Open Data Science Conference 2015
Download as PPTX, PDF1 like894 views
The document discusses the importance of having more data and better algorithms for improving machine learning models, specifically in word alignment. It highlights the effects of various factors like data quantity, feature selection, and data cleanliness on classifier error rates. Additionally, it emphasizes the significance of open data and data enrichment platforms in enhancing data collection and usage.
Open Data Science Conference 2015
- 2. Why We Need More Data
- 4. The Effect of Better Algorithms 4 CrowdFlower, Inc. – Proprietary and Confidential 0% 5% 10% 15% 20% 25% Naïve Bayes Maximum Entropy SVM Classifier Error Rate
- 5. Real World Data 5 Active Semi-Supervised Learning for Improving Word Alignment (Vamshi ACL ’10)
- 6. The Effect of Better Features 6 CrowdFlower, Inc. – Proprietary and Confidential 0% 5% 10% 15% 20% 25% 30% Unigrams Bigrams Unigrams+Bigrams Classifier Error Rate
- 8. The Effect of More Data 8 CrowdFlower, Inc. – Proprietary and Confidential 0% 2% 4% 6% 8% 10% 12% 14% N 2N 4N Classifier Error Rate
- 9. Real World Data 9 Active Semi-Supervised Learning for Improving Word Alignment (Vamshi ACL ’10)
- 10. The Effect of Cleaner Data 10 CrowdFlower, Inc. – Proprietary and Confidential 0% 2% 4% 6% 8% 10% 12% 14% 90% Accurate Data 95% Accurate Data 100% Accurate Data Classifier Error Rate
- 11. 11 Where do Data Scientists Spend Their Time
- 12. The Power of Open Data
- 13. CrowdFlower Data Enrichment Platform 13
- 14. Color Data 14
- 15. 15
- 16. 16
- 17. 17
- 18. 18
- 19. 19
- 20. 20
- 21. Fleshmap 21
- 22. 22
- 24. 24
- 25. 25
- 26. Apple Watch 26
- 27. Apple Watch 27
- 28. Apple Watch 28
- 29. Apple Watch 29
- 31. Collecting the Same Data Over and Over 31
- 32. Open Data 32
- 33. Make Your Data Public Setting 33
- 35. Data For Everyone Library 35
- 40. Open Data API 40
- 41. Record Data 41
- 42. Extracting Names and Titles 42
- 43. Summarization 43
- 44. Is an Image Funny? 44
- 47. 47
- 48. 396 Scripts 48