University of Chicago

zed.uchicago.edu

Rare and Extreme Events In Complex Dynamical Systems

Urban Crime

Extreme weather phenomena

Seismic events

Long-range memory

Non-trivial stochastic effects

Can we learn stochastic phenomena non-parametrically?

Can we learn stochastic phenomena non-parametrically?

?

Can be more complicated...

A lot more complicated...

Algorithm genESeSS

What do the states mean?

State structure :

self-similarity in dynamical systems

The Fractalnet architecture

No back-propagation

Deep Learning without Neural Networks

Lets Predict Crime

• No manual selection of factors
• No creation of "lists"
• Uses only de-identified data

Philadelphia

The Problem:

• Predicting crime sufficiently ahead of time to be actionable
• Prediction precise enough in time and space to be actionable
• Use ONLY data that is realistically and cheaply available

Mean AUC

Property crime: 81%

Violent crime:    84%

Spatial tiles:

0.003 deg latitude, 0.003 deg longitude

0.25 miles across

Time-period:

Training:                      Jan 1 2016 - Dec 31 2018

Out-of-sample test:  Jan 1 2019 - April 1 2019

Prediction Performance (Philadelphia)

sensitivity   0.90
ppv           0.87

100 crimes

Raise 103 flags

90 correct flags

13 false positives

10 missed

3 day ahead prediction

Jan 1 2019

to

April 1 2019

Triangles: actual events

heatmap: predicted risk 3 days ahead

Triple homicide incident

Jan 7 2019

https://www.inquirer.com/crime/kensington-triple-shooting-homicide-philadelphia-police-20190107.html

Triangles: actual events

heatmap: predicted risk 3 days ahead

Predicting

extreme weather

Outperforms pure physics-based models at longer horizons

Next:

Model and predict world events

Nearly

Hands-free

No feature-engineering

The Issue with Neural Networks

only certain specific structures may be identified