Introducing 

Prezi AI.

Your new presentation assistant.

Refine, enhance, and tailor your content, source relevant images, and edit visuals quicker than ever before.

Loading…
Transcript

- usefulness

- displayed behavior

- logical feat

- philosophical feat

- learning capacity

Frame-of-reference problem

Dictate-ive

minimal or non-existent window for "emergence"

Liberate-ive

good window for "emergence"

"... any intelligent system must have access to a body with which to interact and learn from its environment ..." (Brooks, 1999)

"It can also be maintained that it is best to provide the machine with the best sense organs that money can buy,

and then teach it to understand and speak English. That process could follow the normal teaching of a child.

Things would be pointed out and named, etc."

(Turing, 1950).

overall preset rules and behaviors

mainly concerned with "directly instructing" the machine

Problem:

Symbol grounding

Physical Symbol System Hypothesis (PSSH)

how a symbol represent its actual meaning be it physical or abstract

explicit

Application example:

  • simple actions such as: push the big box, move from here to there
  • inaccuracy due to real-world real-time demands

e.g. box moved while it's in "Plan" phase

  • Sense Model Plan Act (SMPA)
  • limited controlled world
  • Expert System

(http://www.ai.sri.com/shakey/)

  • Prolog

everything can be represented as symbols

intelligence is the manipulation of these symbols

Example:

Shakey

Stanford Research Institute

1966-1972

"emergence"

the process of coming into existence or prominence

mid 17th century (in the sense ‘unforeseen

occurrence’): from medieval Latin emergentia, from -

Latin emergere 'bring to light'

(http://oxforddictionaries.com/definition/emergence)

EMERGENCE

  • robot equipped with 2 obstacle sensors, left and right
  • robot simply turn in opposite direction where obstacle detected
  • emergence: clustering behavior
  • frame-of-reference problem: are the robots intelligence?

Example of emergence:

Swiss robot experiment

AI Lab - University of Zurich (1990s)

rule-oriented

GOFAI

What i'm looking at:

Property of approach:

Robotics

Good Old-Fashioned AI

Area of interest:

idea of INTELLIGENCE

What i'm NOT looking at:

relativity

(elaborated later on)

Why not?

Classification of

AI

Approaches in

This is based on my own thinking.

It is my own classification, the way I look at approaches in AI.

Information presented here is not meant to be exhaustive.

And so, property-wise, the 2 approaches are:

I made up these words

Dictate-ive

Liberate-ive

implicit

preset basic components/methods as building block

Likewise any attempt to classify such diverse field, there are bound to be things in-between

mainly concerned with "indirectly instructing" the machine

"dictate"

control or decisively affect; determine

(http://oxforddictionaries.com/definition/dictate)

"liberate"

release (someone) from a situation which limits freedom of thought or behaviour

(http://oxforddictionaries.com/definition/liberate)

Neural

Network

mimicry of property of human biological neural network

Application example:

Developmental Robotics

(controller-centered)

Autonomous Mental Development (AMD)

  • developmental psychology
  • neurobiology

Evolutionary

Computation

Embodied

AI

"Evolution"

the gradual development of something

(http://oxforddictionaries.com/definition/evolution)

Addendum

Application example:

Evolutionary Algorithms

mimicry of biological evolution

  • population
  • reproduction
  • recombination
  • mutation
  • selection

development of system through a process of gradual change

Evolutionary Robotics

(controller-centered)

use of Evolutionary Computation methods to develop robots

Example:

Evolutionary computation and embodied system

Resilient Machines Through Continuous Self-Modeling

Josh Bongar, Viktor Zykov, Hod Lipson (2006)

Artificial Ontogeny

evolution of both robot morphology and controller

  • typically in modular robots

Recommended

reading

Some personal insight:

(Darwinian evolution)

Running Dog Robot Project

AI Lab - University of Zurich (2004)

  • neurobiology
  • No method is particularly inferior or superior, this is application-dependent

Example:

Biomimicry and morphological computation

Example of morphological computation in human:

reflex response (e.g. knee jerk reflex)

How the Body Shapes the Way We Think: A New View of Intelligence

Rold Pfeifer, Josh Bongard

MIT Press, 2006

394 pages

  • Have an appreciation for GOFAI, it is where the other approaches are ultimately at its root built upon
  • Use of mixed approaches

http://www.icub.org/

Current research:

http://eccerobot.org/

Learn more about creating dynamic, engaging presentations with Prezi