Child: A First Step Towards Continual Learning

Ring, Mark

doi:10.1007/978-1-4615-5529-2_11

Cited by 91 publications

(88 citation statements)

References 10 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…One traditional goal in reinforcement learning is for agents to continually improve their performance as they obtain more data (Hutter, 2005;Ring, 1997;Singh, Barto, & Chentanez, 2004;Sutton et al, 2011;Thrun & Mitchell, 1993;Wilson, 1985). Measuring the extent to which this is the case for a given agent can be a challenge, and this challenge is exacerbated in the Arcade Learning Environment, where the agent is evaluated across 60 games.…”

Section: Summarizing Learning Performancementioning

confidence: 99%

Revisiting the Arcade Learning Environment: Evaluation Protocols and Open Problems for General Agents

Machado¹,

Bellemare²,

Talvitie³

et al. 2018

jair

213

201

View full text Add to dashboard Cite

The Arcade Learning Environment (ALE) is an evaluation platform that poses the challenge of building AI agents with general competency across dozens of Atari 2600 games. It supports a variety of different problem settings and it has been receiving increasing attention from the scientific community, leading to some high-profile success stories such as the much publicized Deep Q-Networks (DQN). In this article we take a big picture look at how the ALE is being used by the research community. We show how diverse the evaluation methodologies in the ALE have become with time, and highlight some key concerns when evaluating agents in the ALE. We use this discussion to present some methodological best practices and provide new benchmark results using these best practices. To further the progress in the field, we introduce a new version of the ALE that supports multiple game modes and provides a form of stochasticity we call sticky actions. We conclude this big picture look by revisiting challenges posed when the ALE was introduced, summarizing the state-of-the-art in various problems and highlighting problems that remain open.

show abstract

Section: Summarizing Learning Performancementioning

confidence: 99%

Revisiting the Arcade Learning Environment: Evaluation Protocols and Open Problems for General Agents

Machado¹,

Bellemare²,

Talvitie³

et al. 2018

jair

213

201

View full text Add to dashboard Cite

show abstract

“…This is different to other lifelong learning approaches that use explanation-based neural networks (EBNN) [12], backpropagation neural networks [13], [14] and reinforcement learning approaches [15]. Another important difference is that here we do not work with a task-to-task mapping as in transfer learning.…”

Section: Related Workmentioning

confidence: 95%

A Knowledge Growth and Consolidation Framework for Lifelong Machine Learning Systems

Plumed¹,

Ferri²,

Orallo³

et al. 2014

2014 13th International Conference on Machine Learning and Applications

View full text Add to dashboard Cite

“… CHILD, a developmental learning approach in which learning takes place with the continual process. Learning paradigm is ANN supervised reinforced [40], combination of Q learning and temporal transition hierarchies, with no knowledge at the beginning. System creates knowledge units while acting on the environment continuously.…”

Section: Psychology and Roboticsmentioning

confidence: 99%

“…But the system has limitation for the states of particular sequence. Although system shows effective and fast learning in Maze problem [40], however its learning is environment dependent.  GRASP project is an attempt to design cognitive grasp capability based on novel situations.…”

Section: Psychology and Roboticsmentioning

confidence: 99%

Human Brain inspired Artificial Intelligence & Developmental Robotics: A Review

Kumar

Sha

2017

SJCMS

View full text Add to dashboard Cite

Along with the developments in the field of the robotics, fascinating contributions and developments can be seen in the field of Artificial intelligence (AI). In this paper we will discuss about the developments is the field of artificial intelligence focusing learning algorithms inspired from the field of Biology, particularly large scale brain simulations, and developmental Psychology. We will focus on the emergence of the Developmental robotics and its significance in the field of AI.

show abstract

Child: A First Step Towards Continual Learning

Cited by 91 publications

References 10 publications

Revisiting the Arcade Learning Environment: Evaluation Protocols and Open Problems for General Agents

Revisiting the Arcade Learning Environment: Evaluation Protocols and Open Problems for General Agents

A Knowledge Growth and Consolidation Framework for Lifelong Machine Learning Systems

Human Brain inspired Artificial Intelligence & Developmental Robotics: A Review

Contact Info

Product

Resources

About