Rainbow: Combining Improvements in Deep Reinforcement Learning
- Matteo Hessel, Joseph Modayil, David Silver
- Computer ScienceAAAI Conference on Artificial Intelligence
- 6 October 2017
This paper examines six extensions to the DQN algorithm and empirically studies their combination, showing that the combination provides state-of-the-art performance on the Atari 2600 benchmark, both in terms of data efficiency and final performance.
Horde: a scalable real-time architecture for learning knowledge from unsupervised sensorimotor interaction
- R. Sutton, Joseph Modayil, Doina Precup
- Computer ScienceAdaptive Agents and Multi-Agent Systems
- 2 May 2011
Results using Horde on a multi-sensored mobile robot to successfully learn goal-oriented behaviors and long-term predictions from off-policy experience are presented.
Vector-based navigation using grid-like representations in artificial agents
These findings show that emergent grid-like representations furnish agents with a Euclidean spatial metric and associated vector operations, providing a foundation for proficient navigation, and support neuroscientific theories that see grid cells as critical for vector-based navigation.
Deep Reinforcement Learning and the Deadly Triad
- H. V. Hasselt, Yotam Doron, Florian Strub, Matteo Hessel, Nicolas Sonnerat, Joseph Modayil
- Computer ScienceArXiv
- 6 December 2018
This work investigates the impact of the deadly triad in practice, in the context of a family of popular deep reinforcement learning models - deep Q-networks trained with experience replay - analysing how the components of this system play a role in the emergence of the Deadly triad, and in the agent's performance.
Local metrical and global topological maps in the hybrid spatial semantic hierarchy
- B. Kuipers, Joseph Modayil, P. Beeson, M. MacMahon, F. Savelli
- Computer ScienceIEEE International Conference on Robotics and…
- 7 June 2004
This work describes how a local perceptual map is analyzed to identify a local topology description and is abstracted to a topological place and creates a set of topological map hypotheses that are consistent with travel experience.
Factoring the Mapping Problem: Mobile Robot Map-building in the Hybrid Spatial Semantic Hierarchy
This paper describes how to abstract a symbolic description of the robot’s immediate surround from local metrical models, how to combine these local symbolic models in order to build global symbolic models, and how to create a globally consistent metrical map from a topological skeleton by connecting local frames of reference.
Multi-timescale nexting in a reinforcement learning robot
This paper presents results with a robot that learns to next in real time, making thousands of predictions about sensory input signals at timescales from 0.1 to 8 seconds, and extends nexting beyond simple timescale by letting the discount rate be a function of the state.
Improving the recognition of interleaved activities
The Interleaved Hidden Markov Models for recognizing multitasked activities are introduced and an approximation that is both effective and efficient is described that significantly reduces the error rate.
Bootstrap learning for object discovery
- Joseph Modayil, B. Kuipers
- Computer ScienceIEEE/RJS International Conference on Intelligent…
- 28 September 2004
This work shows how a robot can autonomously learn an ontology of objects to explain aspects of its sensor input from an unknown dynamic world, and makes it possible for the robot to describe a cluttered dynamic world with symbolic object descriptions along with a static environment model, both models grounded in sensory experience, and learned without external supervision.