@InProceedings{alberti2025caisar,
  author           = {Michele Alberti, François Bobot, Julien Girard-Satabin, Alban Grastien, Aymeric Varasse, Zakaria Chihani},
  booktitle        = {Integrated Formal Methods (iFM)},
  title            = {The CAISAR Platform: Extending the Reach ofMachine Learning Specification and Verification},
  year             = {2025},
  address          = {Paris},
  editor           = {Springer},
  month            = nov,
  abstract         = {The formal specification and verification of machine learning
models have advanced remarkably in less than a decade, leading to a pro-
fusion of verification tools that provide mathematical guarantees about
model properties. However, this growing diversity risks ecosystem frag-
mentation, making it difficult to compare tools beyond narrowly defined
benchmarks. Moreover, much of the progress to date has focused on a
limited class of properties, particularly local robustness. While existing
tools are increasingly effective at verifying such properties, more complex
ones, such as those involving multiple neural networks, remain beyond
their capabilities: these properties cannot currently be expressed in their
specification languages, nor can they be directly verified. This applies
even to the winning verification tools of the International Verification of
Neural Networks Competition (VNN-Comp).
In this tool paper, we present CAISAR, an open-source platform for spec-
ifying and verifying properties of machine learning models, with partic-
ular focus on neural networks and support vector machines. CAISAR
provides a high-level language for specifying complex properties and in-
tegrates several state-of-the-art verifiers for their automatic verification.
Through concrete use cases, we show how CAISAR leverages automated
graph-editing techniques to translate high-level specifications into queries
for the supported verifiers, bridging the (embedding) gap between user
specifications and the corresponding ones that are actually verified},
  creationdate     = {2025-09-05T12:11:10},
  doi              = {10.1007/978-3-032-10794-7_15},
  groups           = {team},
  modificationdate = {2025-11-19T13:36:43},
}