Model-Based Learning
Definition
Explicit assumptions form the starting point of an adapted learning solution.
Here, model-based learning refers to the approach of expressing domain knowledge and uncertainties in a probabilistic model and deriving inference from it.
Simply explained
First, one describes how the world is likely connected; then the system learns from data.
Why it matters in practice
Assumptions become visible and can be deliberately changed.
Example from business
Hypothetical example: A company models demand taking seasonality and measurement errors into account.
Limits and pitfalls
- A well-calculated model can be based on incorrect assumptions.
Origin & history
The approach is described, among others, by Bishop's work on Model-Based Machine Learning.
Use in mid-sized companies
- SMEs with domain expertise can have suitable models developed instead of only standardized ones.
Frequently asked questions
Is this the same as model-based reinforcement learning?
No, there a specific environment model is used for decision-making.
Sources
- Model-based machine learning — · Christopher Bishop: Model-based machine learning —
Editorially reviewed on: