Researchers have introduced Behavior Learning (BL), a novel machine learning framework that learns interpretable optimization structures from data.

Key Features

**Accepted at ICLR 2026**
- **Interpretable**: Learn utility functions in symbolic form
- **Identifiable**: Smooth variant (IBL) guarantees identifiability
- **Universal approximation**: Proven theoretical property

## How It Works

BL parameterizes:
- Compositional utility functions
- Built from interpretable modular blocks
- Induces data distribution for prediction

## Applications

- Scientific domains involving optimization
- Economics and decision theory
- Robotics and control
- Any domain where interpretability matters

Installation

```bash
pip install blnetwork
```

GitHub: github.com/MoonYLiang/Behavior-Learning