Post-grad image and deep learning coursework, training CNNs with NVIDIA's DIGITS platform. Achieved 90%
Go to file
2021-04-10 16:56:01 +01:00
cars added 256 node FCs, confusion matrix notebook 2021-04-10 16:56:01 +01:00
cifar100 added cifar/cars manifests, added model templates 2021-03-31 09:15:11 +01:00
models added dense layer variations 2021-04-10 12:20:26 +01:00
report added 256 node FCs, confusion matrix notebook 2021-04-10 16:56:01 +01:00
.gitattributes step down results, set up for exp and sigmoid decay 2021-04-07 22:40:14 +01:00
.gitignore exponential and sigmoid results 2021-04-09 13:04:40 +01:00
pyproject.toml added 256 node FCs, confusion matrix notebook 2021-04-10 16:56:01 +01:00
README.md added initial results for fixed learning rate investigations 2021-04-05 18:14:17 +01:00
split.py added dataset split experiments 2021-04-02 14:43:36 +01:00

Digits CNN Coursework

Coursework evaluating convolutional neural networks for image classification using the DIGITS platform

80/10/10 split

90/5/5 split