JAJames M. Adams

James M. Adams

*This is an algorithmically generated author record based on publicly available information.

James M. Adams has a total of 38 co-authors. Here are the top co-authors by number of shared publications:


Latest Publications

  • Open Access

    Multi‐Objective Reinforcement Learning for Automated Resilient Cyber Defence

    by Ross O'DRiscoll, Claudia Hagen, Joe Bater, James Adams

    Cyber‐attacks pose a security threat to military command and control networks, Intelligence, Surveillance, and Reconnaissance (ISR) systems, and civilian critical national infrastructure. The use of artificial intelligence and autonomous agents in these attacks increases the scale, range, and complexity of this threat and the subsequent disruption they cause. Autonomous Cyber Defence (ACD) agents aimto mitigate this threat by responding at machine speed and at the scale required to address the problem. Additionally, they reduce the burden on the limited number of human cyber experts available to respond to an attack. Sequential decision‐making algorithms such as Deep Reinforcement Learning (RL) provide a promising route to create ACD agents. These algorithms focus on a single objectivesuch as minimising the intrusion of red agents on the network, by using a handcrafted weighted sum of rewards. This approach removes the ability to adapt the model during inference, and fails to address the many competing objectivespresent when operating and protecting these networks. Conflicting objectives, such as restoring a machine from a back‐up image, must be carefully balanced with the cost of associated down‐time or the disruption to network traffic or services that might result. Instead of pursuing a Single‐Objective RL (SORL) approach, here we present a simple example of a multi‐objective network defense game that requires consideration of both defending the network against red‐agents and maintaining the critical functionality of green‐agents. Two Multi‐Objective Reinforcement Learning (MORL) algorithms, namely Multi‐Objective Proximal Policy Optimization (MOPPO) and Pareto‐Conditioned Networks (PCN), are used to create two trained ACD agents whose performance is compared on our Multi‐Objective Cyber Defense game. The benefits and limitations of MORL ACD agents in comparison to SORL ACD agents are discussed based on the investigations of this game.


  • Open Access

    Multi-Objective Reinforcement Learning for Automated Resilient Cyber Defence

    by Ross O’Driscoll, Claudia Hagen, Joe Bater, James Adams

    Cyber-attacks pose a security threat to military command and control networks, Intelligence, Surveillance, and Reconnaissance (ISR) systems, and civilian critical national infrastructure. The use of artificial intelligence and autonomous agents in these attacks increases the scale, range, and complexity of this threat and the subsequent disruption they cause. Autonomous Cyber Defence (ACD) agents aim to mitigate this threat by responding at machine speed and at the scale required to address the problem. Additionally, they reduce the burden on the limited number of human cyber experts available to respond to an attack. Sequential decision-making algorithms such as Deep Reinforcement Learning (RL) provide a promising route to create ACD agents. These algorithms focus on a single objective such as minimizing the intrusion of red agents on the network, by using a handcrafted weighted sum of rewards. This approach removes the ability to adapt the model during inference, and fails to address the many competing objectives present when operating and protecting these networks. Conflicting objectives, such as restoring a machine from a back-up image, must be carefully balanced with the cost of associated down-time, or the disruption to network traffic or services that might result. Instead of pursing a Single-Objective RL (SORL) approach, here we present a simple example of a multi-objective network defence game that requires consideration of both defending the network against red-agents and maintaining critical functionality of green-agents. Two Multi-Objective Reinforcement Learning (MORL) algorithms, namely Multi-Objective Proximal Policy Optimization (MOPPO), and Pareto-Conditioned Networks (PCN), are used to create two trained ACD agents whose performance is compared on our Multi-Objective Cyber Defence game. The benefits and limitations of MORL ACD agents in comparison to SORL ACD agents are discussed based on the investigations of this game.


  • Transient shear banding in the nematic dumbbell model of liquid crystalline polymers

    by J. M. Adams, D. Corbett

    In the shear flow of liquid crystalline polymers (LCPs) the nematic director orientation can align with the flow direction for some materials but continuously tumble in others. The nematic dumbbell (ND) model was originally developed to describe the rheology of flow-aligning semiflexible LCPs, and flow-aligning LCPs are the focus in this paper. In the shear flow of monodomain LCPs, it is usually assumed that the spatial distribution of the velocity is uniform. This is in contrast to polymer solutions, where highly nonuniform spatial velocity profiles have been observed in experiments. We analyze the ND model, with an additional gradient term in the constitutive model, using a linear stability analysis. We investigate the separate cases of constant applied shear stress and constant applied shear rate. We find that the ND model has a transient flow instability to the formation of a spatially inhomogeneous flow velocity for certain starting orientations of the director. We calculate the spatially resolved flow profile in both constant applied stress and constant applied shear rate in start up from rest, using a model with one spatial dimension to illustrate the flow behavior of the fluid. For low shear rates flow reversal can be seen as the director realigns with the flow direction, whereas for high shear rates the director reorientation occurs simultaneously across the gap. Experimentally, this inhomogeneous flow is predicted to be observed in flow reversal experiments in LCPs.


  • Numerical study of stretched smectic-Aelastomer sheets

    by A. W. Brown, J. M. Adams

    We present a numerical study of stretching monodomain smectic-A elastomer sheets, computed using the finite element method. When stretched parallel to their smectic layer normal the smectic layers are unstable to a transition to a buckled state. We model macroscopic deformations by replacing the microscopic energy with a coarse grained effective free energy that accounts for the fine-scale layer buckling. We augment this model with a term to describe the energy of deforming buckled layers, which is necessary to reproduce the experimentally observed Poisson ratios postbuckling. We examine the spatial distribution of the microstructure phases for various stretching angles relative to the layer normal and for different length-to-width aspect ratios. When stretching parallel to the layer normal the majority of the sample forms a bidirectionally buckled microstructure, except at the clamps where a unidirectionally buckled microstructure is predicted. When stretching at small inclinations to the layer normal the phase of the sample is sensitive to the aspect ratio of the sample, with the bidirectionally buckled phase persistent to large angles only for small aspect ratios. We relate these theoretical results to experiments on smectic-A elastomers.


  • Tack energy and switchable adhesion of liquid crystal elastomers

    by D. R. Corbett, J. M. Adams

    The mechanical properties of liquid crystal elastomers (LCEs) make them suitable candidates for pressure-sensitive adhesives (PSAs). Using the nematic dumbbell constitutive model, and the block model of PSAs, we study their tack energy and the debonding process as could be measured experimentally in the probe-tack test. To investigate their performance as switchable PSAs we compare the tack energy for the director aligned parallel, and perpendicular to the substrate normal, with that for the isotropic state. We find that the tack energy is larger in the parallel alignment than the isotropic case by over a factor of two. The tack energy for the perpendicular alignment can be 50% less than the isotropic case. We propose a mechanism for reversibly switchable adhesion based on the reversibility of the isotropic to nematic transition. Finally we consider the influence of several material parameters that could be used to tune the stress–strain response.


  • Negative Poisson's ratio and semisoft elasticity of smectic-Cliquid-crystal elastomers

    by A. W. Brown, J. M. Adams

    Models of smectic-C liquid-crystal elastomers predict that they can display soft elasticity, in which the shape of the elastomer changes at no energy cost. The amplitude of the soft mode and the accompanying shears are dependent on the orientation of the layer normal and the director with respect to the stretch axis. We demonstrate that in some geometries the director is forced to rotate perpendicular to the stretch axis, causing lateral expansion of the sample—a negative Poisson's ratio. Current models do not include the effect of imperfections that must be present in the physical sample. We investigate the effect of a simple model of these imperfections on the soft modes in monodomain smectic-C elastomers in a variety of geometries. When stretching parallel to the layer normal (with imposed strain) the elastomer has a negative stiffness once the director starts to rotate. We show that this is a result of the negative Poisson's ratio in this geometry through a simple scalar model.


  • Transient shear banding in entangled polymers: A study using the Rolie-Poly model

    by J. M. Adams, S. M. Fielding, P. D. Olmsted

    Spatially inhomogeneous shear flow occurs in entangled polymer solutions, both as steady state shear banding and transiently after a large step strain or during startup to a steady uniform shear rate. Theoretically, steady state shear banding is a hallmark of models with a nonmonotonic constitutive relation between total shear stress and applied shear rate, but transient banding is sometimes seen in fluids that do not shear band at steady state. We model this behavior using the diffusive Rolie-Poly model in a Newtonian solvent, whose steady state constitutive behavior can be monotonic or nonmonotonic depending on the degree of convective constraint release. We study monotonic steady state constitutive behavior. Linear stability analysis of the startup to a sufficiently high shear rate shows that spatial fluctuations are unstable at early times. There is a strong correlation between this instability and the negative slope of the (time dependent) constitutive curve. If the time integral of the most unstable...


  • Strain analysis of a chiral smectic-Aelastomer

    by Christopher M. Spillmann, John H. Konnert, James M. Adams, Jeffrey R. Deschamps, Jawad Naciri, Banahalli R. Ratna

    We present a detailed analysis of the molecular packing of a strained liquid crystal elastomer composed of chiral mesogens in the smectic-A phase. X-ray diffraction patterns of the elastomer collected over a range of orientations with respect to the x-ray beam were used to reconstruct the three-dimensional scattering intensity as a function of tensile strain. We show that the smectic domain order is preserved in these strained elastomers. Changes in the intensity within a given scattering plane are due to reorientation, and not loss, of the molecular order in directions orthogonal to the applied strain. Incorporating the physical parameters of the elastomer, a nonlinear elastic model is presented to describe the rotation of the smectic-layered domains under strain, thus providing a fundamental analysis to the mechanical response of these unique materials.


  • Adams and Olmsted Reply:

    by J. M. Adams, P. D. Olmsted

    A Reply to the Comment by Shi-Qing Wang.