hpi-sam Robust-Multi-Agent-Reinforcement-Learning-for-SAS issues

hpi-sam / Robust-Multi-Agent-Reinforcement-Learning-for-SAS

Research project on robust multi-agent reinforcement learning (marl) for self-adaptive systems (sas)

MIT License

0 stars 0 forks source link

issues

Newest

Newest Most commented Recently updated Oldest Least commented Least recently updated

Perform experiments to evaluate agents and learning

#71 caustt opened 2 years ago
0
Keep mRUBIS environment the same

#70 caustt opened 2 years ago
0
create baseline model

#69 caustt opened 2 years ago
0
Future Work - Discuss the impact of each future work feature on the current architecture.

#68 christianadriano opened 2 years ago
0
Move current Future Work to Report

#67 christianadriano opened 2 years ago
0
Future Work - allow for non-stationary utility

#66 christianadriano opened 2 years ago
1
Future Work - allow multiple failures on the same shop

#65 christianadriano opened 2 years ago
1
Future Work - Perfect Failure Masking Phenomenon

#64 christianadriano opened 2 years ago
1
Preliminaries/Definitions: Self-Adaptive Systems.

#63 christianadriano opened 2 years ago
0
Future Work - Add Multi-Armed Bandits Approach to Rank Learner

#62 florenceboettger opened 2 years ago
0
Future Work - Enable asynchronous fault injection for shops managed by distinct agents.

#61 christianadriano opened 2 years ago
1
synchronize run variable between mrubis and agents

#60 caustt opened 2 years ago
0
Write structure for section 7 threats to validity

#59 ulibath opened 2 years ago
0
prepare presentation for friday

#58 jocodeone closed 2 years ago
0
Switch for different configs in data generator

#57 ulibath opened 2 years ago
0
Collect test data to show need of robustness feature

#56 ulibath opened 2 years ago
0
Add README and Makefile for python

#55 ulibath opened 2 years ago
0
Robustness Feature for Agents

#54 ulibath closed 2 years ago
0
Rework mRUBiS to work with MARL

#53 florenceboettger closed 2 years ago
0
MAPE-K definitions and references

#52 christianadriano opened 2 years ago
3
add probability for non causing issues in trace to be left out for the perturbations

#51 caustt opened 2 years ago
0
Characterize the Lack of Robustness phenomenon

#50 jocodeone opened 2 years ago
2
Implement robustness feature

#49 jocodeone opened 2 years ago
0
check parameters of robustness for variance of utility and speed of trend detection

#48 jocodeone opened 2 years ago
0
Future work - NN to train rank learner under total utility loss

#47 christianadriano opened 2 years ago
1
Add socket communicator

#46 florenceboettger closed 2 years ago
0
add predicitng utility on action level and sorting on rank learner level

#45 jocodeone closed 2 years ago
0
impossible 100% accuracy trade-off false negative and false positive

#44 jocodeone opened 2 years ago
0
Future Work - How to consider uncertainty in the utility when ranking actions (probabilities if this component is correct)

#43 ulibath opened 2 years ago
0
Write about malicious agents (learn to not take an action)

#42 ulibath opened 2 years ago
0
Test learning with separate dense layers or more

#41 ulibath closed 2 years ago
0
Automate test data generation

#40 ulibath closed 2 years ago
0
Implement switch of HMM for mock data to test transfer learning

#39 jocodeone closed 2 years ago
0
WIP: test framework python

#38 ulibath closed 2 years ago
0
implement sorting of actions on the agent level

#37 jocodeone closed 2 years ago
0
rethink of reducing the reward if the agent is not fixing the component immediately

#36 ulibath opened 2 years ago
0
add more difficult test data

#35 ulibath closed 2 years ago
0
finish miro board reward flow

#34 ulibath opened 2 years ago
0
Cost of rules should not be the same for given <shop, component, failure>

#33 christianadriano opened 2 years ago
0
Issue are remaining in the model

#32 christianadriano opened 2 years ago
0
Adapt the Injection-Strategy for new Communication

#31 caustt opened 2 years ago
0
Create new Layer between mRUBIS and the python controller

#30 caustt opened 2 years ago
0
add stats and saving/loading of model

#29 jocodeone closed 2 years ago
0
Feature/5 save and load agent models

#28 jocodeone closed 2 years ago
0
Adapt mRUBiS for the new python handler

#27 florenceboettger closed 2 years ago
0
Create socket methods for communicating big JSONs

#26 florenceboettger closed 2 years ago
0
Create sequence diagram for old and new program flow

#25 florenceboettger closed 2 years ago
0
Feature/4 agents start learning

#24 jocodeone closed 2 years ago
0
Review threats to validity

#23 jocodeone opened 2 years ago
0
Review the literature for the report

#22 jocodeone opened 2 years ago
0