Saved in:
| Main Authors: | , , , , |
|---|---|
| Format: | Preprint |
| Published: |
2025
|
| Subjects: | |
| Online Access: | https://arxiv.org/abs/2503.06420 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1866910891701698560 |
|---|---|
| author | Chakraborty, Debraj Dubslaff, Clemens Kanav, Sudeep Kretinsky, Jan Weinhuber, Christoph |
| author_facet | Chakraborty, Debraj Dubslaff, Clemens Kanav, Sudeep Kretinsky, Jan Weinhuber, Christoph |
| contents | Safety-critical controllers of complex systems are hard to construct manually. Automated approaches such as controller synthesis or learning provide a tempting alternative but usually lack explainability. To this end, learning decision trees (DTs) have been prevalently used towards an interpretable model of the generated controllers. However, DTs do not exploit shared decision-making, a key concept exploited in binary decision diagrams (BDDs) to reduce their size and thus improve explainability. In this work, we introduce predicate decision diagrams (PDDs) that extend BDDs with predicates and thus unite the advantages of DTs and BDDs for controller representation. We establish a synthesis pipeline for efficient construction of PDDs from DTs representing controllers, exploiting reduction techniques for BDDs also for PDDs. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2503_06420 |
| institution | arXiv |
| publishDate | 2025 |
| record_format | arxiv |
| spellingShingle | Explaining Control Policies through Predicate Decision Diagrams Chakraborty, Debraj Dubslaff, Clemens Kanav, Sudeep Kretinsky, Jan Weinhuber, Christoph Artificial Intelligence Systems and Control Safety-critical controllers of complex systems are hard to construct manually. Automated approaches such as controller synthesis or learning provide a tempting alternative but usually lack explainability. To this end, learning decision trees (DTs) have been prevalently used towards an interpretable model of the generated controllers. However, DTs do not exploit shared decision-making, a key concept exploited in binary decision diagrams (BDDs) to reduce their size and thus improve explainability. In this work, we introduce predicate decision diagrams (PDDs) that extend BDDs with predicates and thus unite the advantages of DTs and BDDs for controller representation. We establish a synthesis pipeline for efficient construction of PDDs from DTs representing controllers, exploiting reduction techniques for BDDs also for PDDs. |
| title | Explaining Control Policies through Predicate Decision Diagrams |
| topic | Artificial Intelligence Systems and Control |
| url | https://arxiv.org/abs/2503.06420 |