BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Memento EPFL//
BEGIN:VEVENT
SUMMARY:Dissociating curiosity-driven exploration algorithms.
DTSTART:20240502T090000
DTEND:20240502T110000
DTSTAMP:20260407T075521Z
UID:5bf1bafa9e0d0f23c042d030c05e47abd2638a91c406b1b839af1b39
CATEGORIES:Conferences - Seminars
DESCRIPTION:Lucas Gruaz\nEDIC candidacy exam\nExam president: Prof. Martin
  Jaggi\nThesis advisor: Prof. Wulfram Gerstner\nCo-examiner: Prof. Nicolas
  Flammarion\n\nAbstract\nExploration is a fundamental concept both in Rein
 forcement\nLearning (RL) and human behavior. In RL\, exploration\ninvolves
  the agent actively seeking information about\nits environment to discover
  optimal strategies for maximizing\nrewards. Similarly\, in human behavior
 \, exploration manifests\nas curiosity\, experimentation\, and risk-taking
 \, all of which\ncontribute to learning and adaptation. By exploring the u
 nknown\,\nboth RL agents and humans can discover novel solutions\, adapt\n
 to changing circumstances\, and ultimately improve their performance\nand 
 understanding of the world around them. Various\nmethods have been develop
 ed to encourage exploration in RL\nagents. Their specificity and applicati
 on scenarios are varied\, and\ntheir similarity with exploration strategie
 s observed in humans\nremains unclear. In this proposal\, we review three 
 paper related\nto this question. The first paper gives an overview of expl
 oration\ntechniques in deep RL\, the second paper shows an example of\na s
 uccessful application of such techniques\, and the third paper\nexplore hu
 man exploratory behavior\, serving as a starting point\nto assess differen
 ces with previously introduced techniques.\n\nBackground papers\n\n	Paper 
 1: Ladosz et al. “Exploration in deep reinforcement learning: A survey
 ”. Information Fusion\, 85:1–22\, Sept. 2022. https://www.sciencedirec
 t.com/science/article/pii/S1566253522000288\n	Paper 2: Badia et al. “Ag
 ent57: Outperforming the Atari Human Benchmark”. In Proceedings of the 3
 7th International Conference on Machine Learning\, pages 507–517. PMLR\,
  Nov. 2020. https://arxiv.org/abs/2003.13350\n	Paper 3: Brändle et al. 
 “Empowerment contributes to exploration behaviour in a creative video ga
 me”. Nature Human Behaviour\, 7(9):1481–1489\, Sept. 2023. https://www
 .nature.com/articles/s41562-023-01661-2\n
LOCATION:SV 2615 https://plan.epfl.ch/?room==SV%202615
STATUS:CONFIRMED
END:VEVENT
END:VCALENDAR