Meeta Kumar

Bio: Meeta Kumar is an academic researcher from Symbiosis International University. The author has contributed to research in topics: Metaheuristic & Evolutionary algorithm. The author has an hindex of 3, co-authored 8 publications receiving 100 citations.

Socio-inspired Optimization Metaheuristics: A Review

Abstract: The chapter attempts to review the recent literature in the upcoming area of socio-inspired metaheuristics. These optimization methodologies are a novel subbranch of the popular Evolutionary algorithms under the class of nature-inspired algorithms for optimization. The socio-inspired class of algorithms seeks inspiration from human behavior seen during the course of the social and cultural interactions with others. A human being exhibits natural and inherent tendencies of competitive behavior, to collaborate, work together and interact socially and culturally. All such natural behaviors help an individual to learn and imbibe behaviors from other humans, resulting in them to adapt and improve their own behaviors in due course of time. This tendency observed in humans serves as a motivation for socio-inspired optimization algorithms were the agents in the optimizer algorithm work toward achieving some shared goals. This class of optimization algorithms finds their strength in the fact that individuals tend to adapt and evolve faster through interactions in their social setup than just through biological evolution based on inheritance alone. In the article, the authors introduce and summarize the existing socio-inspired algorithms, their sources of inspiration, and the basic functioning. Additionally, the review also sheds light on the limitations and the strengths of each of these socio-inspired optimizers discussed in the article. The problem domains to which these optimizers have been successfully applied to is also presented. The authors note that most of the algorithms developed in this subbranch of nature inspire methodologies in this area are new and are still evolving, thus promising scope of work in this domain.

A Reactive Energy-Alert Algorithm for MANET and Its Impact on Node Energy Consumption

Abstract: Mobile Ad-hoc networks (MANETs) are self organized networks whose nodes are free to move randomly while being able to communicate with each other without the help of an existing network infrastructure. In MANET, the routing protocols have to route the packets depending on the MANET constraints such as battery power in addition to the shortest path. The limited battery supply to mobile node in MANET requires that the routing protocols utilize power efficiently and thus maximize the network life time. The energy aware deterioration in ad hoc networks is a very important aspect of the overall management of ad hoc networks. In this paper, the focus is on the reactive power-alert technique for communication between ad hoc network nodes by continuously alerting their energy status to neighbor nodes. Here the concentration is on reducing the energy consumption by proposing optimal path selection method. In this scheme a threshold value is set on the energy consumed by mobile nodes in ad-hoc network. If the energy level of any node/s in the network reaches a threshold level then such nodes are made inactive and inform other nodes not to establish connections with it in this sleep state. In this paper, experimental results and a comparative analysis are presented based on the use of this threshold. The result shows significant improvement in the throughput and routing load which in turn increases the lifetime of the network.

Techniques for Traffic Sign Classification using Machine Learning- A Survey

Abstract: The Road Sign Recognition is a field of applied computer vision research concerned with the automatically detection and classification of traffic signs in traffic scene images. The aim of the present paper is to study various classification techniques that can be used to construct a system that recognizes road signs in images. The primary objective is to develop an algorithm, which will identify various types of road signs from static digital images in a reasonable time frame. In the current paper, we will study various learning systems that are based on prior knowledge for classification. A road sign recognition system faces a classical problem of pattern recognition, meaning classifying between different road signs. On top of that, the location of the road sign in the picture is unknown. Once these obstacles are overcome, such system could be integrated in a Smart Driver System. A variety of MATLAB Image Processing Toolbox commands can be used to determine if a road sign is present in current image. Neural network or other classification techniques can be applied in order to classify the road signs. Finally, the relevant sign is highlighted and output to the screen. Some of the examples where this technique is used is Ford Focus, BMW-7 series, Mercedes-Benz E-class, Volkswagen, etc. car. But still, identification of road signs invariantly with respect to various natural viewing conditions still remains a challenging task. This is so because color information is affected by varying illumination; Road signs are frequently occluded partially by other vehicles; many objects are present in traffic scenes which make the sign detection hard; road signs exist in hundreds of variants often different from legally defined standard; the algorithms must be suitable for the real-time implementation The study consists of three parts: road sign detection, classification and GUI. The actual imaging processing including color space conversion, color-thresholding is applied to determine if a road sign is present. If present, the sign will be resized and classified. The data which obtained by neural network training is used to classify the road signs. GUI will be created for user to interactive with the algorithm. The system will have the potential to help in improving road safety.

A Hybridized Data Clustering for Breast Cancer Prognosis and Risk Exposure Using Fuzzy C-means and Cohort Intelligence

Abstract: Breast cancer is the most prevailing type of cancer responsible for a large number of deaths every year. However, at the same time, this is largely a curable type of cancer if identified at initial stages. With major advances in research in the areas of image processing, data mining and clustering and machine learning, a more precise prognosis and prediction of breast cancer are possible at earlier stages. A fuzzy clustering model is a popular model used across various researches in image processing to predict the malignancy of breast tumor. The partitional clustering method finds its strength in its fuzzy partitioning such that a data point may belong to different classes with varying degrees of membership (ranging between 0 and 1), which is less rigid as compared to an older and still popular k-means clustering algorithm. The current article attempts to hybridize the fuzzy C-means with the cohort intelligence (CI) algorithm to optimize cluster formation. CI is a robust optimization metaheuristic belonging to the class of socio-inspired optimizers (Kumar M, Kulkarni A Socio-cultural inspired metaheuristics, pp 1–28, Springer International Publishing, 2019 [22]), motivated from self-adapting behavior of candidates in a cohort or a group. CI is typically characterized by its simple algorithmic nature, robust structure and a faster convergence rate, hence gaining popularity. This novel hybridized data clustering algorithm fuzzy-CI imitates the soft clustering and communal learning attitude of clusters and candidates. The hybridized method of fuzzy-CI is validated by testing it on the Breast Cancer Wisconsin (Diagnostic) Dataset. The results validate that the hybridized version exhibits better cluster formation in comparison with the non-hybridized version.

Political Optimizer: A novel socio-inspired meta-heuristic for global optimization

Abstract: This paper proposes a novel global optimization algorithm called Political Optimizer (PO), inspired by the multi-phased process of politics. PO is the mathematical mapping of all the major phases of politics such as constituency allocation, party switching, election campaign, inter-party election, and parliamentary affairs. The proposed algorithm assigns each solution a dual role by logically dividing the population into political parties and constituencies, which facilitates each candidate to update its position with respect to the party leader and the constituency winner. Moreover, a novel position updating strategy called recent past-based position updating strategy (RPPUS) is introduced, which is the mathematical modeling of the learning behaviors of the politicians from the previous election. The proposed algorithm is benchmarked with 50 unimodal, multimodal, and fixed dimensional functions against 15 state of the art algorithms. We show through experiments that PO has an excellent convergence speed with good exploration capability in early iterations. Root cause of such behavior of PO is incorporation of RPPUS and logical division of the population to assign dual role to each candidate solution. Using Wilcoxon rank-sum test, PO demonstrates statistically significant performance over the other algorithms. The results show that PO outperforms all other algorithms, and consistency in performance on such a comprehensive suite of benchmark functions proves the versatility of the algorithm. Furthermore, experiments demonstrate that PO is invariant to function shifting and performs consistently in very high dimensional search spaces. Finally, the applicability on real-world applications is demonstrated by efficiently solving four engineering optimization problems.

Heap-based optimizer inspired by corporate rank hierarchy for global optimization

Abstract: In an organization, a group of people working for a common goal may not achieve their goal unless they organize themselves in a hierarchy called Corporate Rank Hierarchy (CRH). This principle motivates us to map the concept of CRH to propose a new algorithm for optimization that logically arranges the search agents in a hierarchy based on their fitness. The proposed algorithm is named as heap-based optimizer (HBO) because it utilizes the heap data structure to map the concept of CRH. The mathematical model of HBO is built on three pillars: the interaction between the subordinates and their immediate boss, the interaction between the colleagues, and self-contribution of the employees. The proposed algorithm is benchmarked with 97 diverse test functions including 29 CEC-BC-2017 functions with very challenging landscapes against 7 highly-cited optimization algorithms including the winner of CEC-BC-2017 (EBO-CMAR). In the first two experiments, the exploitative and explorative behavior of HBO is evaluated by using 24 unimodal and 44 multimodal functions, respectively. It is shown through experiments and Friedman mean rank test that HBO outperforms and secures 1 st rank. In the third experiment, we use 29 CEC-BC-2017 benchmark functions. According to Friedman mean rank test HBO attains 2 nd position after EBO-CMAR; however, the difference in ranks of HBO and EBO-CMAR is shown to be statistically insignificant by using Bonferroni method based multiple comparison test. Moreover, it is shown through the Friedman test that the overall rank of HBO is 1 st for all 97 benchmarks. In the fourth and the last experiment, the applicability on real-world problems is demonstrated by solving 3 constrained mechanical engineering optimization problems. The performance is shown to be superior or equivalent to the other algorithms, which have been used in the literature. The source code of HBO is publicly available at https://github.com/qamar-askari/HBO.