Maharaja Surajmal Institute

Delhi, India

Maharaja Surajmal Institute

Delhi, India

Time filter

Source Type

Choudhary A.,Maharaja Surajmal Institute | Rishi R.,Maharshi Dayanand University | Ahlawat S.,Maharaja Surajmal Institute of Technology
Advances in Intelligent and Soft Computing | Year: 2012

The recognition accuracy of an Optical Character Recognition (OCR) system mainly depends on the selection of feature extraction technique and the classification algorithm. This paper focuses on the recognition of handwritten digits using projection profile features. Vertical, Horizontal, Left Diagonal and Right Diagonal directions are the four different orientations that are used for abstracting features from each handwritten digit. A feed forward neural network is proposed for recognition of digits. 750 digit samples are collected from 15 writers; each writer contributed each of the 10 digits 5 times. Thus a local database containing 750 digit samples is used for training and testing of the proposed OCR system. Preprocessing of handwritten digits is also done before their classification. The combination of proposed feature extraction method along with back-propagation neural network classifier is found to be very effective as it yields excellent recognition accuracy. © 2012 Springer-Verlag GmbH Berlin Heidelberg.


Choudhary A.,Maharaja Surajmal Institute | Ahlawat S.,Maharaja Surajmal Institute of Technology | Rishi R.,Maharshi Dayanand University
Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics) | Year: 2014

The aim of this work is to judge the efficiency of Multi Layer Perceptron (MLP) and Radial Basis Function (RBF) neural network classifiers for performing the task of cursive handwritten digit recognition. Binarization features are extracted from the preprocessed handwritten digit images. The features thus obtained are used to train MLP and RBF classifiers. A detailed investigation in the proposed experiment was done and it can be summarized that when binarization features of the digit images are extracted and used for training the neural network classifiers in the recognition experiment, RBF classifier outperforms the MLP classifier. The RBF Network delivers 98.40% recognition accuracy whereas the MLP classifier delivers 96.20% accuracy for the proposed experiment of cursive handwritten digit recognition. © 2014 Springer International Publishing Switzerland.


Choudhary A.,Maharaja Surajmal Institute | Rishi R.,TITS | Ahlawat S.,Maharaja Surajmal Institute of Technology | Dhaka V.S.,IMS
2010 The 2nd International Conference on Computer and Automation Engineering, ICCAE 2010 | Year: 2010

Aim of this paper is to analyze the performance of back-propagation feed-forward algorithm using various different activation functions for the neurons of hidden and output layers. For sample creation, 250 numerals were gathered form 35 people. After binarization, these numerals were clubbed together to form training patterns for the neural network. Network was trained to learn its behavior by adjusting the connection strengths at every iteration. The conjugate gradient descent of each presented training pattern was calculated to identify the minima on the error surface for each training pattern. Experiments were performed by selecting different combinations of two activation functions out of the three activation functions 'logsig', 'tansig' and 'purelin' for the neurons of the hidden and output layers and the results revealed that the percentage recognition accuracy of the neural network was observed to be optimum when 'tansig'-'tansig' combination of activation functions was used for neurons of hidden and output layers. ©2010 IEEE.


Choudhary A.,Maharaja Surajmal Institute | Rishi R.,Maharshi Dayanand University | Ahlawat S.,Maharaja Surajmal Institute of Technology
Procedia Computer Science | Year: 2013

Characters extraction is the most critical pre-processing step for any off-line text recognition system because the characters are the smallest unit of any language script. The paper proposes an approach to segment character images from the text containing images and computer printed or handwritten words. This segmentation approach is based on a set of properties for each connected component (object) in the whole binary image of the machine printed or handwritten text containing some other images. These words which are printed along with some images are of different lengths and are printed by different cursive fonts of different sizes. This character extraction technique is applied for the segmentation of untouched characters from the machine printed or handwritten words of varying length written on a noisy background having some images etc. Very promising results are achieved which reveals the robustness of the proposed character detection and extraction technique. © 2013 The Authors. Published by Elsevier B.V.


Choudhary A.,Maharaja Surajmal Institute | Rishi R.,Technological Institute of Textile and science | Ahlawat S.,Maharaja Surajmal Institute
Communications in Computer and Information Science | Year: 2011

In this work the classification efficiency of the feed-forward neural network architecture is analyzed by using various different activation functions for the neurons of hidden and output layer and varying the number of neurons in the hidden layer. 250 numerals were gathered form 35 people to create the samples. After binarization, these numerals were clubbed together to form training patterns for the neural network. Network was trained to learn its behavior by adjusting the connection strengths at every iteration. Experiments were performed by selecting all combinations of two activation functions logsig and tansig for the neurons of the hidden and output layers and the results revealed that as the number of neurons in the hidden layer is increased, the network gets trained in small number of epochs and the percentage recognition accuracy of the neural network was observed to increase up to a certain level and then it starts decreasing when number of hidden neurons exceeds a certain level due to overfitting. © Springer-Verlag Berlin Heidelberg 2011.


Choudhary A.,Maharaja Surajmal Institute | Rishi R.,Maharshi Dayanand University
Advances in Intelligent Systems and Computing | Year: 2014

This paper is focused on evaluating the capability of MLP and RBF neural network classifier algorithms for performing handwritten character recognition task. Projection profile features for the character images are extracted and merged with the binarization features obtained after preprocessing every character image. The fused features thus obtained are used to train both the classifiers i.e. MLP and RBF Neural Networks. Simulation studies are examined extensively and the proposed fused features are found to deliver better recognition accuracy when used with RBF Network as a classifier.


Singh J.,Maharaja Surajmal Institute
Proceedings of the 2014 Conference on IT in Business, Industry and Government: An International Conference by CSI on Big Data, CSIBIG 2014 | Year: 2014

With great power of data comes great responsibility! A big data initiative should not only focus on the volume, velocity or variety of the data, but also on the best way to protect it. Security is usually an afterthought, but Elemental provides the right technology framework to get you the deep visibility and multilayer security any big data project requires. Multilevel protection of your data processing nodes means implementing security controls at the application, operating system and network level while keeping a bird's eye on the entire system using actionable intelligence to deter any malicious activity, emerging threats and vulnerabilities. Advances in Machine Learning (ML) provide new challenges and solutions to the security problems encountered in applications, technologies and theories. Machine Learning (ML) techniques have found widespread applications and implementations in security issues. Many ML techniques, approaches, algorithms, methods and tools are extensively used by security experts and researchers to achieve better results and to design robust systems. © 2014 IEEE.


Kumar M.,Maharaja Surajmal Institute
Proceedings - 2014 13th International Conference on Information Technology, ICIT 2014 | Year: 2014

The rapid growth of elderly population is a global concern and a burden on healthcare services as special health challenges appear for this segment of the population. An innovative way of patient monitoring is possible due to the recent advancements in electronics that have emerged with a number of devices which can provide continuous, real time remote healthcare monitoring to the patients even if they are freely moving around and not in hospital beds. The constant miniaturization of these electronic devices has made it possible to wear these sensors either on the clothing or body or even implanted inside the body. An unprecedented growth of smart phones and Internet technology all over the world would be a boon in this area. The recorded information, sent by these wearable sensors can be collected locally using some PDA or mobile phone. These collected medical data values may be analyzed in brief against the stored threshold values using an app on the mobile phone in real time. Collective Information may be transmitted to a centralized server periodically or immediately in case of emergency medical response required in life critical situations. The submitted medical information is used for clinical diagnosis & experts' advice and long term storage in healthcare database for future references. In this paper we discuss architecture of the Healthcare System and asses the security issues and privacy concerns while collecting patient medical data from sensors to mobile device and further submitting this data to the centralized server. The security and privacy protection of sensitive and private patient medical data is a major unsolved concern and a break into the system is possible. We also discuss the other challenges in the implementation of WBAN and provide a conclusion. © 2014 IEEE.


Pabreja K.,Birla Institute of Technology and Science | Pabreja K.,Maharaja Surajmal Institute | Datta R.K.,Gwalior Academy of Mathematical science
International Journal of Data Analysis Techniques and Strategies | Year: 2012

The multidimensional data model can be effectively utilised for analysing huge and detailed meteorological datasets forecasted by numerical weather prediction (NWP) model. The model cannot predict any weather event directly. The output products of model are interpreted by man-machine mix to infer the idiosyncratic behaviour of weather events. The mathematical tools for analysis and forecasting are able to provide forecast of weather variables only at grid-points. In this paper, the technology of dimension modelling has been adapted for analysing NWP model output datasets corresponding to sub-grid scale events viz. cloudburst, using OLAP technique. The huge datasets of weather variables available directly and derived indirectly, are mined so as to locate the patterns of cloudburst formation. K-means clustering technique has been used to generate clusters of convergence and divergence, for four real-life cases of cloudburst. It has been observed that clustering technique can help in identification of patterns conducive to formation of cloudburst. Copyright © 2012 Inderscience Enterprises Ltd.


Anand N.,Maharaja Surajmal Institute
Proceedings - 2013 International Symposium on Computational and Business Intelligence, ISCBI 2013 | Year: 2013

Web applications are taking popularity in number of ways. Monitoring the client side data allow for gathering valuable information about its behaviour. In this paper an intelligent and integrated system for user activity monitoring for both computer and internet movement is proposed. The system provides on-line and off-line monitoring and allows detecting user behaviour. On-line monitoring is carried in real time and is used to predict user actions. Off-line monitoring is carried out after user has ended his work, and is based on the analysis of statistical parameters of user behaviour. A method for the identifying the category of web sites is also presented. Our system performs clustering on the basis of URL. The URL clustering is very informative, making techniques based on it faster than that make use of text information as well. © 2013 IEEE.

Loading Maharaja Surajmal Institute collaborators
Loading Maharaja Surajmal Institute collaborators