DeepMind Lab

By Charlie Beattie, Joel Z. Leibo, Denis Teplyashin, Tom Ward, Marcus Wainwright, Heinrich Küttler, Andrew Lefrancq, Simon Green, V. Valdés, Amir Sadik and 11 more. 2016

267 citations recorded by the source corpus.

DeepMind Lab is a first-person 3D game platform designed for research and development of general artificial intelligence and machine learning systems. DeepMind Lab can be used to study how autonomous artificial agents may learn complex tasks in large, partially observed, and visually diverse worlds. DeepMind Lab has a simple and flexible API enabling creative task-designs and novel AI-designs to be explored and quickly iterated upon. It is powered by a fast and widely recognised game engine, and tailored for effective use by the research community.

Subfields: Artificial Intelligence, Information Systems and Management, Reinforcement Learning, Evaluation & Benchmarks

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Artificial Intelligence

In Information Systems and Management

In Reinforcement Learning

In Evaluation & Benchmarks

Citations in the corpus

Visual Prompt Tuning (2022, neutral): “800/1000 200 15,000 Clevr/distance [34] 6 15,000 DMLab [3] 6 22,735 KITTI/distance [20] 4 711 dSprites/location [53] 16 73,728 dSprites/orientation [53] 16 73,728 SmallNORB/azimuth [40] 18 12,150 SmallNORB/elevation [40] 9 12,150”
Off-Policy Actor-Critic with Shared Experience Replay (2019, background): “As a result, we present state-of-the-art data efﬁciency in Section 5 in terms of median human normalized performance across 57 Atari games (Bellemare et al., 2013), as well as improved learning efﬁciency on DMLab30 (Beattie et al., 2016) (Table 1).”
Revisiting the Arcade Learning Environment: Evaluation Protocols and Open Problems for General Agents (2017, background): “The lessons learned from that experience may help further that progress and may inform best practices as other testbeds for general competency are developed (e.g., Beattie et al., 2016; Brockman et al., 2016; Johnson, Hofmann, Hutton, & Bignell, 2016; Levine et al., 2013).”
Embodied Question Answering (2017, background): “Deep-Mind Lab [34] or Doom [16]), to more complex, realistic environments ( e.g .”
A Survey of Embodied AI: From Simulators to Research Tasks (2021, background): “DeepMind Lab [18] is the first proof-of-concept of an embodied AI simulator.”
Life-Long Disentangled Representation Learning with Cross-Domain Latent Homologies (2018, methodology): “Hence, we trained VASE on a sequence of two visually challenging DMLab-30 1 [7] datasets: the Exploit Deferred Effects (EDE) environment and a randomized version of the Natural Labyrinth (NatLab) environment (Varying Map Randomized).”
Google Research Football: A Novel Reinforcement Learning Environment (2019, neutral): “The same applies to the DeepMind Lab (Beattie et al. 2016), a navigation-focused maze environment that provides a number of relatively simple tasks with a ﬁrst person viewpoint.”
Learning to Navigate in Complex Environments (2016, methodology): “We consider a set of ﬁrst-person 3D mazes from the DeepMind Lab environment (Beattie et al., 2016) (see Fig.”
AI2-THOR: An Interactive 3D Environment for Visual AI (2017, background): “DeepMind Lab [3] X X VizDoom [13] X X MINOS (Matterport3D) [17] X X MINOS (SUNCG) [17] X X X House3D [2] X X X HoME [5] X X X X AI2-THOR X X X X X Table 1.”
Multi-task Deep Reinforcement Learning with PopArt (2018, neutral): “(Beattie et al. 2016).”
MINOS: Multimodal Indoor Simulator for Navigation in Complex Environments (2017, background): “VizDoom (Kempka et al., 2016) and DeepMind Lab (Beattie et al., 2016) simulate stylized immersive three-dimensional labyrinths.”
State Representation Learning for Control: An Overview (2018, background): “Many of the latter simulated scenarios are part of Universe and OpenAI Gym [ Brockman et al., 2016] or DeepMind Labs [Beattie et al., 2016].”
Look Before You Leap: Bridging Model-Free and Model-Based Reinforcement Learning for Planned-Ahead Vision-and-Language Navigation (2018, background): “Although steady progress has been made on natural language command of robots [2,3,4,5], it is still far from perfect.”
An Introduction to Deep Reinforcement Learning (2018, unknown): “Similarly, Deepmind Lab (Beattie et al., 2016) provides a 3D platform adapted from the Quake video game.”