Write a Blog >>
APLAS 2020
Mon 30 November - Wed 2 December 2020
Mon 30 Nov 2020 19:00 - 19:30 at online - Types Chair(s): Marco Servetto

We present StarChild and Lazuli, two libraries which leverage refinement types to verify neural networks, implemented in F∗ and Liquid Haskell. Refinement types are types augmented, or refined, with assertions about values of that type, e.g., “integers greater than five”, which are checked by an SMT solver. Crucially, these assertions are written in the language itself. A user of our library can refine the type of neural networks, e.g., “neural networks which are robust against adversarial attacks”, and expect F∗ to handle the verification of this claim for any specific network, without having to change the representation of the network, or even having to learn about SMT solvers. Our initial experiments indicate that our approach could greatly reduce the burden of verifying neural networks. Unfortunately, they also show that SMT solvers do not scale to the sizes required for neural network verification.

Mon 30 Nov
Times are displayed in time zone: Osaka, Sapporo, Tokyo change

17:30 - 19:30
TypesResearch Papers at online
Chair(s): Marco ServettoVictoria University Wellington, New Zealand
17:30
30m
Talk
Syntactically Restricting Bounded Polymorphism for Decidable Subtyping
Research Papers
Julian MackayVictoria University of Wellington, Alex PotaninVictoria University of Wellington, Jonathan AldrichCarnegie Mellon University, Lindsay GrovesVictoria University of Wellington
18:00
30m
Talk
A New Refinement Type System for Automated nu-HFLZ Validity Checking
Research Papers
Hiroyuki KatsuraThe University of Tokyo, Naoki IwayamaUniversity of Tokyo, Japan, Naoki KobayashiUniversity of Tokyo, Japan, Takeshi TsukadaChiba University, Japan
18:30
30m
Talk
Behavioural Types for Memory and Method Safety in a Core Object-Oriented Language
Research Papers
Mario BravettiUniversità di Bologna, Adrian FrancalanzaUniversity of Malta, Iaroslav GolovanovDepartment of Computer Science, Aalborg University, Hans HüttelDepartment of Computer Science, Aalborg University, Mathias Steen JakobsenDepartment of Computer Science, Aalborg University, Denmark, Mikkel Klinke KettunenDepartment of Computer Science, Aalborg University, Denmark, Antonio RavaraDepartment of Informatics, Faculty of Sciences and Technology, NOVA University of Lisbon and NOVA LINCS
19:00
30m
Talk
Neural Networks, Secure by Construction: An Exploration of Refinement Types
Research Papers
Wen KokkeUniversity of Edinburgh, Ekaterina KomendantskayaHeriot-Watt University, UK, Daniel KienitzHeriot-Watt University, David AspinallUniversity of Edinburgh, Robert AtkeyUniversity of Strathclyde