DC approximation approaches for sparse optimization

We develop an exact penalty approach for feature selection in machine learning via the zero-norm $$\ell _{0}$$l0-regularization problem. Using a new result on exact penalty techniques we reformulate equivalently the original problem as a Difference of Convex (DC) functions program. This approach permits us to consider all the existing convex and nonconvex approximation approaches to treat the zero-norm in a unified view within DC programming and DCA framework. An efficient DCA scheme is investigated for the resulting DC program. The algorithm is implemented for feature selection in SVM, that requires solving one linear program at each iteration and enjoys interesting convergence properties. We perform an empirical comparison with some nonconvex approximation approaches, and show using several datasets from the UCI database/Challenging NIPS 2003 that the proposed algorithm is efficient in both feature selection and classification.

/pdf/feature-selection-in-machine-learning-an-exact-penalty-36kfolk0q6.pdf

Feature selection in machine learning: an exact penalty approach using a Difference of Convex function Algorithm

COVID-19 has affected the growth of every industry; in the meantime, an enormous amount of demand is present in the field of telecom and automobiles. In this chapter, the authors present case studies based on sales prediction for the Indian market. The analysis of the study is based on the various traditional methods like growth rate (GR), percentage growth rate (PGR), and the evolutionary techniques like genetic algorithms (GA). The data are collected for the report of telecommunication and heavy industry ministry (Republic of India). The results are used to analyze the sale of automobiles and telecommunication devices and to predict the growth at the time of the COVID-19 pandemic. The prediction is used to identify the upcoming sale and counterparts with demand.

A Study on Evolutionary Technique to Predict the Sales During COVID-19

There is need for a scheme to ensure the security and privacy of the administrator and users. Providing security is different to providing privacy, because their goals differ. Security is based on organization goals, but privacy is based on user goals. Providing security and privacy must be according to organization constraints. The most important constraints in any organization are economic issues. A useful scheme must consider all these requirements. In this paper, we present a scheme that provides security and privacy and considers various constraints. We model the problem as a multi-objective optimization problem.

/pdf/multi-objective-optimization-for-computer-security-and-22wd5nfgrp.pdf

Multi-objective Optimization for Computer Security and Privacy.

Security is the most important issue in a network system. Administrators can more easily understand threats to the network by using a model. In this paper, we present an approach for modelling a network that considers the benefits of the network as well as its limitations. In our approach, we model the system as an optimization problem, which is solved using three algorithms. As the proposed approach is stochastic, it works very efficiently in a network environment. This paper, presents a mathematical model of the system. Model provides easy comprehend of system. Presented model is based on multi-objective optimization problem. One parameter in the presented model is security and another parameter is user productivity. Security is the most important issue in a network system. Administrators can more easily understand threats to the network by using a model. In this paper, we present an approach for modelling a network that considers the benefits of the network as well as its limitations.

/pdf/an-evolutionary-multi-objective-approach-for-modelling-1lf84m0hh8.pdf

An Evolutionary Multi-objective Approach for Modelling Network Security.

Ant Colony optimization is a heuristic technique which has been applied to a number of combinatorial optimization problem and is based on the foraging behavior of the ants. Travelling Salesperson problem is a combinatorial optimization problem which requires that each city should be visited once. In this research paper we use the K means clustering technique and Enhanced Ant Colony Optimization algorithm to solve the TSP problem. We show a comparison of the traditional approach with the proposed approach. The simulated results show that the proposed algorithm is better compared to the traditional approach.

Pallavi Arora

Papers

Analysis and synthesis of enhanced ant colony optimization with the traditional ant colony optimization to solve travelling sales person problem