Sparse multi-task reinforcement learning

By Daniele Calandriello, Alessandro Lazaric, Marcello Restelli. 2014

80 citations recorded by the source corpus.

Subfields: Computational Mechanics, Computational Theory and Mathematics, Management Science and Operations Research, Reinforcement Learning

How researchers see this paper

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

In Computational Mechanics

1. Qiang Yang
2. Yu Zhang
3. John Shawe-Taylor
4. Doina Precup
5. Joelle Pineau
6. Thore Graepel
7. Parameswaran Kamalaruban
8. Renjie Liao

In Computational Theory and Mathematics

1. Amy Zhang
2. Joelle Pineau
3. Khimya Khetarpal
4. Matthew Riemer
5. John Shawe-Taylor
6. Doina Precup
7. Qiang Yang
8. Thore Graepel

In Management Science and Operations Research

1. Sanja Fidler
2. Joelle Pineau
3. Doina Precup
4. Qiang Yang
5. Renjie Liao
6. Sergey Levine
7. Shagun Sodhani
8. Matthew Riemer

In Reinforcement Learning

1. Sergey Levine
2. Joelle Pineau
3. Pieter Abbeel
4. Jimmy Ba
5. Matthew Riemer
6. Doina Precup
7. Amy Zhang
8. Khimya Khetarpal

Citations in the corpus

• An Overview of Multi-task Learning (2018, methodology): “In [72], the value functions in different tasks are assumed to share sparse parameters and it applies the multi-task feature selection method with the (cid:2) 2,1 regularization [8] and the MTFL method [5]tolearnallthevaluefunctionssimultaneously.”
• A Survey on Multi-Task Learning (2017, background): “Some works [141], [142], [143], [144], [145], [146], [147], [148], [149], [150] adapt the ideas introduced in Section 2 to MRL.”
• Towards Continual Reinforcement Learning: A Review and Perspectives (2020, background): “Leveraging the assumption of shared common structure between tasks, Borsa, Graepel, and Shawe-Taylor (2016) build on the work of Calandriello, Lazaric, and Restelli (2014) and explicitly model a shared abstraction of the state-action space.”
• Applications of reinforcement learning in energy systems (2021, methodology): “By assuming additional structure on the similarity be-tween the tasks, people have proposed more efficient optimization algorithms for Multi-task RL with provable guarantees [191].”
• Online Multitask Relative Similarity Learning (2017, unknown): “The implicit assumption is that these tasks are re-lated and the labeled data for each task is limited [Hoi et al. , 2014; Bakker and Heskes, 2003; Calandriello et al. , 2014; Cohen and Crammer, 2014; Zhang et al. , 2016 ] .”
• Learning Shared Representations in Multi-task Reinforcement Learning (2016, methodology): “Learning share representations for value functions in Multi-Task RL Using a multi-task RL formulation and following the recent work done in (Calandriello et al., 2014), we firstly outline two general ways of learning RL tasks jointly and sharing knowledge across them by extending two of the most pop”
• Multi-Task Reinforcement Learning with Context-based Representations (2021, background): “(Calandriello et al., 2014; Borsa et al., 2016; Maurer et al., 2016) assume that in the multi-task settings, tasks share a common, low dimensional representation, and therefore advocate for learning a shared representation space across all the tasks.”
• NerveNet: Learning Structured Policy with Graph Neural Networks (2018, methodology): “In Ammar et al. (2014), a multi-task policy gradient is proposed, while, Calandriello et al. (2014) propose multi-task extensions of the fitted Q-iteration algorithm.”