High Impact Factor : 4.396 icon | Submit Manuscript Online icon |

Phishing Detection using Multi-Layer Perceptron And Comparison of Accuracy with Various Neural Network Techniques

Author(s):

Akshay H. Thotwe , College of Engineering Pune; Dr. S. B. Mane , CoEP ,Pune

Keywords:

Phishing Detection, Artificial Neural Network, Multi-Layer Perceptron, Sklearn, Backpropogation, Feed-Forward Neural Network, Voted Perceptron, Weka Tool, Machine Learning

Abstract

In this modern world number of users connected to internet are increasing rapidly with respect to number of attacks on those user to steal an information. The user data is compromised by using various hacking method implemented by hacker on system. In this paper, we will study Phishing Detection using Multi-Layer Perceptron and make comparison of accuracy obtained by using various solving and with respect to activation function. In this model, we have built Neural Network with the help of sklearn. The Dataset used to train the Neural Network has been taken from UCI machine Learning Repository for Phishing web set dataset. The Dataset contains both phishing and non-phishing instances. The number of Phishing and Non-phishing are 4898 and 6157 instances. Instead of feature selection we have extracted those feature that are important to decide legitimacy of the website. We also considered the time taken to build the neural model. After classifying the instances into phishing and legitimate, we got high accuracy. Also we have used Sklearn, it has different solving and activation technique that are defined in Multi-Layer Perceptron. We are classifying solving methods based on accuracy. Also, we have used weka tool to check the accuracy of all neural network algorithm defined in it. In this, we have classified new instances into phishing and non-phishing by using multi-layer perceptron then compared its accuracy with the various neural network techniques.

Other Details

Paper ID: IJSRDV6I30798
Published in: Volume : 6, Issue : 3
Publication Date: 01/06/2018
Page(s): 1674-1678

Article Preview

Download Article