ERC FUNDED PROJECTS

Understanding the rules and mechanisms underlying memory formation, storage and retrieval is a grand challenge in neuroscience. In light of cumulating evidence regarding non-linear dendritic events (dendritic-spikes, branch strength potentiation, temporal sequence detection etc) together with activity-dependent rewiring of the connection matrix, the classical notion of information storage via Hebbian-like changes in synaptic connections is inadequate. While more recent plasticity theories consider non-linear dendritic properties, a unifying theory of how dendrites are utilized to achieve memory coding, storing and/or retrieval is cruelly missing. Using computational models, we will simulate memory processes in three key brain regions: the hippocampus, the amygdala and the prefrontal cortex. Models will incorporate biologically constrained dendrites and state-of-the-art plasticity rules and will span different levels of abstraction, ranging from detailed biophysical single neurons and circuits to integrate-and-fire networks and abstract theoretical models. Our main goal is to dissect the role of dendrites in information processing and storage across the three different regions by systematically altering their anatomical, biophysical and plasticity properties. Findings will further our understanding of the fundamental computations supported by these structures and how these computations, reinforced by plasticity mechanisms, sub-serve memory formation and associated dysfunctions, thus opening new avenues for hypothesis driven experimentation and development of novel treatments for memory-related diseases. Identification of dendrites as the key processing units across brain regions and complexity levels will lay the foundations for a new era in computational and experimental neuroscience and serve as the basis for groundbreaking advances in the robotics and artificial intelligence fields while also having a large impact on the machine learning community.

1 398 000 €

Start date: 2012-10-01, End date: 2017-09-30

Although the Internet is a great technological achievement, more than 40 years after its creation some of its original security and reliability problems remain unsolved. The root cause of these problems is the rigidity of the Internet architecture or in other words the Internet ossification problem, i.e., the basic architectural components of the Internet are set to stone and cannot be changed. The most ossified component of the Internet architecture is the inter-domain routing system. In this project, our goal is to address this challenge and to introduce a new Internet routing architecture that 1) enables innovation at the inter-domain level, 2) is backward-compatible with the present Internet architecture, and 3) provides concrete economic incentives for adopting it. We propose a new Internet routing paradigm based on a novel techno-economic framework, which exploits emerging technologies and meets these three goals. Our novel idea is that the combination of routing control logic outsourcing with Software Defined Networking (SDN) principles enables to innovate at the inter-domain level and therefore has the potential for a major break-through in the architecture of the Internet routing system. SDN is a rapidly emerging new computer networking architecture that makes the routing control plane of a network programmable. Based on our framework, we propose to design, build, and verify a better inter-domain routing system, which solves fundamental security, reliability, and manageability problems of the Internet architecture. Our work will be organized in four core topics 1) build a mutli-domain centralized routing control platform, 2) improve the reliability and security of the current inter-domain routing system, 3) design techniques for resolving tussles between competing network domains, 4) introduce advanced network monitoring and security techniques that intelligently correlate data from multiple domain to diagnose routing outages and attacks.

1 410 600 €

Start date: 2014-01-01, End date: 2018-12-31

An inflation-probing B-mode signal in the polarization of the cosmic microwave background (CMB) would be a discovery of utmost importance in physics. While such a signal is aggressively pursued by experiments around the world, recent Planck results have showed that this breakthrough is still out of reach, because of contamination from Galactic dust. To get to the primordial B-modes, we need to subtract polarized emission of magnetized interstellar dust with high accuracy. A critical piece of this puzzle is the 3D structure of the magnetic field threading dust clouds, which cannot be accessed through microwave observations alone, since they record integrated emission along the line of sight. Instead, observations of a large number of stars at known distances in optical polarization, tracing the same CMB-obscuring dust, can map the magnetic field between them. The Gaia mission is measuring distances to a billion stars, providing an opportunity to produce, the first-ever tomographic map of the Galactic magnetic field, using optical polarization of starlight. Such a map would not only boost CMB polarization foreground removal, but it would also have a profound impact in a wide range of astrophysical research, including interstellar medium physics, high-energy astrophysics, and galactic evolution. Taking advantage of our privately-funded, novel-technology, high-accuracy WALOP optopolarimeters currently under construction, we propose an ambitious optopolarimetric program of unprecedented scale that can meet this challenge: a survey of both northern and southern Galactic polar regions targeted by CMB experiments, covering >10,000 square degrees, which will measure linear optical polarization at 0.2% accuracy of over 360 stars per square degree (over 3.5M stars, a 1000-fold increase over the state of the art), combining wide-field-optimized instruments and an extraordinary commitment of observing time by Skinakas Observatory and the South African Astronomical Observatory.

1 887 500 €

Start date: 2018-06-01, End date: 2023-05-31