Análisis de dos algoritmos de Reinforcement Learning aplicados a OpenAi Gym Retro
El documento se enfoca en hacer una descripción de los algoritmos de reinforcement learning PPO y DQN, los cuales si bien son comparados con el framework Gym-retro de OpenAI (utilizando el juego de NES ice-climbers), cuentan con tantas optimizaciones y estilos que el alcance de las comparaciones es...
- Autores:
-
González Oviedo, Rodrigo José
- Tipo de recurso:
- Trabajo de grado de pregrado
- Fecha de publicación:
- 2023
- Institución:
- Universidad de los Andes
- Repositorio:
- Séneca: repositorio Uniandes
- Idioma:
- spa
- OAI Identifier:
- oai:repositorio.uniandes.edu.co:1992/69749
- Acceso en línea:
- http://hdl.handle.net/1992/69749
- Palabra clave:
- Reinforcement learning
Nintendo entertainment system
Game agents
Artificial intelligence
IceClimber
DQN
PPO
Gym retro
Stable baselines
Ingeniería
- Rights
- openAccess
- License
- Atribución 4.0 Internacional
| id |
UNIANDES2_62839b20792d15b2508cd88e824eebe4 |
|---|---|
| oai_identifier_str |
oai:repositorio.uniandes.edu.co:1992/69749 |
| network_acronym_str |
UNIANDES2 |
| network_name_str |
Séneca: repositorio Uniandes |
| repository_id_str |
|
| dc.title.none.fl_str_mv |
Análisis de dos algoritmos de Reinforcement Learning aplicados a OpenAi Gym Retro |
| title |
Análisis de dos algoritmos de Reinforcement Learning aplicados a OpenAi Gym Retro |
| spellingShingle |
Análisis de dos algoritmos de Reinforcement Learning aplicados a OpenAi Gym Retro Reinforcement learning Nintendo entertainment system Game agents Artificial intelligence IceClimber DQN PPO Gym retro Stable baselines Ingeniería |
| title_short |
Análisis de dos algoritmos de Reinforcement Learning aplicados a OpenAi Gym Retro |
| title_full |
Análisis de dos algoritmos de Reinforcement Learning aplicados a OpenAi Gym Retro |
| title_fullStr |
Análisis de dos algoritmos de Reinforcement Learning aplicados a OpenAi Gym Retro |
| title_full_unstemmed |
Análisis de dos algoritmos de Reinforcement Learning aplicados a OpenAi Gym Retro |
| title_sort |
Análisis de dos algoritmos de Reinforcement Learning aplicados a OpenAi Gym Retro |
| dc.creator.fl_str_mv |
González Oviedo, Rodrigo José |
| dc.contributor.advisor.none.fl_str_mv |
Takahashi Rodríguez, Silvia |
| dc.contributor.author.none.fl_str_mv |
González Oviedo, Rodrigo José |
| dc.contributor.jury.none.fl_str_mv |
Takahashi Rodríguez, Silvia |
| dc.subject.keyword.none.fl_str_mv |
Reinforcement learning Nintendo entertainment system Game agents Artificial intelligence IceClimber DQN PPO Gym retro Stable baselines |
| topic |
Reinforcement learning Nintendo entertainment system Game agents Artificial intelligence IceClimber DQN PPO Gym retro Stable baselines Ingeniería |
| dc.subject.themes.es_CO.fl_str_mv |
Ingeniería |
| description |
El documento se enfoca en hacer una descripción de los algoritmos de reinforcement learning PPO y DQN, los cuales si bien son comparados con el framework Gym-retro de OpenAI (utilizando el juego de NES ice-climbers), cuentan con tantas optimizaciones y estilos que el alcance de las comparaciones es limitado. |
| publishDate |
2023 |
| dc.date.accessioned.none.fl_str_mv |
2023-08-16T13:44:18Z |
| dc.date.available.none.fl_str_mv |
2023-08-16T13:44:18Z |
| dc.date.issued.none.fl_str_mv |
2023-08-15 |
| dc.type.es_CO.fl_str_mv |
Trabajo de grado - Pregrado |
| dc.type.driver.none.fl_str_mv |
info:eu-repo/semantics/bachelorThesis |
| dc.type.version.none.fl_str_mv |
info:eu-repo/semantics/acceptedVersion |
| dc.type.coar.none.fl_str_mv |
http://purl.org/coar/resource_type/c_7a1f |
| dc.type.content.es_CO.fl_str_mv |
Text |
| dc.type.redcol.none.fl_str_mv |
http://purl.org/redcol/resource_type/TP |
| format |
http://purl.org/coar/resource_type/c_7a1f |
| status_str |
acceptedVersion |
| dc.identifier.uri.none.fl_str_mv |
http://hdl.handle.net/1992/69749 |
| dc.identifier.instname.es_CO.fl_str_mv |
instname:Universidad de los Andes |
| dc.identifier.reponame.es_CO.fl_str_mv |
reponame:Repositorio Institucional Séneca |
| dc.identifier.repourl.es_CO.fl_str_mv |
repourl:https://repositorio.uniandes.edu.co/ |
| url |
http://hdl.handle.net/1992/69749 |
| identifier_str_mv |
instname:Universidad de los Andes reponame:Repositorio Institucional Séneca repourl:https://repositorio.uniandes.edu.co/ |
| dc.language.iso.es_CO.fl_str_mv |
spa |
| language |
spa |
| dc.relation.references.es_CO.fl_str_mv |
Kaelbling, L. P., Littman, M. L., & Moore, A. W. (1996). Reinforcement learning: A survey. Journal of artificial intelligence research, 4, 237-285. Gym.openai.com. 2016. Gym: A toolkit for developing and comparing reinforcement learning algorithms. [online] Available at: <https://gym.openai.com/docs/> [Accessed 23 December 2021]. Pfau, V., Nichol, A., Hesse, C., Schiavo, L., Schulman, J. and Klimov, O., 2018. Gym Retro. [online] OpenAI. Available at: <https://openai.com/blog/gym-retro/> [Accessed 25 December 2021]. Lipovetzky, N., & Sardina, S. (2018). Pacman capture the flag in AI courses. IEEE Transactions on Games, 11(3), 296-299. Nichol, A., Pfau, V., Hesse, C., Klimov, O., & Schulman, J. (2018). Gotta learn fast: A new benchmark for generalization in rl. arXiv preprint arXiv:1804.03720. Alemán de León, C. D. Agente Sonic. Deep Reinforcement Learning LeBlanc, D. G., & Lee, G. General Deep Reinforcement Learning in NES Games. Bellman, R.E. 1957. Dynamic Programming. Princeton University Press, Princeton, NJ. Republished 2003: Dover Dusparic, I. and Cardozo, N., 2021. ISIS 4222 RL Markov Decision Processes. Dusparic, I. and Cardozo, N., 2021. ISIS 4222 RL Q learning. Watkins, C.J.C.H. (1989). Learning from Delayed Rewards. PhD thesis, Cambridge University, Cambridge, England Sutton, R. S., & Barto, A. G. (2018). Reinforcement learning: An introduction. MIT press. Odemakinde, E., 2022. Model-Based and Model-Free Reinforcement Learning: Pytennis Case Study - neptune.ai. [online] neptune.ai. Available at: <https://neptune.ai/blog/model-based-and-model-free-reinforcement-learning-pytenniscase- study> [Accessed 8 June 2022]. Spinningup.openai.com. 2018. Part 2: Kinds of RL Algorithms Spinning Up documentation. [online] Available at: <https://spinningup.openai.com/en/latest/spinningup/rl_intro2.html> [Accessed 7 June 2022]. Sutton, R. S., & Barto, A. G. (2018). Reinforcement Learning: An Introduction. MIT Press. Watkins, Christopher J. C. H., and Peter Dayan. "Q-learning." Machine learning 8.3-4 (1992): 279-292. Lin, L.-J. Self-improving reactive agents based on reinforcement learning, planning and teaching. Machine learning, 8(3-4):293-321, 1992 Fedus, W., Ramachandran, P., Agarwal, R., Bengio, Y., Larochelle, H., Rowland, M., & Dabney, W. (2020, November). Revisiting fundamentals of experience replay. In International Conference on Machine Learning (pp. 3061-3071). PMLR. Li, Y. (2017). Deep reinforcement learning: An overview. arXiv preprint arXiv:1701.07274. Mnih, V., Kavukcuoglu, K., Silver, D., Graves, A., Antonoglou, I., Wierstra, D., & Riedmiller, M. (2013). Playing atari with deep reinforcement learning. arXiv preprint arXiv:1312.5602. Schulman, J., Wolski, F., Dhariwal, P., Radford, A., & Klimov, O. (2017). Proximal policy optimization algorithms. arXiv preprint arXiv:1707.06347. Schulman, J., Levine, S., Abbeel, P., Jordan, M., & Moritz, P. (2015, June). Trust region policy optimization. In International conference on machine learning (pp. 1889-1897). PMLR. Estes, R. (2020) Rjalnev - DQN, GitHub. Available at: https://github.com/rjalnev (Accessed: 15 December 2021). |
| dc.rights.license.*.fl_str_mv |
Atribución 4.0 Internacional |
| dc.rights.uri.*.fl_str_mv |
http://creativecommons.org/licenses/by/4.0/ |
| dc.rights.accessrights.none.fl_str_mv |
info:eu-repo/semantics/openAccess |
| dc.rights.coar.none.fl_str_mv |
http://purl.org/coar/access_right/c_abf2 |
| rights_invalid_str_mv |
Atribución 4.0 Internacional http://creativecommons.org/licenses/by/4.0/ http://purl.org/coar/access_right/c_abf2 |
| eu_rights_str_mv |
openAccess |
| dc.format.extent.es_CO.fl_str_mv |
38 páginas |
| dc.format.mimetype.es_CO.fl_str_mv |
application/pdf |
| dc.publisher.es_CO.fl_str_mv |
Universidad de los Andes |
| dc.publisher.program.es_CO.fl_str_mv |
Ingeniería de Sistemas y Computación |
| dc.publisher.faculty.es_CO.fl_str_mv |
Facultad de Ingeniería |
| dc.publisher.department.es_CO.fl_str_mv |
Departamento de Ingeniería Sistemas y Computación |
| institution |
Universidad de los Andes |
| bitstream.url.fl_str_mv |
https://repositorio.uniandes.edu.co/bitstreams/2a89aac7-da6d-4dfc-a7ae-9eca6706d611/download https://repositorio.uniandes.edu.co/bitstreams/85277596-d08d-42ad-95a3-52932dadd0ad/download https://repositorio.uniandes.edu.co/bitstreams/c0054a4b-c528-4aa0-8818-2cef305ff9ab/download https://repositorio.uniandes.edu.co/bitstreams/002e7824-9494-4d20-b994-879a75ee4f0a/download https://repositorio.uniandes.edu.co/bitstreams/dc35e395-6b60-4459-9f6d-4ec8c201bfc1/download https://repositorio.uniandes.edu.co/bitstreams/6358ffbe-6804-41c0-941b-6e545f6d581f/download https://repositorio.uniandes.edu.co/bitstreams/aa1fbda6-f678-4b7d-86ae-abcd6846faea/download https://repositorio.uniandes.edu.co/bitstreams/a1a23570-eb36-427a-a7c5-fe0b061e0766/download |
| bitstream.checksum.fl_str_mv |
0463a7172de3715b6e386d4f286a843c 08b106dfeb12472e88207a069e15ba30 0175ea4a2d4caec4bbcc37e300941108 c62b1df50f559b3045a9365fe7d2a76b aba4f89b20fbe3c0c893b84dbfeb954f 5aa5c691a1ffe97abd12c2966efcb8d6 cb23c16a33e0e9a3431201c153092bd2 7bb240e14694b5da4fce5d3a2297a95a |
| bitstream.checksumAlgorithm.fl_str_mv |
MD5 MD5 MD5 MD5 MD5 MD5 MD5 MD5 |
| repository.name.fl_str_mv |
Repositorio institucional Séneca |
| repository.mail.fl_str_mv |
adminrepositorio@uniandes.edu.co |
| _version_ |
1808390398061576192 |
| spelling |
Atribución 4.0 Internacionalhttp://creativecommons.org/licenses/by/4.0/info:eu-repo/semantics/openAccesshttp://purl.org/coar/access_right/c_abf2Takahashi Rodríguez, Silviad87dbcfc-97cc-4baf-8ff3-c9ea6c4fd89d600González Oviedo, Rodrigo Josébbda3ff1-ef5b-43a7-8e91-56fcec364907600Takahashi Rodríguez, Silvia2023-08-16T13:44:18Z2023-08-16T13:44:18Z2023-08-15http://hdl.handle.net/1992/69749instname:Universidad de los Andesreponame:Repositorio Institucional Sénecarepourl:https://repositorio.uniandes.edu.co/El documento se enfoca en hacer una descripción de los algoritmos de reinforcement learning PPO y DQN, los cuales si bien son comparados con el framework Gym-retro de OpenAI (utilizando el juego de NES ice-climbers), cuentan con tantas optimizaciones y estilos que el alcance de las comparaciones es limitado.El presente trabajo explica detalladamente el funcionamiento de los algoritmos de aprendizaje reforzado, DQN y PPO. Adicionalmente se realiza una comparación entre estos algoritmos utilizando el framework de OpenAI Gym-retro, para entrenar un game agent basado en cada uno de los algoritmos.Ingeniero de Sistemas y ComputaciónPregrado38 páginasapplication/pdfspaUniversidad de los AndesIngeniería de Sistemas y ComputaciónFacultad de IngenieríaDepartamento de Ingeniería Sistemas y ComputaciónAnálisis de dos algoritmos de Reinforcement Learning aplicados a OpenAi Gym RetroTrabajo de grado - Pregradoinfo:eu-repo/semantics/bachelorThesisinfo:eu-repo/semantics/acceptedVersionhttp://purl.org/coar/resource_type/c_7a1fTexthttp://purl.org/redcol/resource_type/TPReinforcement learningNintendo entertainment systemGame agentsArtificial intelligenceIceClimberDQNPPOGym retroStable baselinesIngenieríaKaelbling, L. P., Littman, M. L., & Moore, A. W. (1996). Reinforcement learning: A survey. Journal of artificial intelligence research, 4, 237-285.Gym.openai.com. 2016. Gym: A toolkit for developing and comparing reinforcement learning algorithms. [online] Available at: <https://gym.openai.com/docs/> [Accessed 23 December 2021].Pfau, V., Nichol, A., Hesse, C., Schiavo, L., Schulman, J. and Klimov, O., 2018. Gym Retro. [online] OpenAI. Available at: <https://openai.com/blog/gym-retro/> [Accessed 25 December 2021].Lipovetzky, N., & Sardina, S. (2018). Pacman capture the flag in AI courses. IEEE Transactions on Games, 11(3), 296-299.Nichol, A., Pfau, V., Hesse, C., Klimov, O., & Schulman, J. (2018). Gotta learn fast: A new benchmark for generalization in rl. arXiv preprint arXiv:1804.03720.Alemán de León, C. D. Agente Sonic. Deep Reinforcement LearningLeBlanc, D. G., & Lee, G. General Deep Reinforcement Learning in NES Games.Bellman, R.E. 1957. Dynamic Programming. Princeton University Press, Princeton, NJ. Republished 2003: DoverDusparic, I. and Cardozo, N., 2021. ISIS 4222 RL Markov Decision Processes.Dusparic, I. and Cardozo, N., 2021. ISIS 4222 RL Q learning.Watkins, C.J.C.H. (1989). Learning from Delayed Rewards. PhD thesis, Cambridge University, Cambridge, EnglandSutton, R. S., & Barto, A. G. (2018). Reinforcement learning: An introduction. MIT press.Odemakinde, E., 2022. Model-Based and Model-Free Reinforcement Learning: Pytennis Case Study - neptune.ai. [online] neptune.ai. Available at: <https://neptune.ai/blog/model-based-and-model-free-reinforcement-learning-pytenniscase- study> [Accessed 8 June 2022].Spinningup.openai.com. 2018. Part 2: Kinds of RL Algorithms Spinning Up documentation. [online] Available at: <https://spinningup.openai.com/en/latest/spinningup/rl_intro2.html> [Accessed 7 June 2022].Sutton, R. S., & Barto, A. G. (2018). Reinforcement Learning: An Introduction. MIT Press.Watkins, Christopher J. C. H., and Peter Dayan. "Q-learning." Machine learning 8.3-4 (1992): 279-292.Lin, L.-J. Self-improving reactive agents based on reinforcement learning, planning and teaching. Machine learning, 8(3-4):293-321, 1992Fedus, W., Ramachandran, P., Agarwal, R., Bengio, Y., Larochelle, H., Rowland, M., & Dabney, W. (2020, November). Revisiting fundamentals of experience replay. In International Conference on Machine Learning (pp. 3061-3071). PMLR.Li, Y. (2017). Deep reinforcement learning: An overview. arXiv preprint arXiv:1701.07274.Mnih, V., Kavukcuoglu, K., Silver, D., Graves, A., Antonoglou, I., Wierstra, D., & Riedmiller, M. (2013). Playing atari with deep reinforcement learning. arXiv preprint arXiv:1312.5602.Schulman, J., Wolski, F., Dhariwal, P., Radford, A., & Klimov, O. (2017). Proximal policy optimization algorithms. arXiv preprint arXiv:1707.06347.Schulman, J., Levine, S., Abbeel, P., Jordan, M., & Moritz, P. (2015, June). Trust region policy optimization. In International conference on machine learning (pp. 1889-1897). PMLR.Estes, R. (2020) Rjalnev - DQN, GitHub. Available at: https://github.com/rjalnev (Accessed: 15 December 2021).201516582PublicationTEXTAna¿lisis de dos algoritmos de Reinforcement Learning aplicados a OpenAi Gym Retro.pdf.txtAna¿lisis de dos algoritmos de Reinforcement Learning aplicados a OpenAi Gym Retro.pdf.txtExtracted texttext/plain63161https://repositorio.uniandes.edu.co/bitstreams/2a89aac7-da6d-4dfc-a7ae-9eca6706d611/download0463a7172de3715b6e386d4f286a843cMD56autorizacion tesis(2).pdf.txtautorizacion tesis(2).pdf.txtExtracted texttext/plain1161https://repositorio.uniandes.edu.co/bitstreams/85277596-d08d-42ad-95a3-52932dadd0ad/download08b106dfeb12472e88207a069e15ba30MD58CC-LICENSElicense_rdflicense_rdfapplication/rdf+xml; charset=utf-8908https://repositorio.uniandes.edu.co/bitstreams/c0054a4b-c528-4aa0-8818-2cef305ff9ab/download0175ea4a2d4caec4bbcc37e300941108MD52THUMBNAILAna¿lisis de dos algoritmos de Reinforcement Learning aplicados a OpenAi Gym Retro.pdf.jpgAna¿lisis de dos algoritmos de Reinforcement Learning aplicados a OpenAi Gym Retro.pdf.jpgIM Thumbnailimage/jpeg8890https://repositorio.uniandes.edu.co/bitstreams/002e7824-9494-4d20-b994-879a75ee4f0a/downloadc62b1df50f559b3045a9365fe7d2a76bMD57autorizacion tesis(2).pdf.jpgautorizacion tesis(2).pdf.jpgIM Thumbnailimage/jpeg16709https://repositorio.uniandes.edu.co/bitstreams/dc35e395-6b60-4459-9f6d-4ec8c201bfc1/downloadaba4f89b20fbe3c0c893b84dbfeb954fMD59LICENSElicense.txtlicense.txttext/plain; charset=utf-81810https://repositorio.uniandes.edu.co/bitstreams/6358ffbe-6804-41c0-941b-6e545f6d581f/download5aa5c691a1ffe97abd12c2966efcb8d6MD53ORIGINALAna¿lisis de dos algoritmos de Reinforcement Learning aplicados a OpenAi Gym Retro.pdfAna¿lisis de dos algoritmos de Reinforcement Learning aplicados a OpenAi Gym Retro.pdfTrabajo de gradoapplication/pdf1980143https://repositorio.uniandes.edu.co/bitstreams/aa1fbda6-f678-4b7d-86ae-abcd6846faea/downloadcb23c16a33e0e9a3431201c153092bd2MD54autorizacion tesis(2).pdfautorizacion tesis(2).pdfHIDEapplication/pdf296523https://repositorio.uniandes.edu.co/bitstreams/a1a23570-eb36-427a-a7c5-fe0b061e0766/download7bb240e14694b5da4fce5d3a2297a95aMD551992/69749oai:repositorio.uniandes.edu.co:1992/697492023-10-10 18:24:37.18http://creativecommons.org/licenses/by/4.0/open.accesshttps://repositorio.uniandes.edu.coRepositorio institucional Sénecaadminrepositorio@uniandes.edu.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 |