Completed
What should a well-educated scientist know about sampling?
Class Central Classrooms beta
YouTube videos curated by Class Central.
Classroom Contents
The Power of Sampling by Peter W. Glynn
Automatically move to the next video in the Classroom when playback concludes
- 1 Date: 14 August 2019, 16:00 to
- 2 Introduction to speaker
- 3 The Power of Sampling
- 4 What should a well-educated scientist know about sampling?
- 5 Thesis of Lecture:
- 6 According to Wikipedia:
- 7 Some History:
- 8 First sampling-based survey:
- 9 Second known instance of sampling
- 10 Today, statistical sampling is used everywhere
- 11 Such randomized trials form one setting where one can legitimately claim evidence of a causal effect
- 12 Key Philosophical Underpinning:
- 13 1936 US Presidential Election
- 14 Problems with Literary Digest survey:
- 15 Moral of the Story:
- 16 Subtle modifications in sampling can have a big impact
- 17 Why?
- 18 Many other such subtleties exist in applying sampling methods
- 19 Active research area:
- 20 Sampling in Computational Mathematics
- 21 High-dimensional Integration:
- 22 High-level Perspective:
- 23 History
- 24 Analysis of Monte Carlo Method:
- 25 If it's so slow, why is it so widely used?
- 26 The Numerical Alternative to Monte Carlo
- 27 Better integration rules:
- 28 Tracking discontinuities is easy in d = 1
- 29 Versatility of Monte Carlo Method:
- 30 "Curse of Dimensionality"
- 31 Result
- 32 What about Monte Carlo?
- 33 Example: Compute volume of region A c [0, 1]d
- 34 So, Monte Carlo is dimensionally insensitive...
- 35 Coding Flexibility and Visualization: Aan Example
- 36 Modify model so that time spent in just one state is non-exponential...
- 37 Monte Carlo Alternative:
- 38 A Key Difference between Statistical Sampling and Monte Carlo Sampling
- 39 Sampling in Synthesizing Distributed Controls
- 40 The agents typically need state information to make good decisions
- 41 Example: Assigning incoming jobs to servers on a server farm
- 42 Approach 1: Centralized controller assigns incoming jobs to shortest queue
- 43 Sampling plays a key role in many machine learning algorithms
- 44 Final words:
- 45 Q&A
