Hrudaya Kumar Tripathy

A hybrid improved PSO-DV algorithm for multi-robot path planning in a clutter environment

This paper proposed a novel approach to determine the optimal trajectory of the path for multi-robots in a clutter environment using hybridization of improved particle swarm optimization (IPSO) with differentially perturbed velocity (DV) algorithm. The objective of the algorithm is to minimize the maximum path length that corresponds to minimize the arrival time of all the robots to their respective destination in the environment. The robots on the team make independent decisions, coordinate, and cooperate with each other to determine the next positions from their current position in the world map using proposed hybrid IPSO-DV. The proposed scheme adjusts the velocity of the robots by incorporating a vector differential operator inherited from Differential Evolution (DE) in IPSO. Finally the analytical and experimental results of the multi-robot path planning have been compared to those obtained by IPSO-DV, IPSO, DE in a similar environment. Simulation and khepera environment results are compared with those obtained by IPSO-DV to ensure the integrity of the algorithm. The results obtained from Simulation as well as Khepera environment reveal that, the proposed IPSO-DV performs better than IPSO and DE with respect to optimal trajectory path length and arrival time.

Enhancing the capabilities of Student Result Prediction System

In the age of information and communication technology, technology is being used in every domain. Education is the integral part of our society consisting of the teaching, learning and evaluation process. Information and communication technologies are being used for the purpose of e --learning, measuring students learning, course design, student performance evaluation e.t.c. Using machine learning techniques performance of the students has been studied and useful results have been derived. Predicting the performance of a student accurately in the upcoming exam is of extreme significance. Every machine learning tool heavily depends upon the input data. Studying and implementing the elaborate feature set for students has improved the accuracy of the prediction system. Further use of preprocessing techniques along with classification algorithms has significantly improved the results of the prediction system.

An implementation of Feature ranking using Machine learning techniques for Diabetes disease prediction

Disease diagnosis is an application area where machine learning tools are providing successful results. Diabetes disease is one of the crucial factors of death all over the world. The availability of huge amounts of medical data leads to the need for powerful learning tools to help medical experts to diagnose diabetes disease. Machine learning methods are helpful in the diagnosis of diabetes disease, showing a reasonable level of efficiency. But these data are redundant and are noisy in nature which negatively affects the process of observing knowledge and useful pattern. Machine learning techniques have attracted a big attention to researchers to turn such data into useful knowledge. Further relevant data can be extracted from huge records using filter based feature selection methods. In our study, a comparative analysis is drawn between four different filter based feature selection methods (Chisquare method, Information gain method, Cluster Variation method and Correlation method) based on Diabetes disease. Three classifiers (RBF, IBK and JRip) were implemented to estimate the performance of the algorithms. The study revealed that filter based feature selection methods enhance the performance of learning algorithms in effective prediction and diagnosis of diabetes disease.

Unalike methodologies of feature extraction & feature matching in Speech Recognition

In this present scenario, the application of speech science has a vital role to produce the biometric applications. After so many research and improvement of Automatic Speech Recognition, accuracy of speech recognition is one of the challenging task. Various feature extraction is one of the Linear Predictive Coding, cepstral analysis, Local Discriminant Base, Restricted Boltzmann Machines have been discussed since past days. Similarly, a lot of debates have been arranged among the researchers for feature matching. Some of them are Hidden Markov Model (HMM), Dynamic time warping (DTW), Deep Belief Network.This paper is a clear reflection of automatic speech recognition. It describes various feature extraction and matching and focuses on analytical study based on performance metrics like Word Error Rate (WER) and accuracy of these techniques.

Machine Learning on Big Data: A Developmental Approach on Societal Applications

Machine Learning (ML) is a potential tool that can be used to make predictions on the future based on the past history data. It constructs a model from input examples to make data-driven predictions or decisions. The growing concept “Big Data” need to be brought a great deal accomplishment in the field from claiming data science. It gives data quantifiability in a variety of ways that endow into data science. ML techniques have made huge societal effects in extensive varieties of applications. Effective and interactive ML relies on the design of novel interactive and collaborative techniques based on an understanding of end-user capabilities, behaviors, and necessities. ML could additionally make utilized within conjunction for enormous information to build effective predictive frameworks or to solve complex data analytic societal problems. In this chapter, we concentrate on the most recent progress over researches with respect to machine learning for big data analytic and different techniques in the context of modern computing environments for various societal applications. Specifically, our aim is to investigate opportunities and challenges of ML on big data and how it affects the society. The chapter covers discussion on ML in Big Data in specific societal areas.

A neuro-genetic model to predict hepatitis disease risk

In the present scenario, large quantity of data is generated in the field of medicine. This data contain valuable information which can be utilized in decision making. Machine learning is an active area which may be useful to healthcare experts. Hepatitis disease is a common disease in the world, which may cause damage to hepatocytes. Machine learning techniques can be implemented to reduce the risk of Hepatitis. Our study has demonstrated an intelligent hybrid system for the efficient risk prediction of Hepatitis disease. We developed an intelligent combination of Genetic search algorithm and Multilayer Perceptron technique named MLP-GS. Our proposed system model was analyzed and computed with the help of several performance parameters like Accuracy, Root Mean-Squared Error, Precision, Recall and F-Measure. It was observed that MLP-GS model performs better on Hepatitis data.

Low Complexicity Graph Based Navigation and Path Finding of Mobile Robot Using BFS

Path-finding is a fundamental problem, in mobile robotics which involves finding an optimal collision-free path from the source node to the destination node. Before the robot traces out the required path, the first step to be carried out is, exploring the environment. Localization plays a major role in this case. This paper adopts an RFID (Radio Frequency Identification) based localization technique. A set of RFID(IC) tags arranged in a grid structure in an equidistant manner are used for the purpose of tracing the current co-ordinate/location of the robot. After exploring the environment one virtual map is generated which contains the location of source, destination, obstacles and landmarks. From the map one graph is generated which is composed of a set of vertices that indicates the cells of the grid and a set of edges which indicates the free path in the environment reachable from the source. After analyzing different searching algorithms we found Breadth First Search (BFS) algorithm to be more effective, because it finds the shortest path from the source node to each node in the graph. We improved the BFS algorithm so that the optimal collision free path from source to destination node is generated. We have implemented our techniques in a simulated environment. We have used MATLAB and JAVA for the simulation purpose. Finally we have demonstrated the result of the implementation through examples.

A Modified Real Time A* Algorithm and Its Performance Analysis for Improved Path Planning of Mobile Robot

This paper proposed an online path planning of mobile robot in a grid-map environment using modified real time A* algorithm. This algorithm has implemented in simulated and Khepera-II environment and find the optimized path from an initial predefine position to a predefine target position by avoiding the obstacles in its trajectory of path. The path finding strategy is designed in a grid-map and cluttered environment with static and dynamic obstacles with quadrant concept. The optimization the path is found using this algorithm as the goal is present in any of the four quadrant and restricted the movement of the robot to only one quadrant. Robot will plan an optimal path by avoiding obstructions in its way and minimizing time, energy, and distance as the cost, but the original A* algorithm find the shortest path not optimized. Finally, it is compared with other heuristic algorithms.