Why Lossless Data Compression Is Important and Fun
summary
Claims that lossless compression wins real money and cites win cash in compression competitions and practical tips: find fast simple transforms that boost context modeling. Mentions concrete methods like symmetric numeral systems, Burrows‑Wheeler and LZ77, and concrete limits — Burrows‑Wheeler needs block size >10 KB while LZ77 can act on 5 bytes; DCT uses 64‑pixel blocks. Says >70% of Internet traffic is compressed video and speculates ~30 bits suffice for 90% accuracy in a simple personality model.
AI-written summary · about
chapters (7)
- 0:00Start
- 0:07Why data
- 2:54Context Modeling and Entropy encoding -- Quantitative and qualitative -- Block-oriented and
- 7:26Practical advices: 1. Find simple transforms (and better context modeling methods) 2. Analyze intermediate data
- 8:20Standard mode con
- 13:50The Ultimate con
- 14:11Foundational Questions Institut "Our Mathematical Universe"
related talks
52mI Want to Compress Everything
32mBeyond Zstandard: The Future of Data Compression
24mHow GZIP compression works
01h 04mCompress and Conquer: Compression Algorithms in Databases
09mCompressing the Library of Babel!
32mOn compression of everything in Postgres
27mCompression in Lucene #bbuzz
01h 09mBtune: Making Compression Better