18188024. MODEL-BASED REINFORCEMENT LEARNING simplified abstract (Ford Global Technologies, LLC)
Contents
MODEL-BASED REINFORCEMENT LEARNING
Organization Name
Inventor(s)
Kaushik Balakrishnan of Mountain View CA (US)
Neeloy Chakraborty of Brentwood TN (US)
Devesh Upadhyay of Canton MI (US)
MODEL-BASED REINFORCEMENT LEARNING - A simplified explanation of the abstract
This abstract first appeared for US patent application 18188024 titled 'MODEL-BASED REINFORCEMENT LEARNING
The abstract describes a computer system that trains an agent neural network to input a state and output an action, then input that action to an environment to determine a new state and reward. A Koopman model neural network is used to generate a fake state based on the input and output states. The agent neural network and Koopman model neural network are re-trained using reinforcement learning with the input state, output action, new state, fake state, and reward.
- The computer system includes a processor and memory.
- The memory contains instructions for training an agent neural network.
- The agent neural network inputs a state and outputs an action.
- The action is input to an environment to determine a new state and reward.
- A Koopman model neural network generates a fake state based on the input and output states.
- Both neural networks are re-trained using reinforcement learning.
Potential Applications: - Autonomous vehicles - Robotics - Gaming industry - Financial trading algorithms
Problems Solved: - Improving decision-making processes in complex environments - Enhancing learning capabilities of neural networks
Benefits: - Increased efficiency in training neural networks - Better adaptation to changing environments - Enhanced performance in decision-making tasks
Commercial Applications: Title: "Enhanced Decision-Making Systems for Autonomous Vehicles and Robotics" This technology can be utilized in developing advanced autonomous systems for various industries, including transportation, manufacturing, and entertainment. The market implications include improved safety, efficiency, and productivity in automated processes.
Questions about the technology: 1. How does the Koopman model neural network contribute to the training process? 2. What are the key advantages of using reinforcement learning in training neural networks?
Original Abstract Submitted
A computer that includes a processor and a memory, the memory including instructions executable by the processor to train an agent neural network to input a first state and output a first action, input the first action to an environment and determine a second state and a reward. Koopman model neural network can be trained based on the first state, the first action and the second state to determine a fake state. The agent neural network can be re-trained and the Koopman model neural network can be re-trained based on reinforcement learning including the first state, the first action, the second state, the fake state, and the reward.
