"Representation, Reasoning, Learning"

Acknowledgements

Roadmap

We’ve Come a Long Way

This is 2001 …

… But Where is HAL?

Roadmap

Representation: Design Decisions

Atomic Worlds

Propositional Worlds

Object-Relational Worlds

Representation: Design Decisions

Universal Truths

Probable Worlds

Probable Worlds

Probable Worlds

Bayesian Networks

Bayesian Networks

Taxonomy of Methods

A Bridge

St. Nordaf University

Relational Schema

Possible Worlds

Representing the Distribution

Probabilistic Relational Models

PRM Semantics

PRMs are First-Order

The Ties that Bind

The Web of Influence

The Web of Influence

The Web of Influence

We’re Not in Kansas Anymore

Summary: Representation

Are We Done?

Roadmap

Reasoning: The Fight Continues

Reasoning: The Fight Continues

Expressive Languages

Complexity in Probabilistic Models

Weapon I: Locality

Weapon I: Locality

Weapon I: Locality

Hierarchies

Exploiting Hierarchical Structure

Weapon II: Universal Patterns

First-Order Probabilistic Inference

First-Order Probabilistic Inference

Exploiting Universals

Weapon III: Approximation

Stochasticity is Your Friend

Summary: Inference

Roadmap

Learning for Reasoning

TB Patients in San Francisco

TB Patients in San Francisco

A Web of Data

Standard Approach

What’s in a Link

What’s in an Anchor

Reasoning for Learning

Reasoning for Learning

Discovering Hidden Concepts

Discovering Hidden Concepts

Web of Influence, Yet Again

Summary: Learning

Another Bridge

Roadmap

Some Important Next Steps

Perception

Probabilistic Perception

Hypothetical Bridge

Hypothetical Bridge

Understanding Language

Resolving Ambiguity

Learning Semantic Models

A Tale of Three Bridges

A Tale of Three Bridges

Slide 78