Seminars

## Towards Security-Oriented Program analysis

Speaker: Sébatien Bardin, CEA, Paris-Saclay

Tuesday 7 September 2021, 11:00, (location to come)

Abstract: While digital security concerns increase, we face both a urging demand for more and more code-level security analysis and a shortage of security experts. Hence the need for techniques and tools able to automate part of these code-level security analyses. As source-level program analysis and formal methods for safety-critical applications have made tremendous progress in the past decades, it is extremely tempting to adapt them from safety to security. Yet, security is not safety and, while still useful, a direct adaptation of safety-oriented program analysis to security scenarios remains limited in its scope. In this talk, we will argue for the need of security-oriented program analysis. We will first present some of the new challenges faced by formal methods and program analysis in the context of code-level security scenarios. For example, security-oriented code analysis is better performed at the binary level, the attacker must be taken into account and practical security properties deviate from standard reachability / invariance properties. Second, we will discuss some early results and achievements carried out within the BINSEC group at CEA LIST. Especially, we will show how techniques such as symbolic execution and SMT constraint solving can be tailored to a number of practical code-level security scenarios.

## Distributed Causal Memory: Modular Specification and Verification in Higher-Order Distributed Separation Logic

Speaker: Léon Gondelman, Aarhus University, DK

Tuesday 15.6.2021, 11:00, online

Abstract: In this presentation we are going to talk about modular specification and verification of causally-consistent distributed database, a data structure that guarantees causal consistency among replicas of the database.

With causal consistency, different replicas can observe different data on the same key, yet it is guaranteed that all data are observed in a causally related order: if a node {$N$} observes an update {$X$} originating at node {$M$}, then node {$N$} must have also observed the effects of any other update {$Y$} that took place on node {$M$} before {$X$}. Causal consistency can, for instance, be used to ensure in a distributed messaging application that a reply to a message is never seen before the message itself.Read more...

## Controlling a random population

Speaker: Pierre Ohlmann, IRIF

Tuesday 6.4.2021, 11:00, online

Abstract: Bertrand et al. introduced a model of parameterised systems, where each agent is represented by a finite state system, and studied the following control problem: for any number of agents, does there exist a controller able to bring all agents to a target state? They showed that the problem is decidable and EXPTIME-complete in the adversarial setting, and posed as an open problem the stochastic setting, where the agent is represented by a Markov decision process. In this paper, we show that the stochastic control problem is decidable. Our solution makes significant uses of well quasi orders, of the max-flow min-cut theorem, and of the theory of regular cost functions. We introduce an intermediate problem of independent interest called the sequential flow problem, and study the complexity of solving it.

## Towards synthetic psychology: from phenomenology to cybernetics

Speaker: David Rudrauf

Tuesday 9.3.2021, 11:00, online

Abstract: The role of consciousness in biological cybernetics remains an essential yet open question for science. We introduce the Projective Consciousness Model (PCM) and show how its principles yield a unified model of appraisal and social-affective perspective taking and a method for active inference. We show how the PCM can account for known relationships between appraisal and distance as an inverse distance law, and how it can be generalised to implement Theory of Mind for strategic action planning. We use simulations of artificial agents applied to toy robots to demonstrate how different model parameters can generate a variety of emergent adaptive and maladaptive behaviours: from the ability to be resilient in the face of obstacles through imaginary projections, to the emergence of social approach and joint attention behaviours, and the ability to take advantage of false beliefs attributed to others. The approach opens new paths towards a science of consciousness, and applications, from clinical assessment to the design of artificial (virtual and robotic) agents. We discuss the interest and variety of computational challenges entailed by the approach. Read more...