Quantitative Evaluation of Systems - 20th International Conference, QEST 2023, Antwerp, Belgium, September 20-22, 2023, Proceedings

Nils Jansen, Mirco TribastoneQest 2023 · DOI · DBLP

Introduction

The 20th International Conference on Quantitative Evaluation of SysTems (QEST 2023), held in Antwerp, Belgium, marks a significant milestone in the evolution of formal verification and performance analysis. This year’s proceedings highlight two decades of progress, showcasing a collection of 23 papers that reflect a strategic shift toward the integration of traditional formal methods with modern artificial intelligence.

Addressing the Complexity of Modern Systems

As computer systems transition from deterministic models to highly complex, data-driven, and stochastic architectures, standard testing methodologies often fail to provide sufficient reliability. QEST 2023 addresses the fundamental challenge of "quantitative evaluation"—the application of rigorous mathematical methods to verify that systems, ranging from distributed networks to chemical reaction models, behave correctly and efficiently under conditions of uncertainty.

Methodological Innovations

The 2023 proceedings encompass a wide variety of novel methodologies rather than a single algorithmic focus. Key technical themes include:

  • Probabilistic Model Checking: Significant updates to established tools, such as the modernization of STAMINA in C++, allow for more efficient handling of infinite-state systems and multi-agent environments.
  • Stochastic Modeling and Learning: The introduction of libraries like Jajapy for learning stochastic models and the use of Variational Autoencoders for inferring chemical reaction networks demonstrate a move toward data-integrated modeling.
  • Abstraction Techniques: Researchers presented "Neural Abstraction" methods to balance computational efficiency with precision when analyzing neural networks, alongside "Score-Based Diffusion Models" for model abstraction.
  • Formal Synthesis: New techniques were introduced for synthesizing controllers for Markov Jump Linear Systems and deductive synthesis for probabilistic hyperproperties.
  • Hybrid and Quantum Systems: The collection features novel approaches to include stochasticity in hybrid automata and the use of barrier certificates for the verification of quantum systems.

The AI-Verification Bridge

A defining theme of QEST 2023 is the "growing push towards joint efforts in verification and artificial intelligence." This synergy was emphasized by invited speakers David Parker (Verification) and Frans Oliehoek (AI), signaling a future where formal methods are used to make AI safer, while AI techniques are leveraged to make formal verification more scalable.

Results and Community Impact

The conference maintained a high standard of selectivity, accepting 23 papers out of 44 submissions (an approximately 52% acceptance rate). Beyond theoretical breakthroughs, such as research on causal and time reversibility, the proceedings emphasize practical tool modernization and artifact maturity. Notably, 11 out of 12 submitted artifacts (code and data) were accepted, indicating a robust standard for usable, open-source research within the community.

Limitations and Future Directions

Despite these advancements, several challenges persist:

  • Scalability: The "state-space explosion" remains a recurring obstacle for probabilistic and hybrid systems, necessitating continued tool optimization.
  • Integration Gaps: Finding the ideal balance between "black-box" machine learning models and "white-box" formal proofs is an ongoing research effort.
  • Future Focus: The editors suggest a continued emphasis on data-driven systems, real-world case studies, and the development of robust tool papers to support the global R&D community.

Summary updated

BibTeX
@inproceedings{jansen2023quantitative,
  author = {Nils Jansen and Mirco Tribastone},
  title = {{Quantitative Evaluation of Systems - 20th International Conference, QEST 2023, Antwerp, Belgium, September 20-22, 2023, Proceedings}},
  booktitle = {Qest},
  year = {2023},
  doi = {10.1007/978-3-031-43835-6},
}

Loading related content...

Figure