A Conversation on AI Codecs with Jona Ballé
Date:
AI codecs are replacing traditional block-based compression algorithms developed by standards bodies such as MPEG and ITU with fully AI-based encoder-decoder models. These AI models are trained end-to-end on video data and already outperform traditional codecs. Undoubtedly, all future codecs will be AI-powered.
In this video, Arsalan has an interview style conversation with Jona Ballé and Chris Finlay. Jona published some of the early work in AI-based image compression while at NYU and later at Google. Chris is the Head of Research at Deep Render and, along with the team, pioneered some of the earliest AI image and video codecs at Deep Render. His background is in applied mathematics. Arsalan is the Co-founder and CTO of Deep Render, where he led the team to build the world’s first AI codec, which currently outperforms traditional codecs by 50%. His background is in machine learning and aeronautical engineering.
Jona Ballé, Chris Finlay, and Arsalan Zafar discuss the history and future of AI-based compression. They talk about how they got into compression, their early research, the current challenges, and what they are working on now. They also share stories from their early days in the field and reflect on what it takes to invent an entirely new discipline.