From missile defense to user acquisition – How advanced AI models predict performance
summary
Shows three concrete ML builds: a deepfake lip-sync model trained on 14 hours of Obama video that transfers audio to video with no manual editing; a missile-defense classifier that infers missile type from sparse XY and Z coordinates along a flight path; and yellowHEAD’s predictive engine Alison, built from two years of tracked data and ten million dollars of ad spend, which forecasts D7 ROI with ~80% accuracy and measurably doubles correct campaign decisions.
AI-written summary · about
chapters (9)
- 0:00Introduction
- 2:10How was it done
- 3:32Missile defense
- 6:24User acquisition
- 8:46Allison
- 12:19Examples
- 14:36Results
- 18:15Summary
- 18:45QA
related talks
41mThe High Stakes Security of AI
21mSession: AI x UA – Redefining Creative Strategy For The Next Decade
08mBroadening AI Adoption with AutoML
48mMachine Takeover - Improving Your Game With Machine Learning & AI
26mAI and ML for Product Management
01h 09mAI is eating the world (and Google)
21mDemystifying Machine Learning for Programmatic UA & Retargeting
39mForward Focus: Perspectives on AI, Hype, and Security