**This is a summary of some additional information about me and my research**
(this page is work in progress)
I am an Associate Professor in Computer Science at the University of Southampton. Before that, I graduated with a Diploma and then PhD in theoretical Physics from the University of Leipzig and then worked as Postdoctoral Fellow and later Research Scientist at CSIRO Atmospheric Research in Canberra. My research interests revolve around complex systems, in particular the dynamics and structure of complex networks but also other related applications of non-linear dynamics or statistics. If you are interested in studying for a PhD in any of the areas related to my description of research interest below, get in touch!
(Recent) Research interests
- Opinion formation and control on complex networks (L. Tranh-Thanh, S. Stein, V. Restocchi, S. Chakraborty, G. Moreno) -- Consider a social network with agents holding opinions that change subject to peer influence (i.e. depending on the opinions of their network neighbours). Suppose you are in control of a limited budget that allows you to change the opinion of a certain number of agents in the social network. Which agents/nodes should you target? We are interested in this problem mostly in situations in which agents can dynamically change their opinions back and forth subject to stochastic updating rules. This is a general problem with obvious applications in advertising, political campaigns, government propaganda, but also with a relationship to studies of foreign influence in political elections or the spread of fake news.
- Maximizing influence on complex networks subject to limited budgets -- In general, the influence maximization problem is a complex optimization problem. What are the best ways to solve, approximate, or simplify it? Are there general rules of thumb?
1. In a variant of the voter dynamics we have investigated a scenarios in which agents resist the influence of an external controller. Our recent results how that the optimal control strategy for the external influencer switches from hub control to periphery control if resistance is larger than a critical threshold [#Brede2018a].
2. We have investigated the scenario of external control subject to finite time horizons and have shown that optimal control strategies generally depend on the time horizon of the controller [#Brede2018b].
3. Under which conditions should an optimal controller focus on zealots? (G. Moreno) -- Social systems are not homogenous and some people have inherent bias for certain opinions or change their opinions more reluctantly than others. How does such social heterogeneity influence rules for optimal control?
- Relationships between the structure of networks and their susceptibility to opinion control
- Control with continuously varying strength (S. Chakraborty and S. Stein) -- Most models of opinion control in previous literature have assumed that control is binary -- a node is either being influenced/controlled or not. Here, for the voting dynamics, we study scenarios in which control strengths can vary continuously.
- Adversarial scenarios of network control -- How do rules for optimal control change if more than one party strifes for control?
1. We study the voter dynamics on star networks in adversarial settings (with S. Stein and S. Chakraborty)
2. In an adaptive network model of competitive control I show that in bounded confidence models of continuous incremental opinion change a simultaneous dynamics of network adaptation inspired by agents attempting to maximize their influence on the system can either lead to radicalization and polarization or speed up transients and lead to sharper compromise consensus states (in submission).
- Lanchester dynamics and military applications (with A. Kalloniatis) -- Conflict between two forces (or in general network attack subject to protracted dynamics of attrition) can be modeled by the Lanchester dynamics on a multi-layer network in which intra-layers might represent the logistics of force movement and manoevre and inter-layers model military engagements. Can we design manoeuvre or logistic layers in a way that one force can gain an advantage?
- Synchronization and the Kuramoto model on complex networks (mostly with A. Kalloniatis)
- A general question of interest is the relationship between the structure of a network and its propensity for synchronization. In systems of non-identical Kuramoto oscillators there are well-known rules for optimal designs of network arrangements that promote synchronizations. In a recent paper we have shown that such configurations can be approached in a distributed way in which all nodes (or oscillators) simultaneously strife to maximize their impact on the dynamics [#Brede2018c].
- Perfect phase synchronization in systems of non-identical oscillators can generally only be achieved in the limit of infinite coupling. An interesting modification of the Kuramoto dynamics is to include effects if delays in coupling, the so-called Kuramoto-Sakaguchi model. Uncorrelated delays typically lead to frustration, thus making synchronization much harder or even impossible. In recent work we have shown that appropriately tuned delays can allow for perfect phase synchronization [#Brede2016a].
- In follow-up work we have exploited the above idea for an application of synchronization control in which we show that a scheme of control via tuned lags can be more efficient than standard pinning control [#Brede2017a] (and in submission).
- Interdisciplinary applications of network science to data mining
- Exploring the structure of language with multiplex networks (with M. Stella)
1. In a model of language learning we have shown that multiplex representations of language carry more information about language learning than single layer models [#Stella2017a].
2. In multilayer models of words of learned words at certain ages we have demonstrated the existence of an abrupt transition in the size of a core of the language at a certain age [#Stella2018a].
- Exploring the structure of twitter communities to learn about eating disorders (with T. Wang, E. Mentzakis, and A. Ianni)
- Modelling the structure, stability, and regulation of banking systems (with R. de Caux and F. McGroarty)
- Modeling (ancient) societies (with S. Roman)
- Other applications of complex systems
- Statistics of aggregation and disperal of fish in the North Sea (with M. Cobain and C. Trueman)
- Evolutionary transitions and specialization (with S. Tudge and R. Watson)
[#Brede2018a]: Brede, Markus, Restocchi, Valerio and Stein, Sebastian (2018) Resisting influence: How the strength of predispositions to resist control can change strategies for optimal opinion control in the voter model. Frontiers in Robotics and AI, 5 (34). (http://dx.doi.org/10.3389/frobt.2018.00034 ).
[#Brede2018b]: Brede, Markus, Restocchi, Valerio and Stein, Sebastian (2018) Effects of time horizons on Influence maximization in the voter dynamics. Journal of Complex Networks. (https://doi.org/10.1093/comnet/cny027 ).
[#Brede2018c]: Brede, Markus, Stella, Massimo and Kalloniatis, Alexander C. (2018) Competitive influence maximization and enhancement of synchronization in populations of non-identical Kuramoto oscillators. Scientific Reports, 8. (http://dx.doi.org/10.1038/s41598-017-18961-z ).
[#Stella2018a]: Stella, Massimo, Beckage, Nicole, Brede, Markus and De Domenico, Manlio (2018) Multiplex model of mental lexicon reveals explosive learning in humans. Scientific Reports, 8, 1-11. (https://doi.org/10.1038/s41598-018-20730-5 ).
[#Brede2017a]: Brede, Markus and Kalloniatis, Alexander (2017) Controlling synchronisation through adaptive phase lags. In Proceedings of the 8th International Conference on Physics and Control (PhysCon 2017). 8 pp.
[#Stella2017a]: Stella, Massimo, Beckage, Nicole and Brede, Markus (2017) Multiplex lexical networks reveal patterns in early word acquisition in children. Scientific Reports, 7. (http://dx.doi.org/10.1038/srep46730 ).
[#Brede2016a]: Brede, Markus and Alexander, Kalloniatis (2016) Frustration tuning and perfect phase synchronization in the Kuramoto-Sakaguchi model. Physical Review E, 93 (6), 1-13. (http://dx.doi.org/10.1103/PhysRevE.93.062315 ).