Á¤º¸Ã³¸®ÇÐȸ ³í¹®Áö ¼ÒÇÁÆ®¿þ¾î ¹× µ¥ÀÌÅÍ °øÇÐ
Current Result Document : 1,129 / 1,130
ÇѱÛÁ¦¸ñ(Korean Title) |
°ü°èÇü °È ÇнÀÀ» À§ÇÑ µµ¸ÞÀÎ Áö½ÄÀÇ È¿°úÀûÀÎ È°¿ë |
¿µ¹®Á¦¸ñ(English Title) |
Effective Utilization of Domain Knowledge for Relational Reinforcement Learning |
ÀúÀÚ(Author) |
°¹Î±³
±èÀÎö
MinKyo Kang
InCheol Kim
|
¿ø¹®¼ö·Ïó(Citation) |
VOL 11 NO. 03 PP. 0141 ~ 0148 (2022. 03) |
Çѱ۳»¿ë (Korean Abstract) |
ÃÖ±Ù µé¾î °È ÇнÀÀº ½ÉÃþ ½Å°æ¸Á ±â¼ú°ú °áÇÕµÇ¾î ¹ÙµÏ, ü½º¿Í °°Àº º¸µå °ÔÀÓ, Atari, StartCraft¿Í °°Àº ÄÄÇ»ÅÍ °ÔÀÓ, ·Îº¿ ¹°Ã¼ Á¶ÀÛ ÀÛ¾÷ µî°ú °°Àº ´Ù¾çÇÑ ºÐ¾ß¿¡¼ ¸Å¿ì ³î¶ó¿î ¼º°øÀ» °ÅµÎ¾ú´Ù. ÇÏÁö¸¸ ÀÌ·¯ÇÑ ½ÉÃþ °È ÇнÀÀº Çൿ, »óÅÂ, Á¤Ã¥ µîÀ» ¸ðµÎ º¤ÅÍ ÇüÅ·ΠǥÇöÇÑ´Ù. µû¶ó¼ ±âÁ¸ÀÇ ½ÉÃþ °È ÇнÀÀº ÇнÀµÈ Á¤Ã¥ÀÇ Çؼ® °¡´É¼º°ú ÀϹݼº¿¡ Á¦ÇÑÀÌ ÀÖ°í, µµ¸ÞÀÎ Áö½ÄÀ» ÇнÀ¿¡ È¿°úÀûÀ¸·Î È°¿ëÇϱ⵵ ¾î·Æ´Ù´Â ÇѰ輺ÀÌ ÀÖ´Ù. ÀÌ·¯ÇÑ ÇÑ°èÁ¡µéÀ» ÇØ°áÇϱâ À§ÇØ Á¦¾ÈµÈ »õ·Î¿î °ü°èÇü °È ÇнÀ ÇÁ·¹ÀÓ¿öÅ©ÀÎ dNL-RRLÀº ¼¾¼ ÀÔ·Â µ¥ÀÌÅÍ¿Í Çൿ ½ÇÇà Á¦¾î´Â ±âÁ¸ÀÇ ½ÉÃþ °È ÇнÀ°ú ¸¶Âù°¡Áö·Î º¤ÅÍ Ç¥ÇöÀ» ÀÌ¿ëÇÏÁö¸¸, Çൿ, »óÅÂ, ±×¸®°í ÇнÀµÈ Á¤Ã¥Àº ¸ðµÎ ³í¸® ¼¼úÀÚ¿Í ±ÔÄ¢µé·Î ³ªÅ¸³»´Â °ü°èÇü Ç¥ÇöÀ» ÀÌ¿ëÇÑ´Ù. º» ³í¹®¿¡¼´Â dNL-RRL °ü°èÇü °È ÇнÀ ÇÁ·¹ÀÓ¿öÅ©¸¦ ÀÌ¿ëÇÏ¿© Á¦Á¶ ȯ°æ ³»¿¡¼ ¿î¼Û¿ë ¸ð¹ÙÀÏ ·Îº¿À» À§ÇÑ Çൿ Á¤Ã¥ ÇнÀÀ» ¼öÇàÇÏ´Â È¿°úÀûÀÎ ¹æ¹ýÀ» Á¦½ÃÇÑ´Ù. ƯÈ÷ º» ¿¬±¸¿¡¼´Â °ü°èÇü °È ÇнÀÀÇ È¿À²¼ºÀ» ³ôÀ̱â À§ÇØ, Àΰ£ Àü¹®°¡ÀÇ »çÀü µµ¸ÞÀÎ Áö½ÄÀ» È°¿ëÇÏ´Â ¹æ¾ÈµéÀ» Á¦¾ÈÇÑ´Ù. ¿©·¯ °¡Áö ½ÇÇèµéÀ» ÅëÇØ, º» ³í¹®¿¡¼ Á¦¾ÈÇÏ´Â µµ¸ÞÀÎ Áö½ÄÀ» È°¿ëÇÑ °ü°èÇü °È ÇнÀ ÇÁ·¹ÀÓ¿öÅ©ÀÇ ¼º´É °³¼± È¿°ú¸¦ ÀÔÁõÇÑ´Ù. |
¿µ¹®³»¿ë (English Abstract) |
Recently, reinforcement learning combined with deep neural network technology has achieved remarkable success in various fields such as board games such as Go and chess, computer games such as Atari and StartCraft, and robot object manipulation tasks. However, such deep reinforcement learning describes states, actions, and policies in vector representation. Therefore, the existing deep reinforcement learning has some limitations in generality and interpretability of the learned policy, and it is difficult to effectively incorporate domain knowledge into policy learning. On the other hand, dNL-RRL, a new relational reinforcement learning framework proposed to solve these problems, uses a kind of vector representation for sensor input data and lower-level motion control as in the existing deep reinforcement learning. However, for states, actions, and learned policies, It uses a relational representation with logic predicates and rules. In this paper, we present dNL-RRL-based policy learning for transportation mobile robots in a manufacturing environment. In particular, this study proposes a effective method to utilize the prior domain knowledge of human experts to improve the efficiency of relational reinforcement learning. Through various experiments, we demonstrate the performance improvement of the relational reinforcement learning by using domain knowledge as proposed in this paper. |
Å°¿öµå(Keyword) |
°ü°èÇü °È ÇнÀ
µµ¸ÞÀÎ Áö½Ä
Çൿ Á¤Ã¥
³í¸® ¼¼úÀÚ
ÀϹݼº
Çؼ® °¡´É¼º
Relational Reinforcement Learning
Domain Knowledge
Policy
Logic Predicate
Generality
Interpretability
|
ÆÄÀÏ÷ºÎ |
PDF ´Ù¿î·Îµå
|