Probabilistic programs are a structured way to describe computations or models with access to some source of randomness. They appear naturally in various safety-, security-, and privacy-critical applications, including randomized algorithms, security protocols, and autonomous systems working in uncertain environments, e.g., due to imprecise sensors.
The behavior of probabilistic programs is often counterintuitive — a consequence of the well-known fact that humans have difficulties reasoning about stochastic processes. In combination with their importance in emerging safety-critical domains, the counterintuitive nature of probabilistic programs means that ensuring their correctness must be based on verification and analysis techniques that are rigorous, tool-supported, and, ideally, largely automated. However, such tools have not kept up with the increasing usage and popularity of probabilistic programs.
In part, the lack of tool support can be explained by the fact that research on probabilistic programs is spread over multiple fields. In addition to the classical understanding of probabilistic programs as randomized algorithms, probabilistic programs have received rapidly increasing attention as a modeling formalism for complex probability distributions in machine learning, artificial intelligence, and cognitive science.
This workshop will provide a forum for research on the automated verification of probabilistic systems that are in some way described by a programming language, with a particular focus on both symbolic methods and compositional approaches.
Call for Presentations
VeriProP aims to bring together researchers interested in the tool-supported verification of probabilistic programs, models, and systems. This includes probabilistic model checking, program verification in the presence of a source of randomness, or formal guarantees for statistical machine learning algorithms and artificial intelligence systems.
We solicit presentations of 10 minutes. Topics of interest include, but are not limited to:
- Symbolic approaches to the verification of Markov models
- Exact inference techniques
- Abstract interpretation for probabilistic programs
- Domain-specific probabilistic programming languages
- Verification of inference algorithms
- Automation of deductive approaches to verifying probabilistic programs-
- Probabilistic program reasoning in safety, security, or privacy
- Synthesis of probabilistic programs
We call for extended abstracts (1-2 pages in pdf format) describing either ongoing research or an overview of past research in the workshop’s scope. We welcome abstracts covering work that has been previously published or is currently under review. There will be no formal proceedings.
Submission date: May 10
Notification date: June 15
Submission link: easychair.org/my/conference?conf=veriprop2022
This workshop will be held as a satellite event of FLoC 2022. The workshop is chaired by:
- Ezio Bartocci, TU Wien
- Fredrik Dahlqvist, Queen Mary University of London
- Sebastian Junges, Radboud University
- Benjamin Kaminski, Saarland University and University College London
- Christoph Matheja, Technical University of Denmark