Artificial Intelligence (AI)(ch. 15)

Tasks (Humans)

Computational; Recognition; Reasoning

Computational

adding a column of numbers, sorting a list of numbers, searching for a given name in a list of names, balancing a spreadsheet, calculating trajectory

4 Characteristics of a Knowledge Representation

adequacy, efficiency, extendability, appropriatness

Drone (UAV) Unmanned Aerial Vehicle

aircraft under the autonomous control via a computer system on board or controlled by a human controller at a remote site

Turing Test

allows a human to interrogate two entities both hidden, asks questions and if by the end the interrogate can't determine who's the robot and who's not, then the machine has passed the test

Explanation Facility

allows user to see the assertions and rules used in arriving at a conclusion

Expert Systems

also called rule-based systems or knowledge-based systems; attempts to mimic the human ability to engage pertinent facts and string them all together in a logical fashion to reach a conclusion; "if" clauses

Recommendation Software

an intelligent agent to welcome a returning customer to its webpage and make suggestions on future purchases based on past ones

Neuron

axon sends impulses, dendrites are trees that branch impulses between spaces, synapses do the connecting

Knowledge Engineering

builder of an expert system uses information from experts to build their database/system

Swarm Intelligence Model

captures collective behavior; swarm (think any colony they all have the same goal but are acting individually & communicating along while)

State-Space Graph

each node of the graph represents a state of the problem and we have a goal state or states in mind. path takes us from initial state to winning configuration and the nodes along the way represent intermediate configurations

Back Propagation Algorithm

error estimates that are passed back from the output layer, eventually causes the network to settle into a stable state where it can correctly respond to any desired degree of accuracy to all inputs in training set

Intelligence Agents

form of software technology that is designed to interact collaboratively with a user somewhat in the mode of a personal assistant. assistant must be trained in the beginning, over time it learns and anticipates what needs to be done (think smart house)

Robot

implies a device which may or may not be humanlike form that has the ability to gather sensory information from its surroundings and to autonomously perform mechanical actions in response; must not only gather sensory information but also filter out the possibly vast amount of data its surroundings might present to it to "recognize"

Formal Language

language of formal logic, usually expressed more symbolically

Connectionist Architecture

large number of simple "processors" with multiple interconnections

Types of Knowledge Representation

natural language; formal language (sacrifices richness of expression); pictorial; graphical

Training Data

network learns experience by modifying the weights on its connections. networks can be given a set of weights and thresholds that act as an educated guess

Push Technology

personalized web search engine (pinterest)

Reasoning

planning on what to wear, deciding on strategic planning for a company, leadership

Recognition

recognizing your best friend, understanding a spoken word, finding the tennis ball in the grass in your back yard

Efficiency

representation must be minimalist, avoids redundant info

Adequacy

representation must capture all relevant knowledge: formal language

Appropriateness

representation should befit the knowledge domain being presented

Deliberative Strategy

says that the robot must have an internal representation of its environment and that its actions in response to some stimuli are programmed into the robot based on this model of environment

Winograd Schema

sentence that involves two entities and a pronoun that refers back to one of the two entities, one must decide to whom the pronoun belongs with (simple for humans, hard for humans)

Extendability

should be relatively easy to extend to include new knowledge as it's acquired

Semantic Net

terminology of object orientation, rectangles represent classes or objects, oval represent properties, arcs represent relationships

Reactive Strategy

uses heuristic algorithms to allow robot to respond directly to stimuli from its environment without filtering through some line of reasoning based on its internal understanding of the environment

State-Space Search

what a state space graph represents