Python-Deep-Learning
Python-Deep-Learning copied to clipboard
dlab-berkeley
D-Lab's 6 hour introduction to deep learning in Python. Learn how to create and train neural networks using Tensorflow and Keras.
``` print("Train:", np.unique(y_train, return_counts=True)) print("Validation:",np.unique(y_val, return_counts=True)) print("Test:", np.unique(y_test, return_counts=True)) ``` This is a low code alternative to viewing the distributions of digits in each data set. No visual, but students...
Why do we have a triple with 28 showing up twice? It's a 28 by 28 pixel. We should add this as a comment. Training Data Shape: (5000, 28, 28)...
In notebook 3 it might be worth replacing the image augmentation section for transfer learning. I think it is a much more valuable skill that transfers (pun intended) across several...
Consider adding a couple of slides with high-level definitions of other major architecture types people might also run into (in particular RNN(LSTM) and GAN) and examples of applications where those...
In the slides, day 1 mention: 1. What evaluation metrics exist 2. Use cases (cancer detection, check processing, etc.) 3. Why we use accuracy - for simplicity's sake / for...
Both links lead to "File not found" paths. For DataHub, using the https://dlab.datahub.berkeley.edu link and signing in allowed me to run the notebook.
There may need to be some sort of documentation for participants in this workshop to know to create a conda environment using Python 3.12 or lower.
Metadata
Owner
Metadata
D-Lab's 6 hour introduction to deep learning in Python. Learn how to create and train neural networks using Tensorflow and Keras.