Project acronym AlgTateGro
Project Constructing line bundles on algebraic varieties --around conjectures of Tate and Grothendieck
Researcher (PI) François CHARLES
Host Institution (HI) UNIVERSITE PARIS-SUD
Call Details Starting Grant (StG), PE1, ERC-2016-STG
Summary The goal of this project is to investigate two conjectures in arithmetic geometry pertaining to the geometry of projective varieties over finite and number fields. These two conjectures, formulated by Tate and Grothendieck in the 1960s, predict which cohomology classes are chern classes of line bundles. They both form an arithmetic counterpart of a theorem of Lefschetz, proved in the 1940s, which itself is the only known case of the Hodge conjecture. These two long-standing conjectures are one of the aspects of a more general web of questions regarding the topology of algebraic varieties which have been emphasized by Grothendieck and have since had a central role in modern arithmetic geometry. Special cases of these conjectures, appearing for instance in the work of Tate, Deligne, Faltings, Schneider-Lang, Masser-Wüstholz, have all had important consequences.
My goal is to investigate different lines of attack towards these conjectures, building on recent work on myself and Jean-Benoît Bost on related problems. The two main directions of the proposal are as follows. Over finite fields, the Tate conjecture is related to finiteness results for certain cohomological objects. I want to understand how to relate these to hidden boundedness properties of algebraic varieties that have appeared in my recent geometric proof of the Tate conjecture for K3 surfaces. The existence and relevance of a theory of Donaldson invariants for moduli spaces of twisted sheaves over finite fields seems to be a promising and novel direction. Over number fields, I want to combine the geometric insight above with algebraization techniques developed by Bost. In a joint project, we want to investigate how these can be used to first understand geometrically major results in transcendence theory and then attack the Grothendieck period conjecture for divisors via a number-theoretic and complex-analytic understanding of universal vector extensions of abelian schemes over curves.
Summary
The goal of this project is to investigate two conjectures in arithmetic geometry pertaining to the geometry of projective varieties over finite and number fields. These two conjectures, formulated by Tate and Grothendieck in the 1960s, predict which cohomology classes are chern classes of line bundles. They both form an arithmetic counterpart of a theorem of Lefschetz, proved in the 1940s, which itself is the only known case of the Hodge conjecture. These two long-standing conjectures are one of the aspects of a more general web of questions regarding the topology of algebraic varieties which have been emphasized by Grothendieck and have since had a central role in modern arithmetic geometry. Special cases of these conjectures, appearing for instance in the work of Tate, Deligne, Faltings, Schneider-Lang, Masser-Wüstholz, have all had important consequences.
My goal is to investigate different lines of attack towards these conjectures, building on recent work on myself and Jean-Benoît Bost on related problems. The two main directions of the proposal are as follows. Over finite fields, the Tate conjecture is related to finiteness results for certain cohomological objects. I want to understand how to relate these to hidden boundedness properties of algebraic varieties that have appeared in my recent geometric proof of the Tate conjecture for K3 surfaces. The existence and relevance of a theory of Donaldson invariants for moduli spaces of twisted sheaves over finite fields seems to be a promising and novel direction. Over number fields, I want to combine the geometric insight above with algebraization techniques developed by Bost. In a joint project, we want to investigate how these can be used to first understand geometrically major results in transcendence theory and then attack the Grothendieck period conjecture for divisors via a number-theoretic and complex-analytic understanding of universal vector extensions of abelian schemes over curves.
Max ERC Funding
1 222 329 €
Duration
Start date: 2016-12-01, End date: 2021-11-30
Project acronym altEJrepair
Project Characterisation of DNA Double-Strand Break Repair by Alternative End-Joining: Potential Targets for Cancer Therapy
Researcher (PI) Raphael CECCALDI
Host Institution (HI) INSTITUT CURIE
Call Details Starting Grant (StG), LS1, ERC-2016-STG
Summary DNA repair pathways evolved as an intricate network that senses DNA damage and resolves it in order to minimise genetic lesions and thus preventing tumour formation. Gaining in recognition the last few years, the alternative end-joining (alt-EJ) DNA repair pathway was recently shown to be up-regulated and required for cancer cell viability in the absence of homologous recombination-mediated repair (HR). Despite this integral role, the alt-EJ repair pathway remains poorly characterised in humans. As such, its molecular composition, regulation and crosstalk with HR and other repair pathways remain elusive. Additionally, the contribution of the alt-EJ pathway to tumour progression as well as the identification of a mutational signature associated with the use of alt-EJ has not yet been investigated. Moreover, the clinical relevance of developing small-molecule inhibitors targeting players in the alt-EJ pathway, such as the polymerase Pol Theta (Polθ), is of importance as current anticancer drug treatments have shown limited effectiveness in achieving cancer remission in patients with HR-deficient (HRD) tumours.
Here, we propose a novel, multidisciplinary approach that aims to characterise the players and mechanisms of action involved in the utilisation of alt-EJ in cancer. This understanding will better elucidate the changing interplay between different DNA repair pathways, thus shedding light on whether and how the use of alt-EJ contributes to the pathogenic history and survival of HRD tumours, eventually paving the way for the development of novel anticancer therapeutics.
For all the abovementioned reasons, we are convinced this project will have important implications in: 1) elucidating critical interconnections between DNA repair pathways, 2) improving the basic understanding of the composition, regulation and function of the alt-EJ pathway, and 3) facilitating the development of new synthetic lethality-based chemotherapeutics for the treatment of HRD tumours.
Summary
DNA repair pathways evolved as an intricate network that senses DNA damage and resolves it in order to minimise genetic lesions and thus preventing tumour formation. Gaining in recognition the last few years, the alternative end-joining (alt-EJ) DNA repair pathway was recently shown to be up-regulated and required for cancer cell viability in the absence of homologous recombination-mediated repair (HR). Despite this integral role, the alt-EJ repair pathway remains poorly characterised in humans. As such, its molecular composition, regulation and crosstalk with HR and other repair pathways remain elusive. Additionally, the contribution of the alt-EJ pathway to tumour progression as well as the identification of a mutational signature associated with the use of alt-EJ has not yet been investigated. Moreover, the clinical relevance of developing small-molecule inhibitors targeting players in the alt-EJ pathway, such as the polymerase Pol Theta (Polθ), is of importance as current anticancer drug treatments have shown limited effectiveness in achieving cancer remission in patients with HR-deficient (HRD) tumours.
Here, we propose a novel, multidisciplinary approach that aims to characterise the players and mechanisms of action involved in the utilisation of alt-EJ in cancer. This understanding will better elucidate the changing interplay between different DNA repair pathways, thus shedding light on whether and how the use of alt-EJ contributes to the pathogenic history and survival of HRD tumours, eventually paving the way for the development of novel anticancer therapeutics.
For all the abovementioned reasons, we are convinced this project will have important implications in: 1) elucidating critical interconnections between DNA repair pathways, 2) improving the basic understanding of the composition, regulation and function of the alt-EJ pathway, and 3) facilitating the development of new synthetic lethality-based chemotherapeutics for the treatment of HRD tumours.
Max ERC Funding
1 498 750 €
Duration
Start date: 2017-07-01, End date: 2022-06-30
Project acronym AQUARAMAN
Project Pipet Based Scanning Probe Microscopy Tip-Enhanced Raman Spectroscopy: A Novel Approach for TERS in Liquids
Researcher (PI) Aleix Garcia Guell
Host Institution (HI) ECOLE POLYTECHNIQUE
Call Details Starting Grant (StG), PE4, ERC-2016-STG
Summary Tip-enhanced Raman spectroscopy (TERS) is often described as the most powerful tool for optical characterization of surfaces and their proximities. It combines the intrinsic spatial resolution of scanning probe techniques (AFM or STM) with the chemical information content of vibrational Raman spectroscopy. Capable to reveal surface heterogeneity at the nanoscale, TERS is currently playing a fundamental role in the understanding of interfacial physicochemical processes in key areas of science and technology such as chemistry, biology and material science.
Unfortunately, the undeniable potential of TERS as a label-free tool for nanoscale chemical and structural characterization is, nowadays, limited to air and vacuum environments, with it failing to operate in a reliable and systematic manner in liquid. The reasons are more technical than fundamental, as what is hindering the application of TERS in water is, among other issues, the low stability of the probes and their consistency. Fields of science and technology where the presence of water/electrolyte is unavoidable, such as biology and electrochemistry, remain unexplored with this powerful technique.
We propose a revolutionary approach for TERS in liquids founded on the employment of pipet-based scanning probe microscopy techniques (pb-SPM) as an alternative to AFM and STM. The use of recent but well established pb-SPM brings the opportunity to develop unprecedented pipet-based TERS probes (beyond the classic and limited metallized solid probes from AFM and STM), together with the implementation of ingenious and innovative measures to enhance tip stability, sensitivity and reliability, unattainable with the current techniques.
We will be in possession of a unique nano-spectroscopy platform capable of experiments in liquids, to follow dynamic processes in-situ, addressing fundamental questions and bringing insight into interfacial phenomena spanning from materials science, physics, chemistry and biology.
Summary
Tip-enhanced Raman spectroscopy (TERS) is often described as the most powerful tool for optical characterization of surfaces and their proximities. It combines the intrinsic spatial resolution of scanning probe techniques (AFM or STM) with the chemical information content of vibrational Raman spectroscopy. Capable to reveal surface heterogeneity at the nanoscale, TERS is currently playing a fundamental role in the understanding of interfacial physicochemical processes in key areas of science and technology such as chemistry, biology and material science.
Unfortunately, the undeniable potential of TERS as a label-free tool for nanoscale chemical and structural characterization is, nowadays, limited to air and vacuum environments, with it failing to operate in a reliable and systematic manner in liquid. The reasons are more technical than fundamental, as what is hindering the application of TERS in water is, among other issues, the low stability of the probes and their consistency. Fields of science and technology where the presence of water/electrolyte is unavoidable, such as biology and electrochemistry, remain unexplored with this powerful technique.
We propose a revolutionary approach for TERS in liquids founded on the employment of pipet-based scanning probe microscopy techniques (pb-SPM) as an alternative to AFM and STM. The use of recent but well established pb-SPM brings the opportunity to develop unprecedented pipet-based TERS probes (beyond the classic and limited metallized solid probes from AFM and STM), together with the implementation of ingenious and innovative measures to enhance tip stability, sensitivity and reliability, unattainable with the current techniques.
We will be in possession of a unique nano-spectroscopy platform capable of experiments in liquids, to follow dynamic processes in-situ, addressing fundamental questions and bringing insight into interfacial phenomena spanning from materials science, physics, chemistry and biology.
Max ERC Funding
1 528 442 €
Duration
Start date: 2017-07-01, End date: 2022-06-30
Project acronym BrainConquest
Project Boosting Brain-Computer Communication with high Quality User Training
Researcher (PI) Fabien LOTTE
Host Institution (HI) INSTITUT NATIONAL DE RECHERCHE ENINFORMATIQUE ET AUTOMATIQUE
Call Details Starting Grant (StG), PE7, ERC-2016-STG
Summary Brain-Computer Interfaces (BCIs) are communication systems that enable users to send commands to computers through brain signals only, by measuring and processing these signals. Making computer control possible without any physical activity, BCIs have promised to revolutionize many application areas, notably assistive technologies, e.g., for wheelchair control, and human-machine interaction. Despite this promising potential, BCIs are still barely used outside laboratories, due to their current poor reliability. For instance, BCIs only using two imagined hand movements as mental commands decode, on average, less than 80% of these commands correctly, while 10 to 30% of users cannot control a BCI at all.
A BCI should be considered a co-adaptive communication system: its users learn to encode commands in their brain signals (with mental imagery) that the machine learns to decode using signal processing. Most research efforts so far have been dedicated to decoding the commands. However, BCI control is a skill that users have to learn too. Unfortunately how BCI users learn to encode the commands is essential but is barely studied, i.e., fundamental knowledge about how users learn BCI control is lacking. Moreover standard training approaches are only based on heuristics, without satisfying human learning principles. Thus, poor BCI reliability is probably largely due to highly suboptimal user training.
In order to obtain a truly reliable BCI we need to completely redefine user training approaches. To do so, I propose to study and statistically model how users learn to encode BCI commands. Then, based on human learning principles and this model, I propose to create a new generation of BCIs which ensure that users learn how to successfully encode commands with high signal-to-noise ratio in their brain signals, hence making BCIs dramatically more reliable. Such a reliable BCI could positively change human-machine interaction as BCIs have promised but failed to do so far.
Summary
Brain-Computer Interfaces (BCIs) are communication systems that enable users to send commands to computers through brain signals only, by measuring and processing these signals. Making computer control possible without any physical activity, BCIs have promised to revolutionize many application areas, notably assistive technologies, e.g., for wheelchair control, and human-machine interaction. Despite this promising potential, BCIs are still barely used outside laboratories, due to their current poor reliability. For instance, BCIs only using two imagined hand movements as mental commands decode, on average, less than 80% of these commands correctly, while 10 to 30% of users cannot control a BCI at all.
A BCI should be considered a co-adaptive communication system: its users learn to encode commands in their brain signals (with mental imagery) that the machine learns to decode using signal processing. Most research efforts so far have been dedicated to decoding the commands. However, BCI control is a skill that users have to learn too. Unfortunately how BCI users learn to encode the commands is essential but is barely studied, i.e., fundamental knowledge about how users learn BCI control is lacking. Moreover standard training approaches are only based on heuristics, without satisfying human learning principles. Thus, poor BCI reliability is probably largely due to highly suboptimal user training.
In order to obtain a truly reliable BCI we need to completely redefine user training approaches. To do so, I propose to study and statistically model how users learn to encode BCI commands. Then, based on human learning principles and this model, I propose to create a new generation of BCIs which ensure that users learn how to successfully encode commands with high signal-to-noise ratio in their brain signals, hence making BCIs dramatically more reliable. Such a reliable BCI could positively change human-machine interaction as BCIs have promised but failed to do so far.
Max ERC Funding
1 498 751 €
Duration
Start date: 2017-07-01, End date: 2022-06-30
Project acronym BrainDyn
Project Tracking information flow in the brain: A unified and general framework for dynamic communication in brain networks
Researcher (PI) Mathilde BONNEFOND
Host Institution (HI) INSTITUT NATIONAL DE LA SANTE ET DE LA RECHERCHE MEDICALE
Call Details Starting Grant (StG), LS5, ERC-2016-STG
Summary The brain is composed of a set of areas specialized in specific computations whose outputs need to be transferred to other specialized areas for cognition to emerge. To account for context-dependent behaviors, the information has to be flexibly routed through the fixed anatomy of the brain. The aim of my proposal is to test a general framework for flexible communication between brain areas based on nested oscillations which I recently developed. The general idea is that internally-driven slow oscillations (<20Hz) either set-up or prevent the communication between brain areas. Stimulus-driven gamma oscillations (>30Hz), nested in the slow oscillations, can then be directed to task-relevant areas of the network. I plan to use a multimodal, multi-scale and transversal (human and monkey) approach in experiments manipulating visual processing, attention and memory to test core predictions of my framework. The theoretical approach and the methodological development used in my project will provide the basis for future fundamental and clinical research.
Summary
The brain is composed of a set of areas specialized in specific computations whose outputs need to be transferred to other specialized areas for cognition to emerge. To account for context-dependent behaviors, the information has to be flexibly routed through the fixed anatomy of the brain. The aim of my proposal is to test a general framework for flexible communication between brain areas based on nested oscillations which I recently developed. The general idea is that internally-driven slow oscillations (<20Hz) either set-up or prevent the communication between brain areas. Stimulus-driven gamma oscillations (>30Hz), nested in the slow oscillations, can then be directed to task-relevant areas of the network. I plan to use a multimodal, multi-scale and transversal (human and monkey) approach in experiments manipulating visual processing, attention and memory to test core predictions of my framework. The theoretical approach and the methodological development used in my project will provide the basis for future fundamental and clinical research.
Max ERC Funding
1 333 718 €
Duration
Start date: 2017-02-01, End date: 2022-01-31
Project acronym CollectSwim
Project Individual and Collective Swimming of Active Microparticles
Researcher (PI) Sebastien MICHELIN
Host Institution (HI) ECOLE POLYTECHNIQUE
Call Details Starting Grant (StG), PE8, ERC-2016-STG
Summary Bacteria are tiny; yet their collective dynamics generate large-scale flows and profoundly modify a fluid’s viscosity or diffusivity. So do autophoretic microswimmers, an example of active microscopic particles that draw their motion from physico-chemical exchanges with their environment. How do such ``active fluids'' turn individual microscopic propulsion into macroscopic fluid dynamics? What controls this self-organization process? These are fundamental questions for biologists but also for engineers, to use these suspensions for mixing or chemical sensing and, more generally, for creating active fluids whose macroscopic physical properties can be controlled precisely.
Self-propulsion of autophoretic swimmers was reported only recently. Major scientific gaps impair the quantitative understanding of their individual and collective dynamics, which is required to exploit these active fluids. Existing models scarcely account for important experimental characteristics such as complex hydrodynamics, physico-chemical processes and confinement. Thus, these models cannot yet be used as predictive tools, even at the individual level.
Further, to use phoretic suspensions as active fluids with microscopically-controlled properties, quantitatively-predictive models are needed for the collective dynamics. Instead of ad-hoc interaction rules, collective models must be built on a detailed physico-mechanical description of each swimmer’s interaction with its environment.
This project will develop these tools and validate them against experimental data. This requires overcoming several major challenges: the diversity of electro-chemical processes, the confined geometry, the large number of particles, and the plurality of interaction mechanisms and their nonlinear coupling.
To address these issues, rigorous physical, mathematical and numerical models will be developed to obtain a complete multi-scale description of the individual and collective dynamics of active particles.
Summary
Bacteria are tiny; yet their collective dynamics generate large-scale flows and profoundly modify a fluid’s viscosity or diffusivity. So do autophoretic microswimmers, an example of active microscopic particles that draw their motion from physico-chemical exchanges with their environment. How do such ``active fluids'' turn individual microscopic propulsion into macroscopic fluid dynamics? What controls this self-organization process? These are fundamental questions for biologists but also for engineers, to use these suspensions for mixing or chemical sensing and, more generally, for creating active fluids whose macroscopic physical properties can be controlled precisely.
Self-propulsion of autophoretic swimmers was reported only recently. Major scientific gaps impair the quantitative understanding of their individual and collective dynamics, which is required to exploit these active fluids. Existing models scarcely account for important experimental characteristics such as complex hydrodynamics, physico-chemical processes and confinement. Thus, these models cannot yet be used as predictive tools, even at the individual level.
Further, to use phoretic suspensions as active fluids with microscopically-controlled properties, quantitatively-predictive models are needed for the collective dynamics. Instead of ad-hoc interaction rules, collective models must be built on a detailed physico-mechanical description of each swimmer’s interaction with its environment.
This project will develop these tools and validate them against experimental data. This requires overcoming several major challenges: the diversity of electro-chemical processes, the confined geometry, the large number of particles, and the plurality of interaction mechanisms and their nonlinear coupling.
To address these issues, rigorous physical, mathematical and numerical models will be developed to obtain a complete multi-scale description of the individual and collective dynamics of active particles.
Max ERC Funding
1 497 698 €
Duration
Start date: 2017-09-01, End date: 2022-08-31
Project acronym CombiTop
Project New Interactions of Combinatorics through Topological Expansions, at the crossroads of Probability, Graph theory, and Mathematical Physics
Researcher (PI) Guillaume CHAPUY
Host Institution (HI) CENTRE NATIONAL DE LA RECHERCHE SCIENTIFIQUE CNRS
Call Details Starting Grant (StG), PE1, ERC-2016-STG
Summary "The purpose of this project is to use the ubiquitous nature of certain combinatorial topological objects called maps in order to unveil deep connections between several areas of mathematics. Maps, that describe the embedding of a graph into a surface, appear in probability theory, mathematical physics, enumerative geometry or graph theory, and different combinatorial viewpoints on these objects have been developed in connection with each topic. The originality of our project will be to study these approaches together and to unify them.
The outcome will be triple, as we will:
1. build a new, well structured branch of combinatorics of which many existing results in different areas of enumerative and algebraic combinatorics are only first fruits;
2. connect and unify several aspects of the domains related to it, most importantly between probability and integrable hierarchies thus proposing new directions, new tools and new results for each of them;
3. export the tools of this unified framework to reach at new applications, especially in random graph theory and in a rising domain of algebraic combinatorics related to Tamari lattices.
The methodology to reach the unification will be the study of some strategic interactions at different places involving topological expansions, that is to say, places where enumerative problems dealing with maps appear and their genus invariant plays a natural role, in particular: 1. the combinatorial theory of maps developped by the "French school" of combinatorics, and the study of random maps; 2. the combinatorics of Fermions underlying the theory of KP and 2-Toda hierarchies; 3; the Eynard-Orantin "topological recursion" coming from mathematical physics.
We present some key set of tasks in view of relating these different topics together. The pertinence of the approach is demonstrated by recent research of the principal investigator."
Summary
"The purpose of this project is to use the ubiquitous nature of certain combinatorial topological objects called maps in order to unveil deep connections between several areas of mathematics. Maps, that describe the embedding of a graph into a surface, appear in probability theory, mathematical physics, enumerative geometry or graph theory, and different combinatorial viewpoints on these objects have been developed in connection with each topic. The originality of our project will be to study these approaches together and to unify them.
The outcome will be triple, as we will:
1. build a new, well structured branch of combinatorics of which many existing results in different areas of enumerative and algebraic combinatorics are only first fruits;
2. connect and unify several aspects of the domains related to it, most importantly between probability and integrable hierarchies thus proposing new directions, new tools and new results for each of them;
3. export the tools of this unified framework to reach at new applications, especially in random graph theory and in a rising domain of algebraic combinatorics related to Tamari lattices.
The methodology to reach the unification will be the study of some strategic interactions at different places involving topological expansions, that is to say, places where enumerative problems dealing with maps appear and their genus invariant plays a natural role, in particular: 1. the combinatorial theory of maps developped by the "French school" of combinatorics, and the study of random maps; 2. the combinatorics of Fermions underlying the theory of KP and 2-Toda hierarchies; 3; the Eynard-Orantin "topological recursion" coming from mathematical physics.
We present some key set of tasks in view of relating these different topics together. The pertinence of the approach is demonstrated by recent research of the principal investigator."
Max ERC Funding
1 086 125 €
Duration
Start date: 2017-03-01, End date: 2022-02-28
Project acronym CONNEXIO
Project Physiologically relevant microfluidic neuro-engineering
Researcher (PI) Thibault Frédéric Johan HONEGGER
Host Institution (HI) CENTRE NATIONAL DE LA RECHERCHE SCIENTIFIQUE CNRS
Call Details Starting Grant (StG), PE7, ERC-2016-STG
Summary Developing minimalistic biological neural networks and observing their functional activity is crucial to decipher the information processing in the brain. This project aims to address two major challenges: to design and fabricate in vitro biological neural networks that are organized in physiological relevant ways and to provide a label-free monitoring platform capable of observing neural activity both at the neuron resolution and at large fields of view. To do so, the project will develop a unique microfluidic compartmentalized chips where populations of primary neurons will be seeded in deposition chambers with physiological relevant number and densities. Chambers will be connected by microgrooves in which neurites only can grow and whose dimensions will be tuned according to the connectivity pattern to reproduce. To observe the activity of such complex neural networks, we will develop a disruptive observation technique that will transduce the electrical activity of spiking neurons into optical differences observed on a lens-free platform, without calcium labelling and constantly in-incubo. By combining neuro-engineering patterning and the lens-free platform, we will compare individual spiking to global oscillators in basic neural networks under localized external stimulations. Such results will provide experimental insight into computational neuroscience current approaches. Finally, we will design an in vitro network that will reproduce a neural loop implied in major neurodegenerative diseases with physiological relevant neural types, densities and connectivities. This circuitry will be manipulated in order to model Huntington and Parkinson diseases on the chip and assess the impact of known drugs on the functional activity of the entire network. This project will engineer microfluidics chips with physiological relevant neural network and a lensfree activity monitoring platform to answer fundamental and clinically relevant issues in neuroscience.
Summary
Developing minimalistic biological neural networks and observing their functional activity is crucial to decipher the information processing in the brain. This project aims to address two major challenges: to design and fabricate in vitro biological neural networks that are organized in physiological relevant ways and to provide a label-free monitoring platform capable of observing neural activity both at the neuron resolution and at large fields of view. To do so, the project will develop a unique microfluidic compartmentalized chips where populations of primary neurons will be seeded in deposition chambers with physiological relevant number and densities. Chambers will be connected by microgrooves in which neurites only can grow and whose dimensions will be tuned according to the connectivity pattern to reproduce. To observe the activity of such complex neural networks, we will develop a disruptive observation technique that will transduce the electrical activity of spiking neurons into optical differences observed on a lens-free platform, without calcium labelling and constantly in-incubo. By combining neuro-engineering patterning and the lens-free platform, we will compare individual spiking to global oscillators in basic neural networks under localized external stimulations. Such results will provide experimental insight into computational neuroscience current approaches. Finally, we will design an in vitro network that will reproduce a neural loop implied in major neurodegenerative diseases with physiological relevant neural types, densities and connectivities. This circuitry will be manipulated in order to model Huntington and Parkinson diseases on the chip and assess the impact of known drugs on the functional activity of the entire network. This project will engineer microfluidics chips with physiological relevant neural network and a lensfree activity monitoring platform to answer fundamental and clinically relevant issues in neuroscience.
Max ERC Funding
1 727 731 €
Duration
Start date: 2016-11-01, End date: 2021-10-31
Project acronym CTO Com
Project Context- and Task-Oriented Communication
Researcher (PI) Michèle WIGGER
Host Institution (HI) INSTITUT MINES-TELECOM
Call Details Starting Grant (StG), PE7, ERC-2016-STG
Summary Emergence of a large number of distributed decision and control systems (e.g., in health care, transportation, and energy management), combined with increasing demands of traditional communications (e.g., due to multiview videos), create an imminent need for highly improved communication systems. We advocate that—combined with improvements in battery, antenna, and chip technologies—context- and/or task-oriented communication techniques will bring the desired breakthrough. Specifically, context- oriented techniques will greatly improve performance, because future networks have complex infrastructures (with cache-memories, cloud-RANs, etc.) allowing the terminals to collect side-informations about other terminals’ data or signals, and because many distributed decision systems rely on numerous devices with correlated measurements. Task-oriented techniques promise even larger gains, especially in distributed decision systems where decisions take value on a small range, and thus the traditional approach of communicating sequences of observed signals results in a huge overhead.
Information theory, and in particular distributed joint source-channel coding, provides a general framework for designing context-oriented communication techniques. Such a general framework is missing for task-oriented communication. Previous results indicate that creative usages of information theory on its frontier to statistics and decision theory are well-suited for designing task-oriented communication techniques for applications as diverse as coordination of smart devices, distributed hypothesis testing, and clustering of data.
Our goal is to design context- and/or task-oriented communication techniques for these three applications and for cache-aided communication. Besides the high gains that our new techniques bring directly to these applications, the complementarity of our applications and obtained results will facilitate a future general framework for context- and task-oriented communication.
Summary
Emergence of a large number of distributed decision and control systems (e.g., in health care, transportation, and energy management), combined with increasing demands of traditional communications (e.g., due to multiview videos), create an imminent need for highly improved communication systems. We advocate that—combined with improvements in battery, antenna, and chip technologies—context- and/or task-oriented communication techniques will bring the desired breakthrough. Specifically, context- oriented techniques will greatly improve performance, because future networks have complex infrastructures (with cache-memories, cloud-RANs, etc.) allowing the terminals to collect side-informations about other terminals’ data or signals, and because many distributed decision systems rely on numerous devices with correlated measurements. Task-oriented techniques promise even larger gains, especially in distributed decision systems where decisions take value on a small range, and thus the traditional approach of communicating sequences of observed signals results in a huge overhead.
Information theory, and in particular distributed joint source-channel coding, provides a general framework for designing context-oriented communication techniques. Such a general framework is missing for task-oriented communication. Previous results indicate that creative usages of information theory on its frontier to statistics and decision theory are well-suited for designing task-oriented communication techniques for applications as diverse as coordination of smart devices, distributed hypothesis testing, and clustering of data.
Our goal is to design context- and/or task-oriented communication techniques for these three applications and for cache-aided communication. Besides the high gains that our new techniques bring directly to these applications, the complementarity of our applications and obtained results will facilitate a future general framework for context- and task-oriented communication.
Max ERC Funding
1 495 288 €
Duration
Start date: 2017-05-01, End date: 2022-04-30
Project acronym D3
Project Interpreting Drawings for 3D Design
Researcher (PI) Adrien BOUSSEAU
Host Institution (HI) INSTITUT NATIONAL DE RECHERCHE ENINFORMATIQUE ET AUTOMATIQUE
Call Details Starting Grant (StG), PE6, ERC-2016-STG
Summary Designers draw extensively to externalize their ideas and communicate with others. However, drawings are currently not directly interpretable by computers. To test their ideas against physical reality, designers have to create 3D models suitable for simulation and 3D printing. However, the visceral and approximate nature of drawing clashes with the tediousness and rigidity of 3D modeling. As a result, designers only model finalized concepts, and have no feedback on feasibility during creative exploration.
Our ambition is to bring the power of 3D engineering tools to the creative phase of design by automatically estimating 3D models from drawings. However, this problem is ill-posed: a point in the drawing can lie anywhere in depth. Existing solutions are limited to simple shapes, or require user input to “explain” to the computer how to interpret the drawing. Our originality is to exploit professional drawing techniques that designers developed to communicate shape most efficiently. Each technique provides geometric constraints that help viewers understand drawings, and that we shall leverage for 3D reconstruction.
Our first challenge is to formalize common drawing techniques and derive how they constrain 3D shape. Our second challenge is to identify which techniques are used in a drawing. We cast this problem as the joint optimization of discrete variables indicating which constraints apply, and continuous variables representing the 3D model that best satisfies these constraints. But evaluating all constraint configurations is impractical. To solve this inverse problem, we will first develop forward algorithms that synthesize drawings from 3D models. Our idea is to use this synthetic data to train machine learning algorithms that predict the likelihood that constraints apply in a given drawing.
In addition to tackling the long-standing problem of single-image 3D reconstruction, our research will significantly tighten design and engineering for rapid prototyping.
Summary
Designers draw extensively to externalize their ideas and communicate with others. However, drawings are currently not directly interpretable by computers. To test their ideas against physical reality, designers have to create 3D models suitable for simulation and 3D printing. However, the visceral and approximate nature of drawing clashes with the tediousness and rigidity of 3D modeling. As a result, designers only model finalized concepts, and have no feedback on feasibility during creative exploration.
Our ambition is to bring the power of 3D engineering tools to the creative phase of design by automatically estimating 3D models from drawings. However, this problem is ill-posed: a point in the drawing can lie anywhere in depth. Existing solutions are limited to simple shapes, or require user input to “explain” to the computer how to interpret the drawing. Our originality is to exploit professional drawing techniques that designers developed to communicate shape most efficiently. Each technique provides geometric constraints that help viewers understand drawings, and that we shall leverage for 3D reconstruction.
Our first challenge is to formalize common drawing techniques and derive how they constrain 3D shape. Our second challenge is to identify which techniques are used in a drawing. We cast this problem as the joint optimization of discrete variables indicating which constraints apply, and continuous variables representing the 3D model that best satisfies these constraints. But evaluating all constraint configurations is impractical. To solve this inverse problem, we will first develop forward algorithms that synthesize drawings from 3D models. Our idea is to use this synthetic data to train machine learning algorithms that predict the likelihood that constraints apply in a given drawing.
In addition to tackling the long-standing problem of single-image 3D reconstruction, our research will significantly tighten design and engineering for rapid prototyping.
Max ERC Funding
1 482 761 €
Duration
Start date: 2017-02-01, End date: 2022-01-31
