๋ณธ๋ฌธ ๋ฐ”๋กœ๊ฐ€๊ธฐ

๋ถ„๋ฅ˜ ์ „์ฒด๋ณด๊ธฐ

(116)
[์ธ๊ณต์ง€๋Šฅ] 4. Beyond Classical Search - 2 2. Local Search In Continuous Space : ์ง€๊ธˆ๊นŒ์ง€๋Š” ํ˜„์žฌ ๋‚ด๊ฐ€ ์–ด๋–ค state์— ์žˆ๋Š”์ง€, action์„ ์ทจํ•˜๋ฉด ์–ด๋–ป๊ฒŒ ๋˜๋Š”์ง€, ์ด์‚ฐ์ , ๊ทธ๋ฆฌ๊ณ  ์ด ๊ฒŒ์ž„์˜ ๊ทœ์น™์ด ๋ญ”์ง€ ์•„๋Š” ๊ฒฝ์šฐ์˜€๋‹ค! ์ฆ‰ fully observable, deterministic, discrete, and known environment! discrete์™€ continuousํ•œ ํ™˜๊ฒฝ์˜ ๊ตฌ๋ถ„์€ ์‹œ๊ฐ„์ด ๋‹ค๋ค„์ง€๋Š” ๋ฐฉ๋ฒ•๊ณผ agent์˜ action๊ณผ percept์— ๋”ฐ๋ผ ์ด๋ฃจ์–ด์ง„๋‹ค. ์ง€๊ธˆ๊นŒ์ง€ ์„ค๋ช…ํ–ˆ๋˜ ์•Œ๊ณ ๋ฆฌ์ฆ˜์œผ๋กœ๋Š” continuous state์™€ action space๋ฅผ ๋‹ค๋ฃฐ ์ˆ˜๊ฐ€ ์—†๋‹ค. (๋‹จ, first-choice hill climbing๊ณผ simulated annealing ์ œ์™ธ) ์™œ๋ƒ๋ฉด ์—ฐ์†์ ์ธ ๊ฒฝ์šฐ branchin..
[์ธ๊ณต์ง€๋Šฅ] 4. Beyond Classical Search - 1 1. Local Search algorithm : Local Search algoritms์€ single current node์™€ ์ผ๋ฐ˜์ ์œผ๋กœ ๊ทธ ๋…ธ๋“œ์˜ neighbors node๋กœ ์›€์ง์ด๋Š” operation์„ ์‚ฌ์šฉํ•œ๋‹ค. ์ „ํ˜•์ ์œผ๋กœ search์— ์˜ํ•ด ๋”ฐ๋ผ์˜ค๋Š” path๋Š” ์œ ์ง€๋˜์ง€ ์•Š๋Š”๋‹ค. -> ๊ทธ๋ž˜์„œ little memory๋งŒ ์‚ฌ์šฉ (๋ณดํ†ต constant amount) -> ์•„์ฃผ ํฌ๊ฑฐ๋‚˜ infinite(continuous)ํ•œ state space์—์„œ ๋‚ฉ๋“ํ• ๋งŒํ•œ solution์„ ์ฐพ์„ ์ˆ˜ ์žˆ๋‹ค. (systematic algorithms์€ ์ ํ•ฉํ•˜์ง€ ์•Š๊ฑธ๋ž‘) -> ๋˜ํ•œ local search algorithm์€ optimization problem(objective funtion์— ๋”ฐ๋ผ ๊ฐ€์žฅ best state..
[์ธ๊ณต์ง€๋Šฅ] 3. Solving problems by searching - 3 5. Heuristic Search * Best-First Search : Best-first search๋Š” ์ผ๋ฐ˜์ ์ธ TreeSearch๋‚˜ GraphSearch algorithm์˜ instance์ด๋‹ค. ์ด ๋•Œ node๋Š” evaluation function์ธ f(n)์— ๊ธฐ์ดˆํ•ด expansion๋œ๋‹ค. Evaluation function์€ cost estimate์™€ ๊ฐ™์€ ๊ฒƒ์œผ๋กœ ์ดํ•ดํ•  ์ˆ˜ ์žˆ์œผ๋ฉฐ, ๊ทธ๋ž˜์„œ ๊ฐ€์žฅ ์ž‘์€ evaluation์„ ๊ฐ€์ง„ node๊ฐ€ ๊ฐ€์žฅ ๋จผ์ € expand๋œ๋‹ค. ์ฆ‰, Best-First Search๋Š” ๋Œ€๋ถ€๋ถ„์˜ TreeSearch๋‚˜ GraphSearch ์•Œ๊ณ ๋ฆฌ์ฆ˜์—์„œ ์“ฐ์ด๊ณ , node๋ฅผ ํ™•์žฅํ•ด ๋‚˜๊ฐˆ ๋•Œ ์–ด๋– ํ•œ evaluation function์„ ๊ธฐ์ค€์œผ๋กœ ํ•œ๋‹ค. ์ด ๋•Œ ์ด Evaluation f..
[์ธ๊ณต์ง€๋Šฅ] 3. Solving problems by searching - 2 4. Blind Search 1) Breadth-First Search (๋„ˆ๋น„์šฐ์„  ํƒ์ƒ‰) - FIFO(First in first out) : root node๋ฅผ ์ฒ˜์Œ์œผ๋กœ expandํ•˜๊ณ , ๋ชจ๋“  successors(์ž์† ๋…ธ๋“œ)๋ฅผ ๋‹ค์Œ์œผ๋กœ expandํ•œ ๋‹ค์Œ์— ๋˜ ๋‹ค์Œ successors...... - Performance of Breadth-First Search : complete ํ•˜์ง€๋งŒ optimal ํ•˜์ง€๋Š” ์•Š๋‹ค. (* complete : ๋ฌด์กฐ๊ฑด goal์„ ์ฐพ์„ ์ˆ˜ ์žˆ์Œ) ๋งŒ์•ฝ ๊ฐ€์žฅ ์–•์€ goal node๊ฐ€ depth d์— ์žˆ๋‹ค๋ฉด, ๋ชจ๋“  ๋” ์–•์€ node๋“ค์„ ํƒ์ƒ‰ํ•œ ๋‹ค์Œ์— ์ฐพ์„ ์ˆ˜ ์žˆ์„ ๊ฒƒ. ๊ทธ๋Ÿฌ๋‚˜ ๊ทธ ๊ฐ€์žฅ ์–•์€ goal node๊ฐ€ ํ•ญ์ƒ optimalํ•˜๋‹ค๊ณ ๋Š” ๋ณผ ์ˆ˜ ์—†์Œ. (๋‹จ, ๋ชจ๋“  cost๊ฐ€ ๊ฐ™์„ ..
[์ธ๊ณต์ง€๋Šฅ] 3. Solving problems by searching - 1 1. Problem-Solving Agents 1) Problem Formulation - Problem์€ initial state, actions, transition model, goal test, path cost๋กœ ์ •์˜๋œ๋‹ค. - Problem formulation์€ goal์ด ์ฃผ์–ด์ง„ ์ƒํ™ฉ์—์„œ, ๋ฌด์Šจ action ๊ณผ states๋ฅผ ๊ณ ๋ คํ•  ๊ฒƒ์ธ์ง€ ๊ฒฐ์ •ํ•˜๋Š” ๊ณผ์ •์„ ๋งํ•œ๋‹ค. - Problem์˜ solution์€ initial state๊ฐ€ goal state๋กœ ๊ฐ€๊ฒŒ ๋งŒ๋“œ๋Š” action๋“ค์˜ sequence์ด๋‹ค. - goal์— ๋‹ค๋‹ค๋ฅด๊ธฐ ์œ„ํ•œ sequence of actions์„ ์ฐพ์•„๋‚˜๊ฐ€๋Š” ๊ณผ์ •์ด search์ด๋‹ค. - solution์ด ์ฐพ์•„์ง€๋ฉด action์ด ์ˆ˜ํ–‰๋˜๋„๋ก ์ถ”์ฒœ๋œ๋‹ค. ์ด๊ฒƒ์ด ๋ฐ”๋กœ execution์ด..
[์ธ๊ณต์ง€๋Šฅ] 1.1 What is AI? &2.3 The Nature of Environments 1. AI(์ธ๊ณต์ง€๋Šฅ) ์ด๋ž€? : ์‚ฌ๋žŒ์˜ Intelligent๋ฅผ ๋ชจ๋ฐฉํ•˜๋Š” ๊ธฐ๊ณ„๋ฅผ ๋งŒ๋“ค์–ด๋ณด์ž! - Thinking Humanly : ์‚ฌ๋žŒ๋‹ต๊ฒŒ ์ƒ๊ฐํ•˜๋Š”๊ฒŒ ๋ญ”๋ฐ? - Thinking Rationally : ๋…ผ๋ฆฌํ•™๊ณผ ๊ด€๋ จ๋จ(3๋‹จ๋…ผ๋ฒ•) - Acting Humanly : Turing test - Acting Rationally : ํ•ฉ๋ฆฌ์ ์œผ๋กœ ํ–‰๋™ํ•˜๋Š” ๊ฒƒ๊ณผ ํ•ฉ๋ฆฌ์ ์œผ๋กœ ์ƒ๊ฐํ•˜๋Š” ๊ฒƒ์€ ๋‹ค๋ฆ„ (ํ–‰๋™ํ•œ๋‹ค๊ณ  ์ƒ๊ฐํ•˜๋Š” ๊ฑด ์•„๋‹˜) EX) ๋„ค๋น„๊ฒŒ์ด์…˜ ์ฆ‰, AI = Science(์ƒ๊ฐ) & Engineering(ํ–‰๋™) (๊ด„ํ˜ธ์•ˆ์€ task environment for an automated taxi) 2. The Nature Of Environments 1) Specifying the task environment : agent๊ฐ€ ์žˆ์„..
[์ด์‚ฐ์ˆ˜ํ•™] Chapter 4. Algorithms 1. Introduction - ์•Œ๊ณ ๋ฆฌ์ฆ˜์€ ๋‹ค์Œ๊ณผ ๊ฐ™์€ ํŠน์ง•๋“ค์„ ๊ฐ€์ง€๊ณ  ์žˆ๋‹ค. 1) input 2) output 3) precision : ์ •ํ™•ํ•˜๊ฒŒ ์„œ์ˆ ๋˜์–ด์•ผ ํ•œ๋‹ค 4) determinism : ์ˆ˜ํ–‰์˜๊ฒฐ๊ณผ๋Š” uniqueํ•˜๋‹ค. ๋˜ํ•œ input๊ณผ ์„ ํ–‰๋œ step์— ๋”ฐ๋ผ ๊ฒฐ์ •๋œ๋‹ค. 5) finiteness : ์•Œ๊ณ ๋ฆฌ์ฆ˜์€ ๋๋‚˜์•ผ ํ•œ๋‹ค. (์œ ํ•œ๊ฐœ์˜ instruction์ด ์ˆ˜ํ–‰๋œ ์ดํ›„์—” ๋ฉˆ์ถฐ์•ผ ํ•จ) 6) correctness : ์•Œ๊ณ ๋ฆฌ์ฆ˜์— ์˜ํ•ด ๋„์ถœ๋œ output์€ ๋งž์•„์•ผ ํ•œ๋‹ค. 7) generality : ๋ชจ๋“  input์— ๋Œ€ํ•ด ์ ์šฉ๋˜์–ด์•ผ ํ•œ๋‹ค. 2. Examples of Algorithms - ํฐ ์ˆ˜ ์ฐพ๊ธฐ ์•Œ๊ณ ๋ฆฌ์ฆ˜, ๋ฌธ์ž์—ด ์ฐพ๊ธฐ ์•Œ๊ณ ๋ฆฌ์ฆ˜, ์‚ฝ์ž… ์ •๋ ฌ ์•Œ๊ณ ๋ฆฌ์ฆ˜ - Randomized Algorithms (๋ฌด์ž‘์œ„ ..
[์ด์‚ฐ์ˆ˜ํ•™] Chapter 3. Relations 1. Relations - ์ฃผ์–ด์ง„ ์ง‘ํ•ฉ X, Y์—์„œ Cartesian product X x Y ๋ฅผ ํ•˜๋ฉด ๊ทธ ๊ฒฐ๊ณผ๋Š” x∈X, y∈Y์ธ (x, y)์˜ ๋ชจ๋“  ordered paris ์ด๋‹ค. X x Y = {(x,y) | x∈X and y∈Y} -Binary relation (์ด์ง„ ๊ด€๊ณ„) : ๋‘ ์ง‘ํ•ฉ์˜ ์›์†Œ ์‚ฌ์ด์˜ ๊ด€๊ณ„ ์ง‘ํ•ฉ X ์—์„œ ์ง‘ํ•ฉ Y ๋กœ์˜ binary relation R์€, Cartesian product X x Y ์˜ subset(๋ถ€๋ถ„์ง‘ํ•ฉ)์ด๋‹ค. Ex) X = { 1, 2, 3 } and Y = { a, b } R = {(1, a), (1, b), (2, b), (3, a)} ๋Š” X์™€ Y ์‚ฌ์ด์˜ relation์ด๋‹ค. - Domain : y∈Y์ธ y์— ๋Œ€ํ•˜์—ฌ (x, y)๊ฐ€ relation R์— ์†ํ•  ๋•Œ..
[Node.js] 1. Node.js๋ž€? + ์ž‘์—…ํ™˜๊ฒฝ ์„ค์ •ํ•˜๊ธฐ Node.js๋ž€ ์ž๋ฐ”์Šคํฌ๋ฆฝํŠธ ์—”์ง„์— ๊ธฐ๋ฐ˜ํ•ด ๋งŒ๋“ค์–ด์ง„ ์„œ๋ฒ„ ์‚ฌ์ด๋“œ ํ”Œ๋žซํผ์ด๋‹ค! * Node ์ž์ฒด๋Š” ์›น์„œ๋ฒ„๊ฐ€ ์•„๋‹˜. HTTP ์„œ๋ฒ„๋ฅผ ์ง์ ‘ ์ž‘์„ฑํ•ด์•ผ ํ•จ. ๊ทธ์ € JS ๋Ÿฐํƒ€์ž„์ผ ๋ฟ! ๋‚ด๊ฐ€ ๋ฆฌ์•กํŠธํ• ๋•Œ๋„ babel๊ณผ ๊ฐ™์€ ๋ช‡๋ช‡ ๊ธฐ๋Šฅ์„ ์‚ฌ์šฉํ•  ๋•Œ ์•Œ๊ฒŒ ๋ชจ๋ฅด๊ฒŒ node.js๋ฅผ ์‚ฌ์šฉํ•œ๊ฑฐ๋ผ๊ณ  ํ•œ๋‹ค. ๋˜, npm์ด๋ผ๋Š” package manager ์—ญ์‹œ node.js์˜ ๊ฒƒ Node.js์˜ ํŠน์ง• 1. ๋น„๋™๊ธฐ I/O ์ฒ˜๋ฆฌ, ์ด๋ฒคํŠธ์œ„์ฃผ 2. ๋น ๋ฅธ ์†๋„ 3. ๋‹จ์ผ ์“ฐ๋ ˆ๋“œ, ๋›ฐ์–ด๋‚œ ํ™•์žฅ์„ฑ 4. ๋ฒ„ํผ๋ง์ด ์—†์Œ 5. ๋ผ์ด์„ผ์Šค ๊ต‰์žฅํžˆ ๋งŽ์€ ํšŒ์‚ฌ์™€ ํ”„๋กœ์ ํŠธ์—์„œ Node.js๋ฅผ ์‚ฌ์šฉํ•˜๊ณ  ์žˆ๋‹ค. ๋‚ด๊ฐ€ ํ•ด์•ผ ํ•  ๊ฒƒ์€ express ๊ณต๋ถ€์ธ๋ฐ, ์ด express๋Š” node.js๋ฅผ ๊ธฐ๋ฐ˜์œผ๋กœ ํ•œ ํ”„๋ ˆ์ž„์›Œํฌ์ด๊ธฐ ๋•Œ๋ฌธ์—, ์šฐ์„  node.js์— ๋Œ€ํ•ด ๊ฐ„๋žตํžˆ ๊ณต..
[git] git clone, pull 1. ๋ ˆํฌ์ง€ํ† ๋ฆฌ fork 2. git clone 3. ๋ธŒ๋žœ์น˜ ์ƒ์„ฑ