7.2 Setting

์ด section์—์„œ์˜ ์‹คํ—˜์€ partially observable, cooperative multi-agent์ƒํ™ฉ์„ ๊ฐ€์ •ํ•ฉ๋‹ˆ๋‹ค. ์ด ๋•Œ, Markov state(์‹ค์ œ state)st s_t๋กœ ๋ถ€ํ„ฐ, discrete feature์˜ ์ง‘ํ•ฉ ftf_t๋Š” ๋ชจ๋“  agent๊ฐ€ common knowledge๋กœ ์•Œ๊ณ  ์žˆ๋Š” ftpubf^{\mathrm{pub}}_t์™€ agent ์ „์ฒด๊ฐ€ ์•Œ๊ณ ์žˆ์ง€๋Š” ์•Š๋”๋ผ๋„ ํ•œ agent ์ด์ƒ์ด ์•Œ๊ณ  ์žˆ๋Š” ์ •๋ณด(์ „๋ถ€๋Š” ์•„๋‹Œ) ftpri f^\mathrm{{pri}}_t๋กœ ์ด๋ฃจ์–ด์ ธ ์žˆ์Šต๋‹ˆ๋‹ค. ์˜ˆ๋ฅผ ๋“ค๋ฉด, ํ…Œ์ด๋ธ”์— ์žˆ๋Š” ๋ชจ๋“  ์ •๋ณด๋Š” ftpubf^{\mathrm{pub}}_t์— ์†ํ•˜์ง€๋งŒ, ํ•œ agent๊ฐ€ ๊ฐ์ž๊ฐ€ ๋“ค๊ณ  ์žˆ๋Š” ํŒจ๋Š” ftpri f^\mathrm{{pri}}_t์— ์†ํ•ฉ๋‹ˆ๋‹ค. ์—ฌ๊ธฐ์„œ๋Š” ์ด๋Ÿฌํ•œ state feature์˜ ๋ถ„๋ฆฌ๊ฐ€ ๋ชจ๋‘์—๊ฒŒ ์•Œ๋ ค์ ธ์žˆ๋‹ค๋Š” ๊ฒƒ์„ ๊ฐ€์ •ํ•ฉ๋‹ˆ๋‹ค.

Last updated