Browsing by Author "Varma, V."

Now showing 1 - 3 of 3

Alternate Approaches to Scattering Networks in Image Classification
(Institute of Electrical and Electronics Engineers Inc., 2023) Rao, S.; Varma, V.
Scattering networks are a special class of convolutional neural networks (CNNs) which implement a windowed scattering transform in their initial layers while learning the rest of the parameters. For classification tasks requiring little data, scattering networks beat cutting-edge deep neural networks. When given a huge dataset, their performance is comparable to end-to-end trained networks, but they're better suited for real-time applications due to their lower latency. The use of a windowed scattering transform for tasks involving image classification on the CIFAR-10 dataset is examined in this paper. We replace the 2-D Gabor filterbank in the state-of-the-art scattering network with alternate filterbanks that provide better directional separation in the frequency domain. We also develop a trainable directional filterbank that uses data-driven directional filters in its construction. The directional filters are built using the weights of a 2D convolutional operator. We demonstrate the performance of the alternate approaches in supervised classification tasks and observe that the trainable implementation outperforms the traditional scattering networks. Â© 2023 IEEE.
Maximizing Performance and Efficiency: An Algorithm Approach to Engine Sensor Optimization using Machine Learning
(Institute of Electrical and Electronics Engineers Inc., 2024) Varma, V.; Verma, K.; Mehta, H.; Gangadharan, K.V.
This paper presents an algorithmic methodology developed to reduce the number of sensors required in automotive engines by leveraging machine learning techniques. The sensor data used in this technique was obtained from a standard engine, which exhibited redundancy in data. The high cost associated with sensors and their integration into engine systems necessitates an efficient approach to optimize sensor utilization while maintaining reliable engine performance. By utilizing advanced data analysis and predictive modeling, our algorithm aims to identify redundant or non-critical sensors, enabling a streamlined and cost-effective sensor configuration. We achieved this by developing a tailored dimensionality reduction algorithm based on functional dependency theory. This approach transforms data from a high-dimensional space into a lower-dimensional space, preserving essential features of the original data and ideally approximating its intrinsic dimension. Â© 2024 IEEE.
Simulation of Indoor Localization and Navigation of Turtlebot 3 using Real Time Object Detection
(Institute of Electrical and Electronics Engineers Inc., 2021) Nandkumar, C.; Shukla, P.; Varma, V.
This paper proposes a method for indoor localization and navigation of Turtlebot 3 using Real Time Object Detection (RTOD). The robot is capable of recognizing the room it is placed inside based on the knowledge of positions of certain fixed arbitrary objects. The robot then proceeds to understand its position inside the room and is capable of moving to other locations. The robot is simulated using the ROS and Gazebo framework. The RTOD is trained to identify certain distinct objects like a rover, bowl, quadcopter and wheel based on which the robot is able to ascertain its location. Â© 2021 IEEE.