Enregistré dans:
| Auteurs principaux: | , , , |
|---|---|
| Format: | Preprint |
| Publié: |
2026
|
| Sujets: | |
| Accès en ligne: | https://arxiv.org/abs/2602.13128 |
| Tags: |
Ajouter un tag
Pas de tags, Soyez le premier à ajouter un tag!
|
| _version_ | 1866914328483987456 |
|---|---|
| author | Tarraf, Mohamed Chan, Alex Yakovlev, Alex Shafik, Rishad |
| author_facet | Tarraf, Mohamed Chan, Alex Yakovlev, Alex Shafik, Rishad |
| contents | Binary Neural Networks (BNNs) offer a low-complexity and energy-efficient alternative to traditional full-precision neural networks by constraining their weights and activations to binary values. However, their discrete, highly non-linear behavior makes them difficult to explain, validate and formally verify. As a result, BNNs remain largely opaque, limiting their suitability in safety-critical domains, where causal transparency and behavioral guarantees are essential. In this work, we introduce a Petri net (PN)-based framework that captures the BNN's internal operations as event-driven processes. By "eventizing" their operations, we expose their causal relationships and dependencies for a fine-grained analysis of concurrency, ordering, and state evolution. Here, we construct modular PN blueprints for core BNN components including activation, gradient computation and weight updates, and compose them into a complete system-level model. We then validate the composed PN against a reference software-based BNN, verify it against reachability and structural checks to establish 1-safeness, deadlock-freeness, mutual exclusion and correct-by-construction causal sequencing, before we assess its scalability and complexity at segment, component, and system levels using the automated measurement tools in Workcraft. Overall, this framework enables causal introspection of transparent and event-driven BNNs that are amenable to formal reasoning and verification. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2602_13128 |
| institution | arXiv |
| publishDate | 2026 |
| record_format | arxiv |
| spellingShingle | Eventizing Traditionally Opaque Binary Neural Networks as 1-safe Petri net Models Tarraf, Mohamed Chan, Alex Yakovlev, Alex Shafik, Rishad Machine Learning Binary Neural Networks (BNNs) offer a low-complexity and energy-efficient alternative to traditional full-precision neural networks by constraining their weights and activations to binary values. However, their discrete, highly non-linear behavior makes them difficult to explain, validate and formally verify. As a result, BNNs remain largely opaque, limiting their suitability in safety-critical domains, where causal transparency and behavioral guarantees are essential. In this work, we introduce a Petri net (PN)-based framework that captures the BNN's internal operations as event-driven processes. By "eventizing" their operations, we expose their causal relationships and dependencies for a fine-grained analysis of concurrency, ordering, and state evolution. Here, we construct modular PN blueprints for core BNN components including activation, gradient computation and weight updates, and compose them into a complete system-level model. We then validate the composed PN against a reference software-based BNN, verify it against reachability and structural checks to establish 1-safeness, deadlock-freeness, mutual exclusion and correct-by-construction causal sequencing, before we assess its scalability and complexity at segment, component, and system levels using the automated measurement tools in Workcraft. Overall, this framework enables causal introspection of transparent and event-driven BNNs that are amenable to formal reasoning and verification. |
| title | Eventizing Traditionally Opaque Binary Neural Networks as 1-safe Petri net Models |
| topic | Machine Learning |
| url | https://arxiv.org/abs/2602.13128 |