| 
|
Topic: frame problem
topics >
computer science >
philosophy
Topic:
computer as an intelligent agent
Topic:
context
Topic:
expert systems
Topic:
facts as relationships between entities
Topic:
knowledge representation by frames
Topic:
problem solving
Topic:
information as facts
Topic:
knowledge representation
Topic:
limitations of artificial intelligence and cognitive science
Topic:
limitations of formalism
Topic:
limitations of robots
Topic:
people vs. computers
Topic:
software models of reality
Topic:
using a world model in robotics
Summary
In artificial intelligence, the frame problem is the inability of naming every influence on an event or activity. You can not define a frame that limits the environment. Fodor's fridgeon property is a classic example: x is a fridgeon at time t if x exists and Fodor's fridge is on. Turning on the fridge changes the state of everything in the universe.While fridgeon is a silly example, the frame problem is pervasive. Context and environment is intertwined with events and entities. As Bongard problems illustrate, the very act of identifying entities and assigning meaning depends on the context.
Treating life as a primitive concept provides a way out of this dilemma. Instead of trying to explain meaning in terms of system, you explain system in terms of meaning. In a sense, system is creating a frame to help understand life. (cbb 6/06)
Subtopic: importance of context
Quote: frame problem: fridgeon is the property that x is a particle and my fridge is on; one act changes every state [»fodoJA_1987] Quote: the frame problem--can not name every conceivable thing that may go wrong; i.e., the frame is always incomplete [»mccaJ_1969]
Subtopic: simplifying
Quote: metaphysical assumption: the background/context can have the same structured description as objects [»dreyHL_1979] Quote: truth depends on both language and extralinguistic fact; but can not distinguish analytic statements as not based on fact; act of faith [»quinWV1_1951] Quote: 'stay near me' has many different meanings depending on its context [»dreyHL_1979] Quote: a human, rule-like behavior depends on current context [»dreyHL_1979] Quote: knowledge of background context need not be represented; instead acquired incrementally [»dreyHL_1979] Quote: intelligence can not be separated from the rest of human life [»dreyHL_1979]
Subtopic: understanding
Quote: the alphabet (events) for an object simplifies its behavior; e.g., ignore emptying the coin box of a vending machine [»hoarCA_1985] Quote: no theory for first step in problem solving: distinguishing essential from inessential [»dreyHL_1979] Quote: problem solving requires restricting class of possibly, relevant facts and then choosing relevant ones [»dreyHL_1979]
Subtopic: unexpected events
Quote: experts look at as many alternatives and moves ahead as a grandmaster [»dreyHL_1979]
Quote: robots can not handle unexpected events
[»thriMW_1983]
Group: philosophy (60 topics, 2323 quotes)
Topic: computer as an intelligent agent (49 items)
Topic: context (8 items)
Topic: expert systems (8 items)
Topic: facts as relationships between entities (22 items)
Topic: knowledge representation by frames (18 items)
Topic: problem solving (32 items)
Topic: information as facts (21 items)
Topic: knowledge representation (41 items)
Topic: limitations of artificial intelligence and cognitive science (64 items)
Topic: limitations of formalism (93 items)
Topic: limitations of robots (6 items)
Topic: people vs. computers (55 items)
Topic: software models of reality (24 items)
Topic: using a world model in robotics
(12 items)