Research and Practice of Deep

Learning and Reinforcement Learning in

Smart Grids

READ THIS PAPER

https://ieeexplore.ieee.org/stamp/stamp.jsp?tp=&arnumber=8468674

• Reinforcement Learning in Robotics: How can be a Robot like Human? Emotion, reaction, intelligent are to be as like as human.
• Safety in Reinforcement Learning: How to avoid unwanted behaviour and reward hacking in RL.
• Competitive and Cooperative Multi-agent Reinforcement Learning:
• Data Efficiency: React and take a decision just from a single raining.

you have to search for articles that are published in the last years on google scholar, you can consult the following links:

• https://arxiv.org/pdf/1804.04577
• https://arxiv.org/pdf/1708.05866
• https://ieeexplore.ieee.org/iel7/6287639/7859429/08119919.pdf
• https://dl.acm.org/citation.cfm?id=3068287
• http://www.jmlr.org/papers/volume18/16-634/16-634.pdf
• .....

Training Game agents or bots using Reinforcement Learning

Honestly if you want to research there are two streams here.

How to use reinforcement learning in regular machine learning problems? Industry is really looking up to it.

How reinforcement learning can help in security domain like creating an antivirus with help of reinforcement learning etc.

Might want to look at my own paper on RL.

Preprint Reinforcement Evolutionary Learning Method for self-learning

This might help you to develop further hypothesis.

How to make reinforcement learning tractable and interpretable. This is a problem since when we use deep learning in RL we can't always explain the behavior. In addition the tractable part comes from being able to train it and use it given some limits in time and computation.

Have a look at https://blog.valohai.com/reinforcement-learning-tutorial-part-1-q-learning

Challenges: https://medium.com/ai³-theory-practice-business/here-are-the-9-challenges-dragging-practical-rl-down-c96cc6366527

https://www.kdnuggets.com/2017/03/next-challenges-reinforcement-learning.html

Dear Mahesh S Patil,

Reinforcement learning (RL) continues to be less valuable for business applications than supervised learning, and even unsupervised learning. It is successfully applied only in areas where huge amounts of simulated data can be generated, like robotics and games. RL as a promising path towards Artificial General Intelligence (AGI), or true intelligence. Model based RL and Model free algorithm for RL are available to successful performance of simulator using AI. Bayesian RL Bayesian RL is quite common. Deep Reinforcement Learning (Monte Carlo) Tree Search Theory of RL Learning for optimization, Sequential sub problem generation Parameter control Using learned value functions in MIP models are most interested part of RL study.

Couple of important topics in RL for AI (Breakthrough research topics in RL) are;

1. How to Combine Tree-Search Methods in Reinforcement Learning

2. Social Influence as Intrinsic Motivation for Multi-Agent Deep Reinforcement Learning

3. Efficient Off-Policy Meta-Reinforcement Learning via Probabilistic Context Variables

4. Policy Certificates: Towards Accountable Reinforcement Learning

5. Distributional Reinforcement Learning for Efficient Exploration

6. Better Exploration with Optimistic Actor-Critic

7. Guided Meta-Policy Search

8. Using a Logarithmic Mapping to Enable Lower Discount Factors in Reinforcement Learning

9. Emergent Tool Use From Multi-Agent Autocurricula

10. Solving Rubik’s Cube with a Robot Hand

Ashish

Dear Mahesh S Patil,

Reinforcement learning (RL) continues to be less valuable for business applications than supervised learning, and even unsupervised learning. It is successfully applied only in areas where huge amounts of simulated data can be generated, like robotics and games. RL as a promising path towards Artificial General Intelligence (AGI), or true intelligence. Model based RL and Model free algorithm for RL are available to successful performance of simulator using AI. Bayesian RL Bayesian RL is quite common.

***Deep Reinforcement Learning (Monte Carlo) Tree Search Theory of RL Learning for optimization, Sequential sub problem generation Parameter control Using learned value functions in MIP models are most interesting part of RL study.

***Couple of important topics in RL for AI (Breakthrough research topics in RL) are;

1. How to Combine Tree-Search Methods in Reinforcement Learning

2. Social Influence as Intrinsic Motivation for Multi-Agent Deep Reinforcement Learning

3. Efficient Off-Policy Meta-Reinforcement Learning via Probabilistic Context Variables

4. Policy Certificates: Towards Accountable Reinforcement Learning

5. Distributional Reinforcement Learning for Efficient Exploration

6. Better Exploration with Optimistic Actor-Critic

7. Guided Meta-Policy Search

8. Using a Logarithmic Mapping to Enable Lower Discount Factors in Reinforcement Learning

9. Emergent Tool Use From Multi-Agent Autocurricula

10. Solving Rubik’s Cube with a Robot Hand

Ashish