Distilling Knowledge via Intermediate Classifiers
This repository is for numerical experiments with convolutional neural networks, in particular for testing initialization procedures. The training data is the CIFAR10 dataset and the CNN-architecture is the FitNet-1 from ROMERO, Adriana, et al. Fitnets: Hints for thin deep nets. arXiv preprint arXiv:1412.6550, 2014.
Summer research June 2020- September 2020. This repository contains code for the model and machine learning for growth control.
Reimplementation of my undergraduate solar power prediction project in Python, addressing limitations in the original evaluation and comparing ANN performance across scikit-learn, TensorFlow, and PyTorch.
To associate your repository with the fitnet topic, visit your repo's landing page and select "manage topics."