Welcome to the Hyperbolic Representation Learning for Computer Vision tutorial, ECCV2022!
This website provides you access to the content that we will use for the tutorial session at the ECCV 2022, “Hyperbolic Representation Learning for Computer Vision”.
Learning in computer vision is all about deep networks and such networks operate on Euclidean manifolds by design. But is Euclidean geometry the best choice for deep learning or simply a practical option? Recent literature in machine learning and computer vision has shown that hyperbolic geometry provides a strong alternative, with an improved ability to embed hierarchies, graphs, text, images, and videos.
In light of recent advancements in hyperbolic representation learning for computer vision, this tutorial seeks to advocate hyperbolic geometry and its strong potential for computer vision to a broader audience. The tutorials provides a theoretical and practical starting point for the field. At the conference, we will provide an easy-going introduction to hyperbolic geometry for non-mathematicians, where we focus on intuition and high-level understanding. We then outline the current state of hyperbolic geometry for vision from supervised and unsupervised perspectives. At the end, we dive into open research problems and future potential for hyperbolic geometry and visual understanding.
Unique for this tutorial is that we do not stop at a theoretical foundation. The tutorial website will also host a series of notebook-style code snippets with foundational works on hyperbolic geometry, to get a better understanding of its workings and lower the barrier to start your dive into this exciting new research direction in computer vision.
For any remaining questions regarding the notebooks, please contact us at m.ghadimiatigh@uva.nl.
How to run the notebooks
On this website, you will find the notebooks exported into a HTML format so that you can read them from whatever device you prefer.
There are three main ways of running the notebooks we recommend:
Locally on GPU: If you have a laptop with a build-in NVIDIA GPU, we recommend that you run the notebooks on your own machine. All notebooks are stored on the github repository that also builds this website. You can find them here: to be updated.All notebooks require access to a GPU to keep the training times in a reasonable range. Nonetheless, if you prefer, you can code and test most of your code on a CPU-only system, i.e. your own laptop, and once your code is tested and ready, use one of the remaining options to train the model. To ensure that you have all the right python packages installed, we provide a conda environment in the same repository.
Google Colab: If you do not have access to a GPU on your local machine, you can make use of Google Colab. Google Colab provides you access to GPUs for free, and you can activate the GPU support by
Runtime -> Change runtime type -> Hardware accelerator: GPU. Each notebook on this documentation website has a badge with a link to directly open it on Google Colab. It is highly recommend to copy the notebook to your own Google Drive before starting, since when closing the session, changes might be lost if you don’t save it to your local computer or have copied the notebook to your Google Drive beforehand. In addition, note that for free account, Google Colab is limited to one session at a time, and each session has a time limit.Compute cluster: If you have access to a compute cluster, we recommend to using it to do your final trainings. Depending on your preference, you can run the tutorials either locally or on Google Colab. Once your notebook is ready, you can first convert the notebooks to a script using
jupyter nbconvert --to script ...ipynb, and then start a job on the cluster for running the script. A few advices when running on clusters:Disable the tqdm statements in the notebook. Otherwise your slurm output file might overflow and be several MB large.
Comment out the matplotlib plotting statements, or change
plt.show()toplt.savefig(...).