This paper tackles the problem of active perception: taking actions to minimize one’s uncertainty. It further formalizes the link between information gain and prediction rewards, and uses this to propose a deep-learning approach to optimize active perception from a data set, thus obviating the need for a complex POMDP model.
Category: news
IJCAI paper: Decentralized MCTS via Learned Teammate Models
Aleksander Czechowski got his paper on Decentralized MCTS via Learned Teammate Models accepted at IJCAI 2020.
In this paper we learn the models of other agents that each agent then uses to predict the future with. Stay tuned for the camready.
AAMAS camready: Model-based RL
Together with Thomas Kipf, Max Welling and myself, Elise van der Pol did some excellent work on model-based RL.
See:
- This post on plannable approximations with MDP homomorphisms
- The paper
Co-organizing COMARL at the AAAI Spring Symposia
I will be co-organizing a AAAI spring symposium on “Challenges and Opportunities for Multi-Agent Reinforcement Learning”. We want to make it a workshop with some actual ‘work’. Please read here for more info.
Why should we care about AI playing video games?
This is the question that De Volkskrant asked me to comment on. Find the piece here (in Dutch).
Influence-Based Abstraction in Deep Reinforcement Learning
Scaling Bayesian RL for Factored POMDPs
Reinforcement learning is tough. POMDPs are hard. And doing RL in partially observable problems is a huge challenge. With Sammie and Chris Amato, I have been making some progress to get a principled method (based on Monte Carlo tree search) too scale for structured problems. We can learn both how to act, as well as the structure of the problem at the same time. See the paper and bib.
At AAMAS: Deep learning of Coordination…?
Can deep Q-networks etc. brute force their way through tough coordination problems…? Perhaps not. Jacopo’s work, accepted as an extended abstract at AAMAS’19, takes a first step in exploring this in the one-shot setting.
Not so surprising: “joint Q-learner” can be too large/slow and “individual Q-learners” can fail to find good representations.
But good to know: “factored Q-value functions” which represent the Q-function as a random mixture of components involving 2 or 3 agents, can do quite well, even for hard coordination tasks!
About learning machines and AI
science squared on my research project
A popular piece on my ERC research: