Mostra el registre d'ítem simple

dc.contributorMartín Muñoz, Mario
dc.contributorGarcia Gasulla, Dario
dc.contributor.authorHeidecke, Johannes
dc.contributor.otherUniversitat Politècnica de Catalunya. Departament de Ciències de la Computació
dc.date.accessioned2019-04-11T09:17:00Z
dc.date.available2019-04-11T09:17:00Z
dc.date.issued2019-01-15
dc.identifier.urihttp://hdl.handle.net/2117/131625
dc.description.abstractWe evaluate the robustness of reward functions learned with IRL, when transferred to similar tasks. We exceed state of the art results for one benchmark task and solve another one for the first time. Modifications are proposed that achieve faster and more stable training.
dc.language.isoeng
dc.publisherUniversitat Politècnica de Catalunya
dc.subjectÀrees temàtiques de la UPC::Informàtica
dc.subject.lcshReinforcement learning
dc.subject.lcshAlgorithms
dc.subject.otherinverse reinforcement learning
dc.subject.otherIRL
dc.subject.otherreinforcement learning
dc.subject.otherRL
dc.subject.otherguided cost learning
dc.subject.otherGCL
dc.subject.otheradversarial inverse reinforcement learning
dc.subject.otherAIRL
dc.subject.othersoft actor critic
dc.subject.otherSAC
dc.subject.othertransfer learning
dc.subject.otherrobustness
dc.subject.othermaximum entropy principle
dc.subject.othermaximum causal entropy principle
dc.subject.otherreward shaping
dc.subject.otherpre-training
dc.subject.othermetric
dc.subject.othershaped reward loss
dc.subject.otherpendulum
dc.subject.otherlunar lander
dc.titleEvaluating the Robustness of GAN-Based Inverse Reinforcement Learning Algorithms
dc.typeMaster thesis
dc.subject.lemacAprenentatge per reforç
dc.subject.lemacAlgorismes
dc.identifier.slug134533
dc.rights.accessOpen Access
dc.date.updated2019-02-04T05:00:43Z
dc.audience.educationlevelMàster
dc.audience.mediatorFacultat d'Informàtica de Barcelona
dc.audience.degreeMÀSTER UNIVERSITARI EN INTEL·LIGÈNCIA ARTIFICIAL (Pla 2017)
dc.contributor.covenanteeUniversitat de Barcelona
dc.contributor.covenanteeUniversitat Rovira i Virgili


Fitxers d'aquest items

Thumbnail

Aquest ítem apareix a les col·leccions següents

Mostra el registre d'ítem simple