Neuromorphic computing presents unique challenges when it comes to data handling and preprocessing. Unlike traditional computer vision datasets with static images, neuromorphic systems generate dynamic, event-based data streams that require specialized tools for efficient loading, transformation, and analysis.
This talk introduces Tonic, a Python package designed to bridge this gap by providing a PyTorch-compatible interface for neuromorphic datasets. Tonic enables researchers to work with event-based data using familiar PyTorch/Jax/Tensorflow patterns while offering specialized transformations and optimizations for neuromorphic applications.
The presentation will cover Tonic’s architecture, its integration with the broader PyTorch ecosystem, and practical examples of how it simplifies working with neuromorphic datasets. We’ll explore how Tonic handles various data formats, provides efficient caching mechanisms, and enables seamless batch processing of event streams.
Contents of the talk:
- Introduction to neuromorphic data challenges and requirements
- Overview of Tonic’s PyTorch-compatible design philosophy
- Dataset loading and transformation workflows
- Community contributions and future roadmap
- Q&A session
Date: September 29, 2024
Time: 8:00 PM CEST
Host: Gregor Lenz
Learn more about Tonic at the Tonic software page.
Upcoming Workshops
About the Author
Gregor Lenz
Inspired? Share your work.
Share your expertise with the community by speaking at a workshop, student talk, or hacking hour. It’s a great way to get feedback and help others learn.
