Big Data Oslo v 4 | "When a Perfect Algorithm Meets Real Data" - Alessandra Cagnazzo
Download as PPTX, PDF1 like56 views
The document discusses the intersection of machine learning algorithms and real data, highlighting their ability to learn and provide insights while noting the importance of engineering structures. It warns that relying too heavily on deep learning can lead to inefficiencies, suggesting that shallower methods may sometimes be more effective. The document also touches on various mathematical expressions related to data processing.
Big Data Oslo v 4 | "When a Perfect Algorithm Meets Real Data" - Alessandra Cagnazzo
- 1. When a perfect algorithm meets real data the challenge of getting insight from data Alessandra Cagnazzo – Data Scientist Big Data Oslo 25 June 2019
- 2. Data
- 6. Use the superpower Of Machine Learning • It learns formulas partly by itself • Give insight, indications • One still has to engineer structures and formulas but is a less constrained way.
- 7. Threats are behind the corner
- 8. Careful assuming the generality of a machine learning algorit Deep is not always efficient. Sometimes it is better to go shallow, at the cost of losing s details.
- 9. 1 + 10 + 20 + 3 ≈ 5.83 1 + 10 + 20 + 3 ≈ 8.11 Parallelising, batching, and simplifying ⅇlog 𝑎 2 𝑒log 2 + 𝑒 𝑎 log 𝑎 𝑏 ⅇlog 𝑎 2 + 2 𝑎 𝑏 𝑐 + 𝑎 − 𝑒log 𝑎+log 𝑏 𝑐 −1 = 𝑎 + 𝑏 + 𝑐