Information Technology

This paper will discuss about Machine Learning technology, its algorithms and its applications. In the new era of automations, business intelligence and the growing datacenters capacity, machine learning is a key component. Through literature review this paper demonstrates this new field and its characteristics. Different algorithms exist for various types of works, making it possible for average business to use the one that suits its needs the best. Machine Learning has come to stay for many years, providing added value to data and predictive mechanisms that will make our world a better place.

In many modern computer applications such as Market Analysis, Critical Care, Speech Recognition, Physical Plant Monitoring, Sleep Stage Classification, Biological Population Tracking, data is captured over the course of time, constituting a Time-Series. Time-Series data often contain temporal information dependencies that cause two otherwise identical points of time to belong to different classes or predict different behavior. This inherent characteristic increases the difficulty of processing such data. Deep Machine Learning (DML) techniques possess the inherent ability for analyzing and making predictions about such data. By its nature, DML requires extensive provision of resources key amongst which is the model computation time. Several optimization algorithms have been invented in the recent past and compare differently in terms of their resource needs. The most popular class of optimization algorithms is based on the classical stochastic gradient descent (SGD) algorithm due to its ability to converge within reasonable time bounds. This paper is part of a larger project investigating optimization procedures for deep learning tasks based on the SGD. Specifically, we report on the comparative performance capabilities of the most popular SGD based algorithms for task of Time Series prediction namely. From our analysis of the six of these algorithms, we noted that ADAMAX is most appropriate for online learning while RMSPROP is the least affected by over-fitting for long training cycles.

In last few decades computer scientist has started taking interest in the quantum computing in order to resolve some of the problems in the recognizable classical computers. Relevant efforts have been put into the practice for two instructions: the basic assumptions behind the quantum mechanics has a residential affect, on the other hand a series of project has been used to show the series of implementation of all these algorithms. This will move the quantum from area of assumptions’ to the actual level. The relationship between these two areas is generally the feasible algorithms for quantum, which settle down their model of quantum circuit.

This paper has presented a review on different image segmentation techniques. The image segmentation; segments a given image into separate regions and objects. It is widely used in various vision applications like face detections, motion detection etc. The overall objective of this paper is to review various techniques of image segmentation. The short comings of image segmentation techniques will also be evaluated. This paper ends up with the analytical solution to overcome the problems of the existing techniques.

Handwritten digit recognition is challenging problem researchers had been research into this area for so long especially in the recent years. There are many fields concern with numbers e.g. checks in banks or recognizing numbers in car plates, the subject of digit recognition appears. The main objective is to recognize digits in different applications. E.g. different users had their own handwriting styles where the main challenge falls to let computer system understand these different styles and recognize them as standard writing. Neural network can be proving as an effective tool to solve such problem. Neural network uses various algorithms for learning and classification; such as Back Propagation Algorithm, feed forward etc.

This paper presents the concept of Responsive Web Design and Development for Web Applications meant to be present on screens of different resolution like desktop, tablets and mobile phones. It brings into light some of the features of Web Application and Native Applications and how the difference between them is narrowing down. A brief look into the traditional form of web designing is also mentioned in order to understand its drawbacks. It also focuses on the concepts of Media Queries and Twitter Bootstrap, a new front end framework. It overall brings into limelight the new perspective of Web Development so as to make it more flexible and efficient.

This paper provides a performance comparison between four of the most common symmetric key cryptography algorithms: AES(Rijndael), Blowfish, Twofish and Serpent. The comparison has been conducted by running four encryption algorithms to process different sizes of data blocks with any extension of files to evaluate the algorithm's encryption/decryption speed. Simulation has been conducted using java language

An ad-hoc network is a collection of mobile nodes, which communicate over radio. These networks have an important advantage; they do not require any existing infrastructure or central administration. Therefore, mobile ad-hoc networks are suitable for temporary communication links. This flexibility, however, comes at a price: communication is difficult to organize due to frequent topology changes. Routing in such networks can be viewed as a distributed optimization problem. A new class of algorithms, inspired by swarm intelligence, is currently being developed that can potentially solve numerous problems of modern communications networks. These algorithms rely on the interaction of a multitude of simultaneously interacting agents. A survey of few such algorithms for ad hoc networks is presented here.

This research work proposes a novel technique for Automatic face recognition (AFR) system using cascaded structures and clustering network. Human beings at a very early age are capable of recognizing the varying facial features, due to the Human Visual System (HVS). But it’s difficult to depict the human visual system using computer vision system. The basic idea used in this proposed work is to use divide and conquer method, where we design a particular approach for each processing stage and then embedding the entire strategy for AFR system. In this proposed work, two important factors namely cost efficiency and applied technology for varying characteristics of input image are considered respectively, irrespective of the traditional factors such as accuracy, retrieving rate etc. For facial detection, a heterogeneous cascaded detector based on various features is designed to increase the processing capability and detecting efficiency respectively. For facial feature extraction, sparse graph, component based direct fitting and component based texture fitting methods are used to extract the various features at different orientations.

In this paper we investigate the problem of separation of human voice from a mixture of voice and different music instruments. The human voice may be a part of singing voice in a song or it may be a part of some news broadcasted by a channel and it contains background music. The final outcome of this work is a file containing only vocals. In our approach we consider stereo audio for separation. We process the signal in time frequency domain. In our method of blind source separation we processed the input stereo audio file in the form of frames, windowed them and then applied discrete Fourier transform (DFT) on signal. Then the signal is masked for de-mixing purpose using time frequency filters and non-zero DFT coefficients that are estimated as a part vocals are selected and signal is reconstructed by overlap add (OLA) method to get the final output signal containing only vocals.