Vellore Institute of Technology (VIT) is a private university located in&nbsp;Tamil Nadu, India. Founded in 1984, as Vellore Engineering College, the institution offers 20 undergraduate, 34 postgraduate, four integrated and four research programs. It has campuses in Vellore, Amravati, Bhopal and Chennai.

VIT is one of the top ranked private universities in India according to NIRF, THE and QS Rankings.&nbsp;Govt. of India has recognized&nbsp;VIT, Vellore as an&nbsp;Institution of Eminence. This has allowed VIT to take independent quality initiatives and move up in world ranking.

&nbsp;

&nbsp;

VIT University

AbstractAir is the most essential constituent for the sustenance of life on earth. The air we inhale has a tremendous impact on our health and well-being. Hence, it is always advisable to monitor the quality of air in our environment. To forecast the air quality index (AQI), artificial neural networks (ANNs) trained with conjugate gradient descent (CGD), such as multilayer perceptron (MLP), cascade forward neural network, Elman neural network, radial basis function (RBF) neural network, and nonlinear autoregressive model with exogenous input (NARX) along with regression models such as multiple linear regression (MLR) consisting of batch gradient descent (BGD), stochastic gradient descent (SGD), mini-BGD (MBGD) and CGD algorithms, and support vector regression (SVR), are implemented. In these models, the AQI is the dependent variable and the concentrations of NO2, CO, O3, PM2.5 , SO2, and PM10 for the years 2010–2016 in Houston and Los Angeles are the independent variables. For the final forecast, several ensemble models of individual neural network predictors and individual regression predictors are presented. This proposed approach performs with the highest efficiency in terms of forecasting air quality index.

Department of Communication Engineering

School of Electronics Engineering

Vellore Campus

Department of Instrumentation

School of Electrical Engineering

World University root node

Journal of Intelligent Systems

Forecasting Air Quality Index Using an Ensemble of Artificial Neural Networks and Regression Models

Inhaling particulate matter such as PM2.5 can have a hazardous impact on the human health. In order to predict the PM2.5 concentration, Artificial Neural Networks trained with conjugate gradient descent such as Multi Layer Perceptron (MLP), cascade forward neural network, Elman neural network, Radial Basis Function (RBF) neural network and Non-linear Autoregressive model with exogenous input (NARX) along with regression models such as Multiple Linear Regression (MLR) consisting of batch gradient descent, stochastic gradient descent, mini-batch gradient descent and conjugate gradient descent algorithms and Support Vector Regression (SVR) were implemented. In these models, the concentration of PM2.5 was the dependent variable and the data related to concentrations of PM2.5, SO2, O3 and meteorological data including average Maximum Temperature (MAX T), daily wind speed (WS) for the years 2010–2016 in Houston and New York were the independent variables. For the final forecast, several ensemble models of individual neural network predictors and individual regression predictors have been presented. © 2018 Springer-Verlag GmbH Germany, part of Springer Nature

Journal of Ambient Intelligence and Humanized Computing

Prediction of PM2.5 using an ensemble of artificial neural networks and regression models

Air is the most vital constituent for the sustenance of life on earth. Air pollution is the major problem we have been facing. It is important to address this issue to lead a healthy life. Forecasting of air quality will contribute to a healthy society. In this paper, artificial neural network (ANN) predictors trained with conjugate gradient descent have been implemented to forecast air quality index (AQI) in a particular area of interest. Several neural network models such as multilayer perceptron (MLP), Elman, radial basis function and NARX were applied. In these neural network models, four major pollutant concentrations including NO2, CO, O3 and PM10 for the year 2014 to 2016 in Delhi (India) were used to train each predictor. It can be concluded that, among all these models, radial basis function exhibited more accuracy in terms of measures of quality with mean absolute error (MAE) = 7.33, mean absolute percent error (MAPE) = 4.05%, correlation coefficient (R) = 0.993, root mean square error (RMSE) = 9.69 and index of agreement (IA) = 0.99.

International Journal of Environment and Waste Management

Air quality index forecasting using artificial neural networks - a case study on Delhi

In this paper a PEVD (Polynomial Eigen Value Decomposition) based adaptive kalman filter is proposed for Acoustic Echo Cancellation (AEC) in the presence of noise. Acoustic echo is the phenomenon when the speaker listens to his own voice after some delay while speaking to his fellow mate on a call. The presence of acoustic coupling between the loudspeakers and the near-end microphone signal produces an undesired acoustic echo, which reduces the speech quality. Existing AEC system implemented adaptive kalman filter which results in less output efficiency in the noisy environment. In the proposed method, the near end speech is separated from the acoustic echo as well as from the surrounding noise by using PEVD based adaptive kalman filter. Initially mixed microphone signal is pre-processed by Polynomial Eigen Value Decomposition which strongly de-correlates the signal and also de-correlates out the noise. The pre-processed signal is passed through adaptive kalman filter for estimation of the acoustic echo. The efficiency of the output is calculated using Echo Return Loss Enhancement (ERLE). If the ERLE lies in the range of 30-40 dB indicates the good echo cancellation. The simulations show that the proposed PEVD based adaptive kalman method provides higher cancellation of echo. The simulations are being carried out in Matlab 2016a.

Electronics

Chennai Campus

VIT Unitversity

International Journal of Intelligent Engineering and Systems

Acoustic Echo Cancellation Using PEVD Based Adaptive Kalman Filter

This paper aims to establish a non-linear regression model based on Mamdani fuzzy inference system and examine the performance of the model based on various metrics such as correlation, index of agreement, mean absolute percentage error, mean absolute error and root mean square error. The data used in this operation is collected from Delhi Punjabibagh with pollutants spanning from Nitrogen dioxide to particulate matter 10μm with ozone and carbon monoxide considered to comprehensively cover the base of inputs. The resultant air quality index is then measured with the calculated ones based on the metrics discussed above.

2017 International Conference on Trends in Electronics and Informatics (ICEI)

Forecasting air quality index based on Mamdani fuzzy inference system

It is always important to monitor the quality of air that we inhale to protect ourselves from the respiratory diseases. In this paper, we present different regression models to forecast air quality index (AQI) in particular areas of interest. Support vector regression (SVR) and linear models like multiple linear regression consisting of gradient descent, stochastic gradient descent, mini-batch gradient descent were implemented. In these models, the air quality index (AQI) is dependent on pollutant concentrations of NO 2 , CO, O 3 , PM 2.5 , PM 10 and SO 2 . Among these models, support vector regression (SVR) exhibited high performance in terms of investigated measures of quality.

Forecasting air quality index using regression models: A case study on Delhi and Houston

Next generation wireless communication systems promise the subscribers with Giga-bit-data-rate experience at low Bit Error Rate (BER) under adverse channel conditions. In order to maximize the overall system throughput of Orthogonal Frequency Division Multiplexing (OFDM), adaptive modulation is one of the key solutions. In adaptive modulated OFDM, the subcarriers are allocated with data bits and energy in accordance with the Signal to Interference Ratio (SIR) of the multipath channel, which is referred to as adaptive bit loading and adaptive power allocation respectively. The number of iterations required allocating the target bits and energy to a sub channel is optimized. The key choice of the paper is to allocate the bits with minimum number of iterations after clustering the sub channels using fuzzy logic. The proposed method exhibits a faster convergence in obtaining the optimal solution.

Department of Micro and Nanoelectronics

Handbook of Research on Fuzzy and Rough Set Theory in Organizational Decision Making Advances in Business Strategy and Competitive Advantage

Optimized-Fuzzy-Logic-Based Bit Loading Algorithms

Background/Objectives: Object identification in color images is a complex task, especially when the background and lighting conditions of the environment are uncontrolled. Methods/Statistical Analysis: This paper proposes an object identification algorithm to identify the light green tea leaves from the tea plants. Findings: In this paper we examine the wavelet transform, one of the most recent mathematical tools related to image processing, for identifying our desired object. Here, we concentrated on identifying the light green tea leaves. The paper is divided into two main parts. In the first part, we illustrated the compression for a color image, by using two different wavelets transforms i.e. the Haar wavelet and Integer wavelet transform (IWT). In this we have analyzed the behavior of these two wavelet transform with different tea leaves images and identified the most appropriate wavelet transform that can perform better process to compress the given color image. To analyses the performance of the wavelet transform with the color images, we calculated Peak Signal to Noise Ratio (PSNR) and compression ratio. The second part deals with Segmentation, which includes edge segmentation and clustering i.e. K-means. Applications/Improvements: The enhancement has been performed for edges. K-means clustering is used to separate the colors from the image, whereas edge based segmentation gives the shapes of the desired regions. Finally the light green tea leaves are identified by using above mentioned methods. These theoretical aspects are illustrated through MATLAB.

Fulltext

Indian Journal of Science and Technology

Object Identification using Wavelet Transform

This paper proposes the multimodal biometrics system for identity verification using two traits i.e., fingerprint and iris. The proposed system is designed for applications where the training database contains an iris and fingerprint images for each individual. The final decision is made by fusion at “matching score level architecture” in which feature vectors are created independently for query images and are then compared to the enrolment templates which are stored during database preparation for each biometric trait. When a new iris and fingerprint in given it checks with the database and if it matches both then the person’s details is shown otherwise it displays it is not authorized. Once it gets authorized it displays person’s details. © Research India Publications.

International Journal of Applied Engineering Research

Design and development of biometric recognition and matching system

Verification of speakers is the use of the voice pattern to check the authenticity of the individual. In this paper the speaker verification system has been implemented on a real time DSP processor TMS3206713 with a new technique is proposed. Feature extraction and speaker modeling are the two major steps involved in speaker verification systems. Being the most frequently used features extracted from the human speech signal, the Mel Frequency Cepstral Coefficients (MFCC) have been considered for feature extraction. In the speaker modeling, a new technique known as Multi Section Vector Quantization (MSVQ) is implemented in addition to normal Vector Quantization. These techniques MFCC and VQ are generally used to recognize the speaker from a database of trained users. Using Multi Section Vector Quantization we can compare the trained codebook of particular user with the utterance and the authenticity decision can be made. Thus the advantage of using Multi Section Vector Quantization in speaker verification system is being discussed and the efficiency of this system is presented. To make this real time system robust in most of the environments a noise filter is also added during the pre-processing stage.

Real Time Implementation of Speaker Verification System

This paper proposes for a computer-aided tool to detect the severity of Tuberculosis (TB) in patients from their respective Chest X-ray (CXR) images. The collected dataset of CXRs with the respective severity is used to extract the feature matrix by valid medical observations and monitoring. This extracted feature matrix is further used to prepare the Back Propagation Neural Network in MATLAB using the MATLAB toolbox. The neural network thus created achieves an accuracy of 96% on using 12 hidden neurons and can be used to determine the severity of TB that is affecting the patient. The severity levels are: absent, mild, moderate, severe, and proliferate. This system can help the patients as well as the doctors with what further steps of treatment should be followed, and how prompt the further action needs to be. © Research India Publications.

Severity detection of tuberculosis using neural network

In many applications like video conferencing, teleconferencing, speaker recognition systems and mobile communication etc., the quality of speech is degraded due to the cocktail problem. The solution to this problem is to separate the speech signals using Independent component analysis. In this paper we proposed Wavelet based Maximum likelihood estimation technique and Wavelet based Minimization of mutual information of ICA techniques to separate the speech signals. The performance of the speech separation is measured using Signal to Distortion Ratio (SDR). The wavelet based ICA simulation in Matlab gives improved SDR compared to conventional ICA. © Research India Publications.

Speech separation using wavelet based independent component analysis

Electronic Dance Music (EDM) is a class of music produced primarily using electronic instruments and computers comprising several sub-genres. Disc jockeys (DJs) mix EDM songs from similar sub-genres for live performances in concerts and festivals. Given a large database of electronic music, if a system can classify songs into their respective sub-genres, it can greatly help producers and DJs while making mixes. We describe an approach to classify four major sub-genres of EDM using artificial neural networks. We extracted several rhythmic and statistical features from a database of 400 songs, evenly distributed among the four sub-genres, and used these features to train a Back Propagation Network (BPN) and a Probabilistic Neural Network (PNN). In addition to classification, we also employed the RELIEFF feature selection algorithm to reduce the number of features and improve our results. Comparing the performances of the two networks on the basis of their correct classifications, we observed that BPN combined with the features proposed yielded a better result of 94.16%. © 2014 IEEE.

2014 IEEE International Symposium on Signal Processing and Information Technology (ISSPIT)

Intelligent classification of electronic music

Neural networks have been an increasingly growing field since the past 50 years, both in terms of research and implementation. There have been numerous applications in industrial automation, medicine, computers, consumer appliances, electronic gadgets, etc., which have had a significant impact by changing the previously used conventional technologies. Neural networks have redefined the way machines are designed to function more accurately, and more adaptively. In this paper, we design and analyze few of the popular neural network algorithms for their application in a fire detection system, and implemented for the application, which makes the system "intelligent". The algorithms are designed based on sample inputs from the sensors; they are trained and then analyzed on standard performance metrics. © Research India Publications.

Design and analysis of neural network algorithm for intelligent fire detection system

A new implementation scheme of high speed mixcolumn based on sharing the use of sbox is introduced in this paper. The single MC (mixcolumn) shares the single SB(sbox-subbyte) based on the time slot. For each time slot SB and MC performed parallelly. Earlier they use 16 individual sbox for each input. In our paper, we introduce sharing concept of sbox which can eliminate the use of 16 individual sbox and reduce the delay and cost. Normal AES uses shiftrows followed by sbox needs 128 bit for their operations that consumes large time. By eliminating shiftrows, we can increase the speed of the AES operation. LUT based sbox consumes more than 75% of power. In our paper we design the Composite field sbox which reduces the power consumption of AES architecture. Sbox is the main source of information leakage since the values are fixed one. In our paper the values of sbox are masked by using particular fixed value thus increase the system security.