S. Lakshmi ¹, R. Siva Kumar ², S. Rajendran ³

Affiliations
1 Jeppiaar Engineering College, Chennai, IN
2 Barathiar University, IN
3 Department of Chemistry, R. V. S. School of Engineering and Technology, Dindigul-5, IN

Abstract

The blue green microalgae Spirulina platensis is an important source of nutrients. An important obstacle for the production of Spirulina is Rotifer. In this paper we present a method for automated identification and recognition of rotifer contamination in Spirulina using image processing techniques. Getting the expert's advice is an essential one to improve the production. Our primary objective is to provide high nutrients Spirulina free from contaminations and secondary objective is to solve the quality issue by identifying rotifer contaminations at the earliest and giving suggestions through the mobile phones. Using microscopic examination for identifying rotifier in live culture is time consuming and manmade process. There is a chance for man made error and our new technique is machine made and the errors to be minimized. Existing system need qualified persons for operation and our proposed system can be operated by anyone. We examine lab scale Spirulina culture through both existing and new techniques for recognizing the rotifer. Our proposed tool for identifying contaminations at beginning stage is easy to eradicate the contaminations from live culture. It will be helpful to biologist to take necessary actions against contaminations of the same before further multiplications of contaminations. According to the Survey of Sun food Super Food, in 2020 the expected world wide Spirulina production is about 220 thousand tons. Hence Spirulina can be a solution for solving the world nutrition problems due to its very high growth rate and high nutritional value. Our research-automatic identification of contamination in algae through the processing of mobile phone images is an easiest way for improving and evaluating the growth of Spirulina for effective cultivation of algae.

Keywords

Automatic Detection and Contamination, Image Processing, Recognition, Rotifer, Spirulina.

Full Text

   

S. Rajendran ¹, K. Balasubramanian ²

Affiliations
1 Dr. M. G. R. Educational and Research Institute, University, Chennai - 600107, Tamil Nadu, IN
2 Department of Mechanical Engineering, Kalasalingam Academy of Research and Education, Krishnankoil - 626126, Tamil Nadu, IN

Abstract

The flowshop scheduling problems are solved by heuristic methods in view of the NP-hardness. This paper proposes to explore the near optimal sequences based on heuristic algorithms with the objective to minimize Makespan. Metaheuristic based methods are used in scheduling problems where exact methods are not sufficient to provide a solution. Swarm intelligence systems are generally made up of a population of naturally existing phenomena or agents and in this paper the consideration is the foraging behavior of honey bees. Such swarm based algorithms are used to duplicate the methods of nature to conduct a search towards the near optimal solution. This paper presents an application of parallel flow line scheduling using metaheuristic method of swarm intelligence based on bee colony algorithm. The algorithm used is to minimize the Makespan, which is one of the important requirements to reduce the overall lead time in manufacturing. The results of this algorithm has been compared with the results of other comparable research algorithms and verified. Computational results show that GA based algorithm outperforms ABC in all instances of the chosen problem sets. In order to measure the efficiency of the solution methods, the execution time (sec) taken by each solution method to obtain solution to an instance of a problem was computed. The mean value of execution time over the hundred problem instances solved under various metaheuristics shows that for the small size problems the computation time is same for all the three algorithms and when the problem size increases the computation time of algorithms also increases and vary exponentially and from the experiments it is inferred that the GA algorithm is faster than ABC. The application of this algorithm discussed can be applied over a number of applications related to manufacturing in parallel assembly lines like packaging industry, manufacturing industry like paper industry, plastics injection molding industry etc.

Keywords

Bee Colony, Heuristics, Parallel Flow-Line, Scheduling

Full Text

   

Vibhor Singh ¹, Priyansh Saxena ¹, Siddharth Singh ¹, S. Rajendran ¹

Affiliations
1 Department of Information Technology, SRM University, Kattankulathur, Chennai – 603203, Tamil Nadu, IN

Abstract

Background/Objective: Customer reviews are important for various fields (e.g. Movies,Products, Services). Moviereviews plays vital role in describing its success and failure. People have now become very specific on what movies to watch and what not to watch. Hence people don't want to waste time on a movie that has bad reviews. Nowadaysonline reviews are important for personal recommendation. Methods/Statistical Analysis: There are various works done on opinion mining and text mining. This paper focuses on analyzing sentiment from semantic orientation of words that occur in a text by manually defining dictionary for positive, negative and intensifier words stored in different text file. In this method we extract data from three different sources. Opinions that has to be analyzed is preprocessed and stored in text file. The data are then compared with our bag of words in order to find the number of positive and negative sentiments in the reviews of that particular movie. To predict movie rating three machine learning algorithms have been used to create three different classifier using a trained data set. Findings: After rating prediction, the accuracy of each classifier is calculated on the data set containing predicted rating of various movies given by each classifier. Out of all three machine learning algorithms used, Naive Bayes is found to be more accurate than Decision Tree and K Nearest Neighbour algorithm. Improvements: There can be more many ways where we understand how the users write reviews as the reviews only support English language. Hence we can also bring in multiple languages, so that we do not even miss out on a single review. Most of the time,a single user may not provide review for all the movies. So, our database will resemble a sparse matrix. A prediction algorithm may be designed to mathematically guess the rating for movies that are not originally reviewed by the user.

Keywords

Machine Learning, Opinion Mining, Sentiment Analysis, Text Mining.

Full Text