Article Directory
What is Artificial Intelligence
AI: AI (Artificial Intelligence) is the science of making machines that:
- think humanly (cognitive modeling);
- think rationally (thinking law);
- act humanly (Turing test);
- act rationally (rational Agent)
The above definition is relatively comprehensive, and said a total of four dimensions. Four dimensions corresponding methods / Domain / Technique.
1) Acting humanly: The Turing Test approach Turing Test
Turing test is designed to provide a satisfactory operational definition for the smart (intelligent machines) . If a machine can pretend to humans can not distinguish between reality and its differences, then we say that this machine is smart (intelligent).
2) Thinking humanly: The cognitive modelling approach cognitive modeling
this dimension of cognitive science point of view (cognitive science approach) to explain AI.
3) Thinking rationally: The laws of thought law approach thinking
Thinking rationally (rational thinking) mainly related to solving some of derivation (reasoning). The classic is: Socrates famous syllogism (syllogisms) [4].
But the derivation of solving artificial intelligence (reasoning) There are two difficulties: 1) informal knowledge (problem) is difficult to describe in formal terms, especially in knowledge are usually not 100% sure of (uncertainty, such as the autopilot suddenly someone out from the side of the road); 2) "theoretically" be able to solve the problem, and the "fact" can be solved completely two concepts (unequal).
4) Acting rationally: The rational agent approach rational Agent
Acting rationally: rational action is action towards specific goals
Economics, philosophy, mathematics, psychology is the foundation of artificial intelligence.
Reason: refers to the property of a system, that is doing the right thing in a known environment.
History of artificial intelligence
Is divided into five stages:
forty-five sixties: early AI research (connectionism and symbolism development)
sixties and seventies: early AI program encountered difficulties (machine translation fails, the combinatorial explosion problem, Perceptron has " limitations ")
seventies and eighties: the prevalence of knowledge-based systems, expert systems commercialized
nineties 00s: Five other AI plan failed. AI research using the scientific method. Neural network regression. Intelligent Agent appear.
00's 10's: big data appears. Rise of the Machines learning
The first stage
Connectionist (neural network)
1943 Warren McCulloch and Walter Pitts proposed artificial neuron model, is considered the first artificial intelligence work.
In 1949 Donald Hebb proposed update a connection strength between neurons learning rules, called Hebbian learning (Hebbian learning), it is still affected.
In 1950 Marvin Minsky and DeanEdmonds built the first neural network computer (SNARC)
Symbolism (derivation system)
CMU Newell and Simon's logic theorist program LT, to prove that most of the "Principia Mathematica" theorem in Chapter 2. Subsequently developed a universal problem solver GPS simulate human problem solving (formally described by the sign and problem solving.)
IBM's Arthur Samuel (starting from 1952) to write checkers program.
second stage
Machine translation requires background knowledge
for the basic structure to produce intelligent behavior has some fundamental limitations; large-scale problems, so solving artificial intelligence inoperable
third phase of
the expert system
robust language understanding would require general knowledge about the world and the use of knowledge general method
The fourth stage
Science for studying AI: HMM (ie the Hidden Markov Model), Bayesian networks, neural networks (Neural Network), Agent appears
Bayesian networks provide a mathematical theory uncertain inference, the expert system can effectively deal with the uncertainty of knowledge.
The fifth stage
Large Data + machine learning
stage AI = large machine learning data + + HPC
Information sharing
Experiment download:
https://github.com/yyl424525/AI_Homework
Artificial Intelligence - A Modern Approach Chinese third edition pdf, courseware, operations and solutions, after-school exercise answers, test code and reports, over the years Kaobo title Download: HTTPS : //download.csdn.net/download/yyl424525/11310392