Deep reinforcement learning at the edge of the statistical precipice

R Agarwal, M Schwarzer, PS Castro… - Advances in neural …, 2021‏ - proceedings.neurips.cc
Deep reinforcement learning (RL) algorithms are predominantly evaluated by comparing
their relative performance on a large suite of tasks. Most published results on deep RL …

A review of reinforcement learning in chemistry

S Gow, M Niranjan, S Kanza, JG Frey - Digital Discovery, 2022‏ - pubs.rsc.org
The growth of machine learning as a tool for research in computational chemistry is well
documented. For many years, this growth was heavily driven by the paradigms of supervised …

Language as a cognitive tool to imagine goals in curiosity driven exploration

C Colas, T Karch, N Lair, JM Dussoux… - Advances in …, 2020‏ - proceedings.neurips.cc
Developmental machine learning studies how artificial agents can model the way children
learn open-ended repertoires of skills. Such agents need to create and represent goals …

Measuring the reliability of reinforcement learning algorithms

SCY Chan, S Fishman, J Canny, A Korattikara… - arxiv preprint arxiv …, 2019‏ - arxiv.org
Lack of reliability is a well-known issue for reinforcement learning (RL) algorithms. This
problem has gained increasing attention in recent years, and efforts to improve it have …

Exploring safer behaviors for deep reinforcement learning

E Marchesini, D Corsi, A Farinelli - … of the AAAI Conference on Artificial …, 2022‏ - ojs.aaai.org
Abstract We consider Reinforcement Learning (RL) problems where an agent attempts to
maximize a reward signal while minimizing a cost function that models unsafe behaviors …

A domain-agnostic approach for characterization of lifelong learning systems

MM Baker, A New, M Aguilar-Simon, Z Al-Halah… - Neural Networks, 2023‏ - Elsevier
Despite the advancement of machine learning techniques in recent years, state-of-the-art
systems lack robustness to “real world” events, where the input distributions and tasks …

Grounding language to autonomously-acquired skills via goal generation

A Akakzia, C Colas, PY Oudeyer, M Chetouani… - arxiv preprint arxiv …, 2020‏ - arxiv.org
We are interested in the autonomous acquisition of repertoires of skills. Language-
conditioned reinforcement learning (LC-RL) approaches are great tools in this quest, as they …

Semantic helm: A human-readable memory for reinforcement learning

F Paischer, T Adler, M Hofmarcher… - Advances in Neural …, 2023‏ - proceedings.neurips.cc
Reinforcement learning agents deployed in the real world often have to cope with partially
observable environments. Therefore, most agents employ memory mechanisms to …

Universal morphology control via contextual modulation

Z **ong, J Beck, S Whiteson - International Conference on …, 2023‏ - proceedings.mlr.press
Learning a universal policy across different robot morphologies can significantly improve
learning efficiency and generalization in continuous control. However, it poses a challenging …

Deep reinforcement learning for navigation in AAA video games

E Alonso, M Peter, D Goumard, J Romoff - arxiv preprint arxiv:2011.04764, 2020‏ - arxiv.org
In video games, non-player characters (NPCs) are used to enhance the players' experience
in a variety of ways, eg, as enemies, allies, or innocent bystanders. A crucial component of …