Course Outline
Introduction
- Learning through positive reinforcement for government applications
Elements of Reinforcement Learning
Important Terms (Actions, States, Rewards, Policy, Value, Q-Value, etc.)
Overview of Tabular Solution Methods
Creating a Software Agent for Government Use
Understanding Value-based, Policy-based, and Model-based Approaches in Public Sector Applications
Working with the Markov Decision Process (MDP) for Government Scenarios
How Policies Define an Agent's Way of Behaving in Government Contexts
Using Monte Carlo Methods for Government Analysis
Temporal-Difference Learning for Government Projects
n-step Bootstrapping for Enhanced Government Decision-Making
Approximate Solution Methods for Complex Government Challenges
On-policy Prediction with Approximation in Government Settings
On-policy Control with Approximation for Government Operations
Off-policy Methods with Approximation for Government Applications
Understanding Eligibility Traces for Government Use
Using Policy Gradient Methods for Government Initiatives
Summary and Conclusion
Requirements
- Experience in machine learning
- Programming expertise
Audience
- Data scientists for government
