Black Magic: How to apply Machine Learning to real-world problems
summary
Mattermark builds an automated funding-extraction pipeline that turns news articles into structured records (amount, series, investors, date) and pulls 2015 data of $156B across 8,532 events. The pipeline slices the problem into classifiers (TF‑IDF + SVM/gradient boosting with word2vec and semantic role labeling), regex and NER for amounts and names, and a meta confidence scoring model that outputs probabilities; items below threshold go to a small human team — a human plus machine workflow that raised precision into the 90s.
AI-written summary · about
chapters (8)
- 0:00Start
- 22:56Problem to solve
- 24:25Smaller Problems
- 25:17Classify Funding Article
- 27:51Classify Funding Sentences
- 29:26Extract Funding Entities
- 32:34Confidence Scoring
- 33:37Human Administration
related talks
08mQ n A with Dr Arun Verma
35mIndustrial Applications of Machine Learning in Search
01h 30mBecome a data-driven Organization with Machine Learning
37mMachine Learning models utilization
31mBuild threat detection models using ML
22mReal-world applications of machine learning: Beyond the hype
24mMachine Intelligence gets real
27mSuccessful Business With Machine Learning