MH - AI Summary

Chapter1

Learning
Data Mining - The process of analyzing large amounts of data in order to extract new kinds of useful information
Neural Network - A network of smple processors designed to emulate the fnctioning of human's brain
Case Based Reasoning - Solve problem by retrieving the solution of previous similar problems

Rule-based Systems
Rule-based system represents knowledge in terms of rules so that the system could conclude in different situations.
Knowledge base = IF-THEN rules and facts
inference engine = interpreter controlling the application of the fules given the facts
eg - prolog

Searching
Find efficient search strategy  to find a solution
eg - game theory/route planning

Planning
Coming up with actions that will increase the autonomy and flexibility of an agent through the construction of sequences of actions to achieve the goal.

Ability-Based Areas
Computer Vision
Natural language recognition/generation
Speech recognition/generation

Roboticsdevice that works automatically or by remote control
Programmed to perform tasks normally done by people



Chapter 2

Measuring An Agent's PerformanceReceives percepts >> enerates sequence of actions >> Performed sequence of states

Rationality of an agent

• Performance measures that define the criterion of successs
• Agent's initial knowledge of the environment
• Actions that the agent can perform
• Agent's percept sequence to date

Properties of Task Environments

Fully Observable - has access to complete state of the environment any time. Need to maintain any    internal state to keep track of the world
Partially Observable - caused by noise and inaccurate sensors or parts of the state are simply missing from sensor data

Deterministic - Next state is determined by the current state and action execute by agent
Stochastic - When the environment is Partially Observable (behavior cannot be predicted)

Episodic - Choice of ation in each episode depends only on the episode itself
Sequencial - Current decision coule affect all future decisions

Single/Multiagent - got brain then know =D



Solo Assignment Tittle *HAWT*

Examples are in italic and might be incorrect, correct me if so =D 

Logical AI

·       Represents knowledge of an agent's world, its goals and the current situation by sentences in mathematical logic language. 

·         The program decides what to do by inferring that certain actions are appropriate for achieving its goals.
eg. Stabilizing robot which obtain input from various sensors, formulate actions and maintain stable position 

Search

·         AI programs often examine large numbers of possibilities

·         Discoveries are continually made about how to do this more efficiently in various domains.

 e.g. moves in a chess game or inferences by a theorem proving program.

Pattern Recognition

·         When a program makes observations of some kind, it is often programmed to compare what it sees with a pattern.

·         For example, a vision program may try to match a pattern of eyes and a nose in a scene in order to find a face.

·         More complex patterns, e.g. in a natural language text, in a chess position, or in the history of some event.

·         These more complex patterns require different methods of approach.
eg. Speech and fingerprint recognition

Representation

·         Facts about the world have to be represented in some way.

·         Usually languages of mathematical logic are used.
eg. how to represent an object such as knife in computer world. How does the knife looks/feels like. 

Common sense knowledge and reasoning

·         This is the area in which AI is farthest from human-level.

·         It has been an active research area since the 1950s.

·         e.g. in developing systems of non-monotonic reasoning and theories of action

eg. The Cyc project aims at making a large base of common sense knowledge.
eg. Using a barometer to determine the height of a building. There are various methods to solve this problem and this topic deals with this sort of problem.

Planning

·        Planning programs start with general facts about the world (especially facts about the effects of actions), facts about the particular situation and a statement of a goal.

·         From these, they generate a strategy for achieving the goal.

·         The strategy is sequence of actions.
eg.  Planning a staff time table for an University?
eg. IBM Deep Blue Machine, planning chess strategy to beat an opponent

Inference

·         From some facts, others can be inferred.

·         Mathematical logical deduction is adequate for some purposes, but new methods of non-monotonic inference have been added to logic.

·         Default reasoning = simplest kind of non-monotonic reasoning.

·         >>conclusion is to be inferred by default but it can be withdrawn if there is evidence to the contrary.

·         For example, when we hear of a bird, we infer that it can fly, but this conclusion can be reversed when we hear that it is a penguin.

·         Ordinary logical reasoning is monotonic.

·         Conclusions that can be drawn from a set of premises are a monotonic increasing function of the premises.

·         Circumscription is another form of non-monotonic reasoning.
eg. Prolog (Programming in Logic) is an example of programming language which deals with inference machine in an expert system using an algorithm called backward chaining.

Learning from Experience

·         The approaches to AI based on connectionism and neural nets specialize in that.

·         There is also learning of laws expressed in logic.

·         Programs can only learn what facts or behaviors their formalisms can represent. Learning systems are almost all based on very limited abilities to represent information.
eg. In pattern recognition, the system can be fed many similar images and from there generate experience so that it could detect similar object in the future

Epistemology

·         This is a study of the kinds of knowledge that are required for solving problems in the world.
eg. The skills, rules and knowledge taxonomy of a human behavior being integrated into a system


 

Ontology

·         Ontology is the study of the kinds of things that exist.

·         In AI, the programs and sentences deal with various kinds of objects, and we study what these kinds are and what their basic properties are.

·         Emphasis on ontology begins in the 1990s.
eg. Semantic web, parsing used in Natural Language Processing

Heuristics

·         A heuristic is a way of trying to discover something or an idea imbedded in a program.

·         Heuristic functions are used in some approaches to search to measure how far a node in a search tree seems to be from a goal.

·         Heuristic predicates that compare two nodes in a search tree to see if one is better than the other
eg. Games such as tic tac toe

Genetic Programming

·         Genetic programming is a technique for getting programs to solve a task by mating random Lisp programs and selecting fittest in millions of generations.
eg. ECJ - Evolutionary Computation (written in Java)

Thanks Anson for helping me filling in the examples =D
The rest in hardcopy, treat me MooMoo ice cream then I MIGHT lend you photocopy =D