a particle swarm optimized featu
Deep Learning Projects , Python Projects

A Particle Swarm Optimized Feature selection and Learning for Fault Classification in Web-Apps

Web attacks are considered to be serious attack as the users of website may not get proper services if web application got an attack. The proposed work aim to classify web attacks as malicious and normal. There are 2 datasets used in proposed work are

  1. HTTP dataset CSIS 2010
  2. FWAF dataset created by FSECURITY

Deep neural network DNN algorithm is used for this study.

The efficient feature selection algorithm applied in this proposed work is PSO (Particle Swarm Optimization). The hybrid model of web attack detection is implemented.

The model outperformed with highest accuracy of 99.5% using FWAF dataset

The dataset is classified as binary classification model using DNN algorithm and source code is available in Python. Keras and Tensorflow are used for deploying and building a deep learning model for web attack detection. The following output image shows the accuracy of the DNN algorithm.

Particle Swarm Optimized Feature selection and Learning for Fault Classification

Further, the working model of the proposed work is available in following demo

A Particle Swarm Optimized Feature selection and Learning for Fault Classification in Web-Apps

Such kind of intrusion detection and attack detection and classification is available in the following link and provided more information

Intrusion detection and classification

privacy preserving cloud and dat
Cloud Computing Project , Java Projects

Privacy-preserving cloud and data dissemination scheme for vehicular cloud

Cloud and data dissemination scheme for vehicular cloud

Cloud and data dissemination scheme for Vehicular cloud (VC) is a part of cloud computing to vehicles that participate in vehicular ad hoc networks, with the goal of providing computing and storage services to vehicles at a minimal price, trying to improve traffic safety and efficiency, it guarantees real-time services.

Because of the highly dynamic nature of VC, it is difficult to efficiently form a securely VC network and to securely deliver messages to the VC without potentially invading cloud users’ privacy.

The proposed work provides a concrete secure and privacy-preserving communication scheme for VC formation and data dissemination.

The proposed scheme enables a group of vehicles in close proximity to form a vehicular cloud with security authentication and dynamically.

Cloud and data dissemination scheme for vehicular cloud

Modules Used in this Project

Vehicle node deployment
Trusted Authority
Data security
Receiver model

Proposed System

Proposed system is a secure and privacy-preserving communication scheme for VC establishment and data dissemination in VC.
Specifically, this scheme allows a group of vehicles located close to each other in the VANET to form a VC securely, anonymously and dynamically.
This allows the vehicles’ resources to be integrated and shared anonymously and securely, using pseudonyms and dynamic identity based authenticated asymmetric group key agreement
In the scheme, the public key of a vehicle is its (one-time use) pseudonym. Thus, no certificate is required to bind the public key with the vehicle, and this simplifies certificate management.
The private key corresponding to the pseudonym is generated by a trusted authority based on the pseudonym of the vehicle. Using the pseudonym and the corresponding private key, each vehicle in the group can anonymously run the security protocol to form a secure VC in one round

Advantages

Cloud users to send encrypted messages to the VC, and an eavesdropper is not able to learn the exchanged messages and the identities of the participants.
Privacy and Security of proposed system is ensured.

Java Project Source code implementation for Secure Vehicular Cloud

Similar Project in Vehicular cloud also implemented in NS2 simulation project, listed below

Identity-Based Privacy-Preserving Authentication Scheme for VANETs

Improving Safety on Drivers for Vehicular Ad Hoc Networks

IEEE Project,Ns2 Project,Python Project
Machine Learning Projects , Python Projects

Citywide Bike Usage Prediction and Forecast

Citywide Bike Usage Prediction and Forecast

City wide bike usage prediction and forecast using machine learning is the aim of the project. Bike sharing demand is increasing nowadays. In this work, dataset from Capital Bikeshare system, Washington DC, USA is used.

Bike demand may affect or change due to various parameters such as weather, holiday or working day, temperature, humidity and more. These features are considered in this dataset.

Project methodology

  1. Data Pre-processing
  2. Data Visualization
  3. Training using Lasso
  4. and Ridge regression

As the dataset considered is bike demand, which is a variable parameter according to each condition and attributes. Regression is applied on this work.

Experimental analysis showed accuracy of bike demand prediction is around 62%.

Data science is useful when the dataset is having multiple influencing parameters. Here we have visualized data and done Explanatory data analysis as shown below

Citywide Bike Usage Prediction and Forecast

Project implementation demo is given below, source code for bike usage demand prediction and forecast is written in Python

Citywide Bike Usage Prediction and Forecast | ieee project demo

Similarly, forecasting is an interesting subject, it can be applied in given work check mode links

Stock Price prediction and Forecast

Air-pollution prediction through feature selection and deep learning
Deep Learning Projects , Python Projects

Air pollution prediction through feature selection and deep learning

Air pollution detection

Air pollution detection and monitoring is one the important area as these days air pollution is increasing through various parameters such as vehicles, industries etc. Air pollution prediction is made easier through artificial intelligence. This proposed work used feature selection algorithm for choosing important attributes from dataset.

The dataset from Central Air pollution board is used for this study. It has many attributes and the labels are considered as multi class classification as given below

  1. Residential
  2. Industrial
  3. RIRUO
  4. Sensitive

Regression algorithm and Neural Network algorithm is used for air pollution detection of above class type. Logistic regression gives best accuracy of around 63% for air pollution prediction.

Air Pollution Detection

The following link shows you the implementation process and methodology used. Source code used is Python and dataset downloaded from Central pollution board of India.

Air-pollution prediction through feature selection and deep learning | ieee project demo

More Machine Learning projects are available, some listed below

Regional Detection of Traffic Congestion Using in a Large-Scale Surveillance System via Deep Learning