data which help in predicting the major biological functions of genes and proteins with respect to their structural behavior. Thus, in this study, we use a new hybrid approach for features selection and classifying data using Support Vector Machine (SVM) classifiers with Quadratic Discriminant Analysis (QDA) as generative classifiers to increase more performance and accuracy. We compare our results with previous results and seem to be much promising. The proposed method provides the higher recognition ratio rather than other method used in previous studies. The obtained results are also compared with other different classifiers and our hybrid classifiers give more accuracy and achieve better results than any other classifiers.

Using hybrid neural networks for identifying the brain abnormalities from MRI structural images

In this study, we present the investigations being pursued in our research laboratory on magnetic resonance images (MRI) of various states of brain by extracting the most significant features, and to classify them into normal and abnormal brain images. We propose a novel method based on deep and extreme machine learning on wavelet transform to initially decompose the images, and then use various features selection and search algorithms to extract the most significant features of brain from the MRI images. By using a comparative study with different classifiers to detect the abnormality of brain images from publicly available neuro-imaging dataset, we found that a principled approach involving wavelet based feature extraction, followed by selection of most significant features using PCA technique, and the classification using deep and extreme machine learning based classifiers results in a significant improvement in accuracy and faster training and testing time as compared to previously reported studies.

A novel machine learning approach for detecting the brain abnormalities from MRI structural images

In this study, we present the investigations being pursued in our research laboratory on magnetic resonance images (MRI) of various states of brain by extracting the most significant features, and to classify them into normal and abnormal brain images. We propose a novel method based ondeep and extreme machine learning on wavelet transform to initially decompose the images, and then use various features selection and search algorithms to extract the most significant features of brain from the MRI images. By using a comparative study with different classifiers to detect the abnormality of brain images from publicly available neuro-imaging dataset, we found that a principled approach involving wavelet based feature extraction, followed by selection of most significant features using PCA technique, and the classification using deep and extreme machine learning based classifiers results in a significant improvement in accuracy and faster training and testing time as compared to previously reported studies.

An Optimal Approach for Pruning Annular Regularized Extreme Learning Machines

Larger datasets, with many samples are problematic for solving problems in data mining and machine learning, due to increase in computational times, increased complexity, and bad generalization due to outliers. Further, the accuracy and performance of machine learning and statistical models are still based on tuning of some parameters and optimizing them for generating better predictive models of learning. In this paper, we propose a novel formulation of Extreme Learning Machines - the Annular ELM, with RANSAC multi model response regularization for pruning large number of hidden nodes to acquire better optimality, generalization and classification accuracy. Experimental evaluation of the proposed ELM formulation on different benchmark datasets showed that the algorithm optimally prunes the hidden nodes, with better generalization and higher classification accuracy as compared to other algorithms, including the well-known SVM, OP-ELM for binary and multi-class classification and regression problems. Also, we extended the proposed algorithm to a more complex application context involving MRI Brain Image classification. For this study, we examine the performance of the proposed algorithm on magnetic resonance images (MRI) of various states of brain by extracting the most significant features, and to classify them into normal and abnormal brain images.

A novel approach to protein structure prediction using PCA or LDA based extreme learning machines

In the area of bio-informatics, large amount of data is harvested with functional and genetic features of proteins. The structure of protein plays an important role in its biological and genetic functions. In this study, we propose a protein structure prediction scheme based novel learning algorithms --- the extreme learning machine and the Support Vector Machine using multiple kernel learning, The experimental validation of the proposed approach on a publicly available protein data set shows a significant improvement in performance of the proposed approach in terms of accuracy of classification of protein folds using multiple kernels where multiple heterogeneous feature space data are available. The proposed method provides the higher recognition ratio as compared to other methods reported in previous studies.

A Comparative Study of Recognition of Speech Using Improved MFCC Algorithms and Rasta Filters

Automatic Speech Recognition has been an active topic of research for the past four decades. The main objective of the automatic speech recognition task is to convert a speech segment into an interpretable text message without the need of human intervention. Many different algorithms and schemes based on different mathematical paradigms have been proposed in an attempt to improve recognition rates. Cepstral coefficients play an important part in speech theory and in automatic speech recognition in particular due to their ability to compactly represent relevant information that is contained in a short time sample of a continuous speech signal. The goal of this paper is to discuss comparison of speech parameterization methods: Mel-Frequency Cepstrum Coefficients (MFCC) and improved Mel-Frequency Cepstrum Coefficients (MFCC) using RASTA filters. Thus, in this study, we try to improve the MFCC algorithms to achieve much accuracy reducing the error rates in Automatic Speech Recognition. First, we remove signal correlation through normalization, then we use RASTA filter to filtering the cepstral coefficients. Finally, we reduce dimension of the cepstral coefficients by the variances of cepstral coefficients in different dimension and obtain our features. By using various classifiers, we try to simulate the speech feature extraction at much optimal and least error rate providing robust method for Automatic Speech Recognition (ASRs).

A novel approach to protein structure prediction using PCA based extreme learning machines and multiple kernels

In the area of bio-informatics, large amount of data is harvested with functional and genetic features of proteins. The structure of protein plays an important role in its biological and genetic functions. In this study, we propose a protein structure prediction scheme based novel learning algorithms --- the extreme learning machine and the Support Vector Machine using multiple kernel learning, The experimental validation of the proposed approach on a publicly available protein data set shows a significant improvement in performance of the proposed approach in terms of accuracy of classification of protein folds using multiple kernels where multiple heterogeneous feature space data are available. The proposed method provides the higher recognition ratio as compared to other methods reported in previous studies.

Email Personalization and User Profiling Using RANSAC Multi Model Response Regression Based Optimized Pruning Extreme Learning Machines and Gradient Boosting Trees

Email personalization is the process of customizing the content and structure of email according to member’s specific and individual needs taking advantage of member’s navigational behavior. Personalization is a refined version of customization, where marketing is done automated on behalf of customer’s user’s profiles, rather than customer requests on his own behalf. There is very thin line between customization and personalization which is achieved by leveraging customer level information using analytical tools. E-commerce is growing fast, and with this growth companies are willing to spend more on improving the online experience.

Pruned Annular Extreme Learning Machine Optimization Based on RANSAC Multi Model Response Regularization

The accuracy and performance of machine learning and statistical models are still based on tuning some parameters and optimization for generating better predictive models of learning is based on training data. Larger datasets and samples are also problematic, due to increase in computational times, complexity and bad generalization due to outliers. Using the motivation from extreme learning machine (ELM), we proposed annular ELM based on RANSAC multi model response regularization to prune the large number of hidden nodes to acquire better optimality, generalization and classification accuracy of the network in ELM. Experimental results on different benchmark datasets showed that proposed algorithm optimally prunes the hidden nodes, better generalization and higher classification accuracy compared to other algorithms, including SVM, OP-ELM for binary and multi-class classification and regression problems.