Video Compression Insights
Auke discusses the complexities of mapping learned encoders and decoders to meaningful physical properties, revealing the unpredictable nature of their responses to different images. He highlights the team's simultaneous efforts in developing neural codecs, emphasizing the potential of transformer-based models as foundational elements for both real-time and streaming video compression applications.In this clip
From this podcast
The TWIML AI Podcast (formerly This Week in Machine Learning & Artificial Intelligence)
Advances in Neural Compression with Auke Wiggers - #570
Related Questions