Advanced ML

Fall 2019    UTD

Course website:    .../~sriraam.natarajan/Courses/…

Upload scribes of paper discussions here.

Reading List

Date

Paper

Owner

Scribe

08/22

Natarajan, Sriraam, and Prasad Tadepalli. "Dynamic preferences in multi-criteria reinforcement learning." ICML 2005.

Brian

Srijita

Athresh

Arjun

08/29

Tadepalli, Prasad, Robert Givan, and Kurt Driessens. "Relational reinforcement learning: An overview." ICML-2004 workshop on relational reinforcement learning. 2004.

Harsha

Dev

Brian

Athresh

08/29

Ponsen, Marc, et al. "Learning with whom to communicate using relational reinforcement learning." Interactive Collaborative Information Systems. Springer, Berlin, Heidelberg, 2010.

Sagnik

Nandini

Omeed

Vignesh

09/05

Dietterich, Thomas G. "The MAXQ Method for Hierarchical Reinforcement Learning." ICML. 1998.

Mike

Omeed

Srijita

Dev

09/05

Sutton, R.S., Precup, D., Singh, S. (1998). Intra-option learning about temporally abstract actions. Proceedings of the 15th International Conference on Machine Learning

Siwen

Vignesh

Justin

Arjun

09/12

C. Wang, S. Joshi and R. Khardon (2007) First Order Decision Diagrams for Relational MDPs

Navdeep

Arjun

Yuqiao

Mike

09/19
(updated)

K.Kersting and K. Driessens, Non–parametric policy gradients: A unified treatment of propositional and relational domains. ICML 2008

Mike

Yuqiao

Navdeep

Siwen

09/19

Guiding autonomous systems towards better behavior using human advice

Arjun

Justin

Nandini

Omeed

09/26
(updated)

Minimal Sufficient Explanations for Factored Markov Decision Processes,

Athresh

Arjun

Harsha

Srijita

09/26

Zambaldi et al., Deep reinforcement learning with relational inductive biases, 2019.

Srijita

Harsha

Athresh

Sagnik

10/03

 Natarajan et al., Multi-agent  IRL

Nandini

Sagnik

Yibo

Navdeep

10/03

Odom and Natarajan, Active advice seeking for IRL 2016

Yibo

Omeed

Harsha

Brian

10/10

Where to add actions in human-in-the-loop reinforcement learning

Navdeep

Brian

Nandini

Dev

10/10

Combining manual feedback with subsequent mdp reward signals for reinforcement learning

Dev

Vignesh

Sagnik

Yibo

10/17

Evaluation of interactive machine learning

Navdeep

Sagnik

Srijita

Athresh

10/17

Interactive Machine Learning for Health Informatics> When do we need HIL?

Harsha

Mike

Omeed

Dev

10/24

Towards accountable reinforcement learning

Athresh

Siwen

Mike

Vignesh

10/24

Exploration in action space

Srijita

Dev

Yibo

Sagnik

10/31

Agent Agnostic HIL RL

Nandini

Vignesh

Siwen

Harsha

10/31

Interaction in NLP

Omeed

Yibo

Athresh

Navdeep

Project

S.No.

Team members

Project topic

1

Brian Ricks and Justin Dula

Using a framework to act as a simulator for computer networks, we plan to study scenarios using multi-criteria agents, agent-side weight changes, and an adversary that can also manipulate weights.

2

Sagnik Dakshit and <Yibo>

I work in the Multimedia Lab. My lab has two VR games for Education and Rehabilitation. The first one is a plant walk scene used both for rehabilitation and education and the other game is for rehabilitation of Phantom pain patient. . I will be happy to arrange a demo for anyone interested. I have the data but the idea for implementation of RL is not consolidated yet. We can explore difficulty level adjustment, Quality of experience or more.

3

Harsha Kokel, Omeed Ashtiani, Mike Skinner

To be added

4

Yuqiao Chen, Navdeep Kaur, Siwen Yan

To be added

5

Athresh, Vignesh

To be added

6

Devendra, Srijita

To be added (May be Fairness)

7

Nandini Ramanan

TBA

Project Presentations - Initial ideas, papers read and proposed roadmap (20 minutes each and 10 minutes for questions)

October 15 - Skinner et al,

October 22 - Dakshit and Yibo,Ricks and Dula

October 29 - SN Away (PI Meeting) -- RL background Vignesh

November 5 - Chen et al, Athresh and Vignesh

November 12 - Srijita and Devendra, Nandini,