Harnessing implicit neural representations for scientific data compression

TL;DR - A hierarchical implicit neural representation compresses large scientific datasets into compact functions while better preserving fine-scale details. This could make massive scientific measurements more practical to store and process.

• Implicit neural representations encode measurements as learned continuous functions rather than explicit data arrays.
• The hierarchical approach is designed to retain richer small-scale structure during compression.
• The provided summary does not report specific datasets, compression ratios, or reconstruction results.