In this project, we designed a novel ResNet architecture to maximize classification accuracy on the CIFAR-10 dataset while ensuring that the total number of parameters are less than 5M. We implement a random search approach for hyperparameter optimization. Our best performing architecture, TPSNet achieved a test accuracy of 94.84%.
License: MIT License
Python 96.40%Shell 3.60%
tpsnet's Introduction
Hi 👋, I'm Sudharsan Ananth
A passionate python developer👨💻, with specialities in Machine learning, Computer vision, and Robotics🤖. Outside work, I like to play with Arduino, Raspberry pi, and robots.
By the way don't forget to check out my projects.🚀