d665: NIPS 2017 Notes – David Abel (Brown University)

NIPS 2017 Notes – David Abel (Brown University) – learn more | reddit discussion

NIPS 2017 Notes Contents:
1 Conference Summary and Highlights
2 Monday
2.1 Tutorial: Deep Probablistic Modeling w/ Gaussian Processes
2.2 Tutorial: Reverse Engineering Intelligence
2.2.1 Part Two: Implementing These Ideas
2.3 NIPS 2017 Welcome
2.3.1 Statistics about the Conference
2.4 Keynote: John Platt on the next 100 years of Human Civilization
2.4.1 Radical Research into Zero-Carbon Energy
2.4.2 Machine Learning for Fusion
3 Tuesday
3.1 Kate Crawford: The Trouble with Bias
3.2 Best Paper Talk: Safe and Nested Subgame Solving for Imperfect Information
3.3 Theory Spotlights
3.3.1 Bandit Optimization
3.3.2 Computational Perspective on Shallow Learning
3.3.3 Monte-Carlo Tree Search by Best Arm Identification
3.4 Deep Learning Spotlights
3.4.1 Deep Mean-shift priors for image restoration
3.4.2 Deep voice-2: Text to Speech
3.4.3 Graph Matching via MWU
3.4.4 Dynamic Routing in Capsules
4 Wednesday
4.1 Speaker: Pieter Abbeel on Deep RL for Robots
4.1.1 Sample Efficient Reinforcement Learning
4.1.2 Hierarchies
4.1.3 Imitation Learning
4.1.4 Lifelong Learning
4.1.5 Leverage Simulation
4.1.6 Yann LeCun’s Cake
4.2 RL Session
4.2.1 ELF: Framework for RL + Game Research
4.2.2 Imagination-Augmented Agents for Deep RL
4.2.3 Simple module for relational reasoning
4.2.4 Scalable TRPO w/ Kronecker Approximation
4.2.5 Off-Policy evaluation for slate recommendation
4.2.6 Transfer learning with HIP-MDPs
4.2.7 Inverse Reward Design
4.2.8 Safe Interruptibility
4.2.9 Unifying PAC and Regret
4.2.10 Repeated IRL
5 Thursday
5.1 Yael Niv on Learning State Representations
5.2 Deep RL Symposium
5.2.1 David Silver: AlphaGo and AlphaZero
5.2.2 Soft Actor Critic
6 Friday
7 Saturday: Hierarchical Reinforcement Learning Workshop
7.1 Invited Talk: David Silver on Subgoals and HRL
7.2 Contributed
7.2.1 Landmark Options via Reflection in Multi-task RL
7.2.2 Cross Modal Skill Learner
7.3 Invited Talk: Jurgen Schmidhuber on HRL and Metalearning
7.4 Invited Talk: Pieter Abbeel on HRL
7.5 Best Paper Talk: Learning with options that terminate off policy
7.6 Posters
7.7 Invited Talk: Jan Peters on imitation HRL for Robotics
7.8 Contributed Talks
7.8.1 Deep Abstract Q-Networks
7.8.2 Hierarchical Multi-Agent Deep RL
7.8.3 Master-Slave Communication for Multi-Agent Deep RL
7.9 Invited Talk: Emma Brunskill on Sample Efficiency in Hierarchical RL
7.10 Invited Talk: Matt Botvonick on Information Bottleneck in HRL
7.11 Invited Talk: Doina Precup on Progress in Deep Temporal RL
7.12 Panel: Doina, Jurgen, Matt, David, Jan, Marcos