As cyber threats grow increasingly sophisticated, reinforcement learning is emerging as a promising technique to create intelligent, self-improving defensive systems. However, most existing autonomous agents have overlooked the inherent graph structure of computer networks subject attacks, potentially missing critical information. To address this gap, we developed custom version Cyber Operations Research Gym (CybORG) environment that encodes observable network state directed graph, utilizing...

Two automated knowledge discovery methodologies (ADoK-S & ADoK-W) are created whereby symbolic regression, parameter estimation, information criteria and model-based design of experiments synergize for the optimized kinetic rate models.

Neural ordinary differential equations (Neural ODEs) define continuous time dynamical systems with neural networks. The interest in their application for modelling has sparked recently, spanning hybrid system identification problems and series analysis. In this work we propose the use of a control policy capable satisfying state constraints to solve nonlinear optimal problems. optimization is posed as ODE problem efficiently exploit availability model. We showcase efficacy type deterministic...

Dynamic real-time optimization (DRTO) is a challenging task due to the fact that optimal operating conditions must be computed in real time. The main bottleneck industrial application of DRTO presence uncertainty. Many stochastic systems present following obstacles: 1) plant-model mismatch, 2) process disturbances, 3) risks violation constraints. To accommodate these difficulties, we constrained reinforcement learning (RL) based approach. RL naturally handles uncertainty by computing an...

PC-Gym is an open-source tool designed to facilitate the development and evaluation of reinforcement learning (RL) algorithms for chemical process control problems. It provides a suite environments that model range processes, incorporating nonlinear dynamics, disturbances, constraints. Key features include flexible constraint handling mechanisms, customizable disturbance generation, modular reward function design. The framework enables benchmarking state-of-the-art RL against Model...

The 2D O(3) model is widely used as a toy for ferromagnetism and quantum chromodynamics. With the latter it shares---among other basic aspects---the property that continuum functional integral splits into topological sectors. Topology can also be defined in its lattice regularized version, but semiclassical arguments suggest susceptibility ${\ensuremath{\chi}}_{\mathrm{t}}$ does not scale towards finite limit. Previous numerical studies confirmed quantity...

Chemical process optimization and control are affected by 1) plant-model mismatch, 2) disturbances, 3) constraints for safe operation. Reinforcement learning policy would be a natural way to solve this due its ability address stochasticity, directly account the effect of future uncertainty feedback in proper closed-loop manner; all without need an inner loop. One main reasons why reinforcement has not been considered industrial processes (or almost any engineering application) is that it...

We study the impact of Gradient Flow on topology in various models lattice field theory. The topological susceptibility X t is measured directly, and by slab method , which based content sub-volumes (“slabs”) estimates even when system remains trapped a fixed sector. results obtained both methods are essentially consistent, but characteristic quantity seems to be different 2-flavour QCD 2d O(3) model. In latter model, we further address question whether or not leads finite continuum limit...

The 2d Heisenberg model --- or O(3) is popular in condensed matter physics, and particle physics as a toy for QCD. Along with other analogies, it shares 4d Yang-Mills theories, QCD, the property that configurations are divided topological sectors. In lattice regularisation charge $Q$ can still be defined such $Q \in \mathbb{Z}$. It has generally been observed, however, susceptibility $\chi_{\rm t} = \langle Q^2 \rangle / V$ does not scale properly continuum limit, i.e. quantity \xi^2$...

The industrialization of catalytic processes requires reliable kinetic models for their design, optimization and control. Mechanistic require significant domain knowledge, while data-driven hybrid lack interpretability. Automated knowledge discovery methods, such as ALAMO (Automated Learning Algebraic Models Optimization), SINDy (Sparse Identification Nonlinear Dynamics), genetic programming, have gained popularity but suffer from limitations needing model structure assumptions, exhibiting...

Bioprocesses have received a lot of attention to produce clean and sustainable alternatives fossil-based materials. However, they are generally difficult optimize due their unsteady-state operation modes stochastic behaviours. Furthermore, biological systems highly complex, therefore plant-model mismatch is often present. To address the aforementioned challenges we propose Reinforcement learning based optimization strategy for batch processes. In this work, applied Policy Gradient method...

Dynamic real-time optimization (DRTO) is a challenging task due to the fact that optimal operating conditions must be computed in real time. The main bottleneck industrial application of DRTO presence uncertainty. Many stochastic systems present following obstacles: 1) plant-model mismatch, 2) process disturbances, 3) risks violation constraints. To accommodate these difficulties, we constrained reinforcement learning (RL) based approach. RL naturally handles uncertainty by computing an...

Neural ordinary differential equations (Neural ODEs) define continuous time dynamical systems with neural networks. The interest in their application for modelling has sparked recently, spanning hybrid system identification problems and series analysis. In this work we propose the use of a control policy capable satisfying state constraints to solve nonlinear optimal problems. optimization is posed as ODE problem efficiently exploit availability model. We showcase efficacy type deterministic...