Security Assessment of Artificial Intelligence in Connected Systems
Motion Estimation and Predication for the Autonomous Vehicles using Computer Vision
One of the major problem associated with the autonomous vehicle or mobile robots is to operate it within the unpredicted complex environment. This scenario lead to the question of perfectly predicting the future motion direction of pedestrians and surrounding vehicles in a road for the autonomous robots, so that source vehicle can move with safety in the busy area. The goal of this research is to find an optimal solution using deep-learning to estimate and predict the future direction of other object with higher degree of accuracy use the feed from the camera installed on the autonomous vehicles.
Cloud Data center workload prediction using deep learning based approaches
Task scheduling and load balancing is the key innovation of cloud computing. However, for these tasks accurately predicting the future resource load is very necessary. Furthermore, forecasting resource usage depends on the past resources as well as the current usage of the resource, which defined workload prediction as the time series problem. Although several technique have already been there for solving time series problem including, deep learning models(LSTM, GRU, RNN), Statistical model(ES, ARIMA, AR), Or some hybrid of RNN and Statistical model (GRU-ES), but the the performance is dropped for longer time period and sudden peak load request.
Thus, It is necessary to find an optimized model, which can accurately and precisely forecast the workload, with minimal computation. In this research, we go through different models and try to optimized them by extensive tuning of hyperparameters in-order to accurately predict the sudden change along with normal changes. From this research we concluded that perfectly optimized hybridization of deep learning methods works outstanding than the individual RNN based model or other statistical model.
Fingerprinting AI/Ml models running in federated learning systems via wireless traffic
Anamorphic Adversarial attack on connected autonomous system
Real-time Gesture based Home automation system using Raspberry Pi
Stock Market Price Forecasting using parallel RNN-based Network
Object detection and classification of road traffic in different environments
Tool review on Wireshark
SpO2 and Blood Pulse detection and heart disease prediction system using Machine learning
Home Security system using IoT and finger-print-based sensor
Mobile app and website-based home automation system
Arduino-based maze solving line following robot
Line following robot with obstacle avoidance feature
Dynamic website for online bus ticketing system