1st Workshop on Transfer in Reinforcement Learning (TiRL)

Schedule

Time	Program
8h00 – 9h00	Registration
9h00 – 9h10	Opening
9h10 - 10h00	Invited talk - Next Steps in Transfer and Lifelong Learning Matthew E. Taylor
Session 1	Chair: Anna Reali
10h00 – 10h10	Transfer Learning Through Graph-based Skill Acquisition (highlight) Farzaneh Shoeleh and Masoud Asadpour
10h10 – 10h20	Reinforcement Learning for Multi-Step Expert Advice (highlight) Patrick Philipp and Achim Rettinger
10h20 – 10h30	Inverse Reinforcement Learning in Swarm Systems (highlight) Adrian Šošić, Wasiur R. Khudabukhsh, Abdelhak Zoubir, and Heinz Koeppl
10h30 – 10h50	Towards Zero-Shot Autonomous Inter-Task Mapping through Object-Oriented Task Description (full) Felipe Leno Da Silva and Anna Helena Reali Costa
10h50 – 11h00	Poster session I
11h00 – 11h30	Coffee break
Session 2	Chair: Felipe Leno da Silva
11h30 – 11h50	Towards a fast detection of opponents in repeated stochastic games (full) Pablo Hernandez-Leal and Michael Kaisers
11h50 – 12h10	Feature Selection for Reinforcement Learning by Learning Process Evaluation (full) Cleiton Alves da Silva and Valdinei Freire
12h10 – 12h15	Case-based Policy Inference (short) Ruben Glatt, Felipe Leno Da Silva, and Anna Helena Reali Costa
12h15 – 12h20	Automatic Skill Transfer Learning Through Domain Adaptation (short) Farzaneh Shoeleh and Masoud Asadpour
12h20 – 12h25	Knowledge Transfer from a Human Perspective (short) Ramya Ramakrishnan and Julie Shah
12h25 – 12h30	Transferring Probabilistic Options in Reinforcement Learning (short) Rodrigo Cesar Bonini, Felipe Leno Da Silva, Ruben Glatt and Anna Helena Reali Costa
12h30 – 13h00	Poster session II
12h30 - 14h30	Lunch break
14h30 – 15h45	Invited talk: Using cases as heuristics in Tranfer Learning Reinaldo Bianchi
Session 3	Chair: Ruben Glatt
15h45 – 16h25	Causal inference and Reinforcement Learning Counterfactual Data-Fusion for Online Reinforcement Learners (full) Transfer Learning in Multi-Armed Bandit: A Causal Approach (full) Elias Bareinboim
16h25 – 16h30	Poster session III
16h30 – 17h00	Coffee break
17h00 – 18h15	Invited talk: Curriculum Construction for RL Agents Jivko Sinapov
18h15 - 18h30	Closing / Community Meeting / Integration

Accepted Papers - Full paper

• Towards Zero-Shot Autonomous Inter-Task Mapping through Object-Oriented Task Description
  Felipe Leno Da Silva and Anna Helena Reali Costa, Escola Politécnica of the University of São Paulo, Brazil.
• Towards a fast detection of opponents in repeated stochastic games
  Pablo Hernandez-Leal and Michael Kaisers, Centrum Wiskunde & Informatica, Netherlands.
• Transfer Learning in Multi-Armed Bandit: A Causal Approach
  Junzhe Zhang and Elias Bareinboim, Purdue University, USA.
• Counterfactual Data-Fusion for Online Reinforcement Learners
  Andrew Forney, Judea Pearl and Elias Bareinboim, Purdue University, USA.
• Feature Selection for Reinforcement Learning: Transfer knowledge assessed in the learning process
  Cleiton Alves Da Silva and Valdinei Freire, University of São Paulo (EACH), Brazil.

Accepted Papers - Short paper

• Automatic Skill Transfer Learning Through Domain Adaptation
  Farzaneh Shoeleh and Masoud Asadpour, University of Tehran, Iran.
• Knowledge Transfer from a Human Perspective
  Ramya Ramakrishnan and Julie Shah, Massachusetts Institute of Technology, USA.
• Case-based Policy Inference
  Ruben Glatt, Felipe Leno Da Silva and Anna Helena Reali Costa, Escola Politécnica of the University of São Paulo, Brazil.
• Transferring Probabilistic Options in Reinforcement Learning
  Rodrigo Cesar Bonini, Felipe Leno Da Silva, Ruben Glatt and Anna Helena Reali Costa, Escola Politécnica of the University of São Paulo, Brazil.

Accepted Papers - Highlight paper

• Transfer Learning Through Graph-based Skill Acquisition
  Farzaneh Shoeleh and Masoud Asadpour, University of Tehran, Iran.
  (Full paper: http://www.sciencedirect.com/science/article/pii/S0167865516302112)
• Inverse Reinforcement Learning in Swarm Systems
  Adrian Šošić, Wasiur R. Khudabukhsh, Abdelhak Zoubir and Heinz Koeppl, Technische Universität Darmstadt, Germany.
  (Full paper: Proceedings of AAMAS-17)
• Reinforcement Learning for Multi-Step Expert Advice
  Patrick Philipp and Achim Rettinger, Karlsruhe Institute of Technology, Germany.
  (Full paper: Proceedings of AAMAS-17)