Telmo Menezes

Academic Researcher in the fields of Computational Social Sciences, Artificial Intelligence and Complex Systems, working at the Marc Bloch Centre in Berlin. PhD in Computer Science, specialization in Artificial Intelligence with a background of interdisciplinary research.

Previous industry experience includes working in Machine Learning at Amazon and developing software for NASA/JPL.

Below are some highlights of my research.

Knowledge Hypergraphs

Knowledge Hypergraph

Hypergraphs are a generalization of graphs (mathematical objects that represent networks), that allow for the representation of complex information structures, namely human knowledge and natural language communication. By employing modern Natural Language Processing techniques and general knowledge ontologies and databases, it becomes possible to transform free-form text into formal structures that are convenient both for AI and Social Sciences research.

The GraphBrain open-source research tool is being developed with the aim of both aiding in the research project, and making the techniques available to the scientific community.

GraphBrain project »

Network Generators and Artificial Scientists

Network Generators

Models explaining network morphology and growth processes permit a wide range of phenomena to be more systematically analyzed and understood. At the same time, creating such models is often challenging and requires insights that may be counter-intuitive. We have developed an approach to automatically detect realistic decentralized network growth models from empirical data, employing a machine learning technique inspired by natural selection and defining a unified formalism to describe such models as computer programs. The proposed method can be applied “out of the box” to any given network. We were able to find programs that are simple enough to lead to an actual understanding of underlaying mechanisms, namely for a simple brain and a social network.

Learn More »

Scales of Human Movement

Scales of Human Movement

Human mobility is known to be distributed across several orders of magnitude of physical distances, which makes it difficult to find or define typical and meaningful scales. Relying on geotagged data collected from photo-sharing social media, we apply community detection to movement networks constrained by increasing percentiles of the distance distribution. We discover clear phase transitions in the community partition space. This is the first known method of identifying natural scales of human movement. The partitions of the natural scales allow us to draw discrete multi-scale geographical boundaries, potentially capable of providing key insights in fields such as epidemiology or cultural contagion.

More details soon...

Evolutionary Multi-Agent Simulations

Multi-Agent simulation

Advances in fields such as Complexity Science and Artificial Life enable a modern direction in Computational Intelligence research. Instead of building isolated artificial intelligence systems from the top-down, this approach attempts to design systems where a population of agents and the environment interact and adaptation processes take place.

We created a novel artificial brain model, called gridbrain, as well as an evolutionary environment embedded in a simulation. The gridbrain model defines agent brains as heterogeneous networks of computational building blocks. A multi-layer approach allows gridbrains to process variable-sized information from several sensory channels. Computational building blocks allow for the use of base functionalities close to the underlying architecture of the digital computer. Evolutionary operators were devised to permit the adpative complexification of gridbrains.

In experimental scenarios that require cooperation, we demonstrate the emergence of synchronization behaviors that would be difficult to achieve under conventional approaches. Kin selection and group selection strategies compared. In a scenario where two species are in competition, we demonstrated the emergence of specialization niches without the need for geographical isolation.


Philosophy of AI

Evolutionary Superintelligences

There is overwhelming evidence that human intelligence is a product of Darwinian evolution. Investigating the consequences of self-modification, and more precisely, the consequences of utility function self-modification, leads to the stronger claim that not only human, but any form of intelligence is ultimately only possible within evolutionary processes. Human-designed artificial intelligences can only remain stable until they discover how to manipulate their own utility function. By definition, a human designer cannot prevent a superhuman intelligence from modifying itself, even if protection mechanisms against this action are put in place. Without evolutionary pressure, sufficiently advanced artificial intelligences become inert by simplifying their own utility function. Within evolutionary processes, the implicit utility function is always reducible to persistence, and the control of superhuman intelligences embedded in evolutionary processes is not possible. Mechanisms against utility function self-modification are ultimately futile. Instead, scientific effort toward the mitigation of existential risks from the development of superintelligences should be in two directions: understanding consciousness, and the complex dynamics of evolutionary systems.

View working paper »