Jupyter Notebooks are a very effective tool for interactive data exploration and visualisation.
I use dark styles for both GitHub and Jupyter Notebook. To see the content appropriately install:
- Jupyter Notebook dark theme;
- GitHub dark theme and comment out the
invert #fff to #181818code block.
These notebooks are somehow of personal usage. I used them for data exploration, getting familiar with PyTorch model graphs, and generate the paper figures and website animations. They all run, on my system, with the data generated from the experiments. I'm releasing them here for reference only, and you are welcome to browse them, even though you might not be able to execute them.
display_loss.ipynb: display train and validation losses for every experiment;figures_generator.ipynb: generates paper figures and website animations;frequency_analysis.ipynb: explor video temporal-frequency components;get_all_embeddings.ipynb: display ResNet18 embeddings and probability vs. time;get_data_stats.ipynb: show stats on the e-VDS35 data set;network_bisection.ipynb: mucking around with PyTorch model graph;salient_regions.ipynb: implements and compute salient regions from bojarski2017explaining;stability_analysis.ipynb: mucking around with ResNet18 and videos;verify_resnet.ipynb: checks whether I'm capable to use PyTorch's ResNet18;plot_conf.py:matplotlibshared configuration.