Building upon the success of the workshop on "Security for All in an AI-enabled Society" held in Manchester, UK (https://enncore.github.io/events/secaiws/), we are excited to announce AFRiTS, a workshop co-located with the 26th Brazilian Symposium on Formal Methods (https://sbmf23.ufam.edu.br/) to occur in Manaus/Brazil. AFRiTS will focus on automated formal reasoning techniques to build trustworthy AI systems. We invite the community to submit a short paper to AFRiTS 2023, where we will explore the development, dissemination, and utilization of formal methods to ensure the reliability of AI systems.

In particular, AFRiTS seeks to address the security challenges of ensuring safety, transparency, and robustness in neural-based architectures and their implementation in computer-based systems using automated formal reasoning techniques. The shift from traditional fixed mathematical models to machine learning (ML) approaches has made achieving safety and robustness more complex. ML algorithms adapt and update their models based on data, making future behavior uncertain. This presents a dual role for ML algorithms in automated formal reasoning: "model evolution" and "verification enhancement." We aim to delve deeply into these topics during the AFRiTS workshop.

We look forward to engaging in fruitful discussions about recent achievements and future initiatives in the domain of automated formal reasoning for building trustworthy AI systems. By joining AFRiTS, you will have the opportunity to be part of a vibrant community that is actively working towards the advancement of AI safety and reliability.

Please consider submitting your research to AFRiTS 2023 and be part of shaping the future of AI systems built on strong formal foundations. Together, let's explore the challenges and opportunities in ensuring the trustworthiness of AI in an increasingly AI-driven world.

Submission Guidelines

We invite researchers to submit an extended abstract (at most four pages) describing the research they intend to present in the workshop. These short papers will be peer-reviewed and published at https://ceur-ws.org.  All papers should present unpublished and original work that has a clear contribution to the state-of-the-art on the theory and practice of formal methods for building trustworthy AI systems.

List of Topics

• Formal Methods
• Formal Verification
• Automated Reasoning
• Security for AI and AI for Security
• Trustworthy AI
• AI Repair
• Vulnerability Detection

Committees

Program Committee

• Artur d'Avila Garcez (City, University of London, UK)
• Ekaterina Komendantskaya (Heriot-Watt University, UK)
• Youcheng Sun (The University of Manchester, UK)
• Edjard Mota (Federal University of Amazonas)
• Lucas C. Cordeiro (The University of Manchester, UK) 
• Louise Dennis (The University of Manchester, UK)
• Sergio Campos (Federal University of Minas Gerais, Brazil)

Organizing committee

• Lucas C. Cordeiro (The University of Manchester, UK) 
• Edjard Mota (Federal University of Amazonas)

Publication

AFRiTS 2023 proceedings will be published at https://ceur-ws.org.

Venue

The conference will be held at the Federal University of Amazonas, Manaus, Brazil.

Contact

All submission questions should be emailed to Lucas Cordeiro (lucas.cordeiro@manchester.ac.uk) and Edjard Mota (edjard@icomp.ufam.edu.br).