Project acronym 4DRepLy
Project Closing the 4D Real World Reconstruction Loop
Researcher (PI) Christian THEOBALT
Host Institution (HI) MAX-PLANCK-GESELLSCHAFT ZUR FORDERUNG DER WISSENSCHAFTEN EV
Call Details Consolidator Grant (CoG), PE6, ERC-2017-COG
Summary 4D reconstruction, the camera-based dense dynamic scene reconstruction, is a grand challenge in computer graphics and computer vision. Despite great progress, 4D capturing the complex, diverse real world outside a studio is still far from feasible. 4DRepLy builds a new generation of high-fidelity 4D reconstruction (4DRecon) methods. They will be the first to efficiently capture all types of deformable objects (humans and other types) in crowded real world scenes with a single color or depth camera. They capture space-time coherent deforming geometry, motion, high-frequency reflectance and illumination at unprecedented detail, and will be the first to handle difficult occlusions, topology changes and large groups of interacting objects. They automatically adapt to new scene types, yet deliver models with meaningful, interpretable parameters. This requires far reaching contributions: First, we develop groundbreaking new plasticity-enhanced model-based 4D reconstruction methods that automatically adapt to new scenes. Second, we develop radically new machine learning-based dense 4D reconstruction methods. Third, these model- and learning-based methods are combined in two revolutionary new classes of 4DRecon methods: 1) advanced fusion-based methods and 2) methods with deep architectural integration. Both, 1) and 2), are automatically designed in the 4D Real World Reconstruction Loop, a revolutionary new design paradigm in which 4DRecon methods refine and adapt themselves while continuously processing unlabeled real world input. This overcomes the previously unbreakable scalability barrier to real world scene diversity, complexity and generality. This paradigm shift opens up a new research direction in graphics and vision and has far reaching relevance across many scientific fields. It enables new applications of profound social pervasion and significant economic impact, e.g., for visual media and virtual/augmented reality, and for future autonomous and robotic systems.
Summary
4D reconstruction, the camera-based dense dynamic scene reconstruction, is a grand challenge in computer graphics and computer vision. Despite great progress, 4D capturing the complex, diverse real world outside a studio is still far from feasible. 4DRepLy builds a new generation of high-fidelity 4D reconstruction (4DRecon) methods. They will be the first to efficiently capture all types of deformable objects (humans and other types) in crowded real world scenes with a single color or depth camera. They capture space-time coherent deforming geometry, motion, high-frequency reflectance and illumination at unprecedented detail, and will be the first to handle difficult occlusions, topology changes and large groups of interacting objects. They automatically adapt to new scene types, yet deliver models with meaningful, interpretable parameters. This requires far reaching contributions: First, we develop groundbreaking new plasticity-enhanced model-based 4D reconstruction methods that automatically adapt to new scenes. Second, we develop radically new machine learning-based dense 4D reconstruction methods. Third, these model- and learning-based methods are combined in two revolutionary new classes of 4DRecon methods: 1) advanced fusion-based methods and 2) methods with deep architectural integration. Both, 1) and 2), are automatically designed in the 4D Real World Reconstruction Loop, a revolutionary new design paradigm in which 4DRecon methods refine and adapt themselves while continuously processing unlabeled real world input. This overcomes the previously unbreakable scalability barrier to real world scene diversity, complexity and generality. This paradigm shift opens up a new research direction in graphics and vision and has far reaching relevance across many scientific fields. It enables new applications of profound social pervasion and significant economic impact, e.g., for visual media and virtual/augmented reality, and for future autonomous and robotic systems.
Max ERC Funding
1 977 000 €
Duration
Start date: 2018-09-01, End date: 2023-08-31
Project acronym Active-DNA
Project Computationally Active DNA Nanostructures
Researcher (PI) Damien WOODS
Host Institution (HI) NATIONAL UNIVERSITY OF IRELAND MAYNOOTH
Call Details Consolidator Grant (CoG), PE6, ERC-2017-COG
Summary During the 20th century computer technology evolved from bulky, slow, special purpose mechanical engines to the now ubiquitous silicon chips and software that are one of the pinnacles of human ingenuity. The goal of the field of molecular programming is to take the next leap and build a new generation of matter-based computers using DNA, RNA and proteins. This will be accomplished by computer scientists, physicists and chemists designing molecules to execute ``wet'' nanoscale programs in test tubes. The workflow includes proposing theoretical models, mathematically proving their computational properties, physical modelling and implementation in the wet-lab.
The past decade has seen remarkable progress at building static 2D and 3D DNA nanostructures. However, unlike biological macromolecules and complexes that are built via specified self-assembly pathways, that execute robotic-like movements, and that undergo evolution, the activity of human-engineered nanostructures is severely limited. We will need sophisticated algorithmic ideas to build structures that rival active living systems. Active-DNA, aims to address this challenge by achieving a number of objectives on computation, DNA-based self-assembly and molecular robotics. Active-DNA research work will range from defining models and proving theorems that characterise the computational and expressive capabilities of such active programmable materials to experimental work implementing active DNA nanostructures in the wet-lab.
Summary
During the 20th century computer technology evolved from bulky, slow, special purpose mechanical engines to the now ubiquitous silicon chips and software that are one of the pinnacles of human ingenuity. The goal of the field of molecular programming is to take the next leap and build a new generation of matter-based computers using DNA, RNA and proteins. This will be accomplished by computer scientists, physicists and chemists designing molecules to execute ``wet'' nanoscale programs in test tubes. The workflow includes proposing theoretical models, mathematically proving their computational properties, physical modelling and implementation in the wet-lab.
The past decade has seen remarkable progress at building static 2D and 3D DNA nanostructures. However, unlike biological macromolecules and complexes that are built via specified self-assembly pathways, that execute robotic-like movements, and that undergo evolution, the activity of human-engineered nanostructures is severely limited. We will need sophisticated algorithmic ideas to build structures that rival active living systems. Active-DNA, aims to address this challenge by achieving a number of objectives on computation, DNA-based self-assembly and molecular robotics. Active-DNA research work will range from defining models and proving theorems that characterise the computational and expressive capabilities of such active programmable materials to experimental work implementing active DNA nanostructures in the wet-lab.
Max ERC Funding
2 349 603 €
Duration
Start date: 2018-11-01, End date: 2023-10-31
Project acronym BITCRUMBS
Project Towards a Reliable and Automated Analysis of Compromised Systems
Researcher (PI) Davide BALZAROTTI
Host Institution (HI) EURECOM
Call Details Consolidator Grant (CoG), PE6, ERC-2017-COG
Summary "The vast majority of research in computer security is dedicated to the design of detection, protection, and prevention solutions. While these techniques play a critical role to increase the security and privacy of our digital infrastructure, it is enough to look at the news to understand that it is not a matter of ""if"" a computer system will be compromised, but only a matter of ""when"". It is a well known fact that there is no 100% secure system, and that there is no practical way to prevent attackers with enough resources from breaking into sensitive targets. Therefore, it is extremely important to develop automated techniques to timely and precisely analyze computer security incidents and compromised systems. Unfortunately, the area of incident response received very little research attention, and it is still largely considered an art more than a science because of its lack of a proper theoretical and scientific background.
The objective of BITCRUMBS is to rethink the Incident Response (IR) field from its foundations by proposing a more scientific and comprehensive approach to the analysis of compromised systems. BITCRUMBS will achieve this goal in three steps: (1) by introducing a new systematic approach to precisely measure the effectiveness and accuracy of IR techniques and their resilience to evasion and forgery; (2) by designing and implementing new automated techniques to cope with advanced threats and the analysis of IoT devices; and (3) by proposing a novel forensics-by-design development methodology and a set of guidelines for the design of future systems and software.
To provide the right context for these new techniques and show the impact of the project in different fields and scenarios, BITCRUMBS plans to address its objectives using real case studies borrowed from two different
domains: traditional computer software, and embedded systems.
"
Summary
"The vast majority of research in computer security is dedicated to the design of detection, protection, and prevention solutions. While these techniques play a critical role to increase the security and privacy of our digital infrastructure, it is enough to look at the news to understand that it is not a matter of ""if"" a computer system will be compromised, but only a matter of ""when"". It is a well known fact that there is no 100% secure system, and that there is no practical way to prevent attackers with enough resources from breaking into sensitive targets. Therefore, it is extremely important to develop automated techniques to timely and precisely analyze computer security incidents and compromised systems. Unfortunately, the area of incident response received very little research attention, and it is still largely considered an art more than a science because of its lack of a proper theoretical and scientific background.
The objective of BITCRUMBS is to rethink the Incident Response (IR) field from its foundations by proposing a more scientific and comprehensive approach to the analysis of compromised systems. BITCRUMBS will achieve this goal in three steps: (1) by introducing a new systematic approach to precisely measure the effectiveness and accuracy of IR techniques and their resilience to evasion and forgery; (2) by designing and implementing new automated techniques to cope with advanced threats and the analysis of IoT devices; and (3) by proposing a novel forensics-by-design development methodology and a set of guidelines for the design of future systems and software.
To provide the right context for these new techniques and show the impact of the project in different fields and scenarios, BITCRUMBS plans to address its objectives using real case studies borrowed from two different
domains: traditional computer software, and embedded systems.
"
Max ERC Funding
1 991 504 €
Duration
Start date: 2018-04-01, End date: 2023-03-31
Project acronym Browsec
Project Foundations and Tools for Client-Side Web Security
Researcher (PI) Matteo MAFFEI
Host Institution (HI) TECHNISCHE UNIVERSITAET WIEN
Call Details Consolidator Grant (CoG), PE6, ERC-2017-COG
Summary The constantly increasing number of attacks on web applications shows how their rapid development has not been accompanied by adequate security foundations and demonstrates the lack of solid security enforcement tools. Indeed, web applications expose a gigantic attack surface, which hinders a rigorous understanding and enforcement of security properties. Hence, despite the worthwhile efforts to design secure web applications, users for a while will be confronted with vulnerable, or maliciously crafted, code. Unfortunately, end users have no way at present to reliably protect themselves from malicious applications.
BROWSEC will develop a holistic approach to client-side web security, laying its theoretical foundations and developing innovative security enforcement technologies. In particular, BROWSEC will deliver the first client-side tool to secure web applications that is practical, in that it is implemented as an extension and can thus be easily deployed at large, and also provably sound, i.e., backed up by machine-checked proofs that the tool provides end users with the required security guarantees. At the core of the proposal lies a novel monitoring technique, which treats the browser as a blackbox and intercepts its inputs and outputs in order to prevent dangerous information flows. With this lightweight monitoring approach, we aim at enforcing strong security properties without requiring any expensive and, given the dynamic nature of web applications, statically infeasible program analysis.
BROWSEC is thus a multidisciplinary research effort, promising practical impact and delivering breakthrough advancements in various disciplines, such as web security, JavaScript semantics, software engineering, and program verification.
Summary
The constantly increasing number of attacks on web applications shows how their rapid development has not been accompanied by adequate security foundations and demonstrates the lack of solid security enforcement tools. Indeed, web applications expose a gigantic attack surface, which hinders a rigorous understanding and enforcement of security properties. Hence, despite the worthwhile efforts to design secure web applications, users for a while will be confronted with vulnerable, or maliciously crafted, code. Unfortunately, end users have no way at present to reliably protect themselves from malicious applications.
BROWSEC will develop a holistic approach to client-side web security, laying its theoretical foundations and developing innovative security enforcement technologies. In particular, BROWSEC will deliver the first client-side tool to secure web applications that is practical, in that it is implemented as an extension and can thus be easily deployed at large, and also provably sound, i.e., backed up by machine-checked proofs that the tool provides end users with the required security guarantees. At the core of the proposal lies a novel monitoring technique, which treats the browser as a blackbox and intercepts its inputs and outputs in order to prevent dangerous information flows. With this lightweight monitoring approach, we aim at enforcing strong security properties without requiring any expensive and, given the dynamic nature of web applications, statically infeasible program analysis.
BROWSEC is thus a multidisciplinary research effort, promising practical impact and delivering breakthrough advancements in various disciplines, such as web security, JavaScript semantics, software engineering, and program verification.
Max ERC Funding
1 990 000 €
Duration
Start date: 2018-06-01, End date: 2023-05-31
Project acronym CharFL
Project Characterizing the fitness landscape on population and global scales
Researcher (PI) Fyodor Kondrashov
Host Institution (HI) INSTITUTE OF SCIENCE AND TECHNOLOGYAUSTRIA
Call Details Consolidator Grant (CoG), LS2, ERC-2017-COG
Summary The fitness landscape, the representation of how the genotype manifests at the phenotypic (fitness) levels, may be among the most useful concepts in biology with impact on diverse fields, including quantitative genetics, emergence of pathogen resistance, synthetic biology and protein engineering. While progress in characterizing fitness landscapes has been made, three directions of research in the field remain virtually unexplored: the nature of the genotype to phenotype of standing variation (variation found in a natural population), the shape of the fitness landscape encompassing many genotypes and the modelling of complex genetic interactions in protein sequences.
The current proposal is designed to advance the study of fitness landscapes in these three directions using large-scale genomic experiments and experimental data from a model protein and theoretical work. The study of the fitness landscape of standing variation is aimed at the resolution of an outstanding question in quantitative genetics: the extent to which epistasis, non-additive genetic interactions, is shaping the phenotype. The second aim of characterizing the global fitness landscape will give us an understanding of how evolution proceeds along long evolutionary timescales, which can be directly applied to protein engineering and synthetic biology for the design of novel phenotypes. Finally, the third aim of modelling complex interactions will improve our ability to predict phenotypes from genotypes, such as the prediction of human disease mutations. In summary, the proposed study presents an opportunity to provide a unifying understanding of how phenotypes are shaped through genetic interactions. The consolidation of our empirical and theoretical work on different scales of the genotype to phenotype relationship will provide empirical data and novel context for several fields of biology.
Summary
The fitness landscape, the representation of how the genotype manifests at the phenotypic (fitness) levels, may be among the most useful concepts in biology with impact on diverse fields, including quantitative genetics, emergence of pathogen resistance, synthetic biology and protein engineering. While progress in characterizing fitness landscapes has been made, three directions of research in the field remain virtually unexplored: the nature of the genotype to phenotype of standing variation (variation found in a natural population), the shape of the fitness landscape encompassing many genotypes and the modelling of complex genetic interactions in protein sequences.
The current proposal is designed to advance the study of fitness landscapes in these three directions using large-scale genomic experiments and experimental data from a model protein and theoretical work. The study of the fitness landscape of standing variation is aimed at the resolution of an outstanding question in quantitative genetics: the extent to which epistasis, non-additive genetic interactions, is shaping the phenotype. The second aim of characterizing the global fitness landscape will give us an understanding of how evolution proceeds along long evolutionary timescales, which can be directly applied to protein engineering and synthetic biology for the design of novel phenotypes. Finally, the third aim of modelling complex interactions will improve our ability to predict phenotypes from genotypes, such as the prediction of human disease mutations. In summary, the proposed study presents an opportunity to provide a unifying understanding of how phenotypes are shaped through genetic interactions. The consolidation of our empirical and theoretical work on different scales of the genotype to phenotype relationship will provide empirical data and novel context for several fields of biology.
Max ERC Funding
1 998 280 €
Duration
Start date: 2019-01-01, End date: 2023-12-31
Project acronym CoCoSym
Project Symmetry in Computational Complexity
Researcher (PI) Libor BARTO
Host Institution (HI) UNIVERZITA KARLOVA
Call Details Consolidator Grant (CoG), PE6, ERC-2017-COG
Summary The last 20 years of rapid development in the computational-theoretic aspects of the fixed-language Constraint Satisfaction Problems (CSPs) has been fueled by a connection between the complexity and a certain concept capturing symmetry of computational problems in this class.
My vision is that this connection will eventually evolve into the organizing principle of computational complexity and will lead to solutions of fundamental problems such as the Unique Games Conjecture or even the P-versus-NP problem. In order to break through the current limits of this algebraic approach, I will concentrate on specific goals designed to
(A) discover suitable objects capturing symmetry that reflect the complexity in problem classes, where such an object is not known yet;
(B) make the natural ordering of symmetries coarser so that it reflects the complexity more faithfully;
(C) delineate the borderline between computationally hard and easy problems;
(D) strengthen characterizations of existing borderlines to increase their usefulness as tools for proving hardness and designing efficient algorithm; and
(E) design efficient algorithms based on direct and indirect uses of symmetries.
The specific goals concern the fixed-language CSP over finite relational structures and its generalizations to infinite domains (iCSP) and weighted relations (vCSP), in which the algebraic theory is highly developed and the limitations are clearly visible.
The approach is based on joining the forces of the universal algebraic methods in finite domains, model-theoretical and topological methods in the iCSP, and analytical and probabilistic methods in the vCSP. The starting point is to generalize and improve the Absorption Theory from finite domains.
Summary
The last 20 years of rapid development in the computational-theoretic aspects of the fixed-language Constraint Satisfaction Problems (CSPs) has been fueled by a connection between the complexity and a certain concept capturing symmetry of computational problems in this class.
My vision is that this connection will eventually evolve into the organizing principle of computational complexity and will lead to solutions of fundamental problems such as the Unique Games Conjecture or even the P-versus-NP problem. In order to break through the current limits of this algebraic approach, I will concentrate on specific goals designed to
(A) discover suitable objects capturing symmetry that reflect the complexity in problem classes, where such an object is not known yet;
(B) make the natural ordering of symmetries coarser so that it reflects the complexity more faithfully;
(C) delineate the borderline between computationally hard and easy problems;
(D) strengthen characterizations of existing borderlines to increase their usefulness as tools for proving hardness and designing efficient algorithm; and
(E) design efficient algorithms based on direct and indirect uses of symmetries.
The specific goals concern the fixed-language CSP over finite relational structures and its generalizations to infinite domains (iCSP) and weighted relations (vCSP), in which the algebraic theory is highly developed and the limitations are clearly visible.
The approach is based on joining the forces of the universal algebraic methods in finite domains, model-theoretical and topological methods in the iCSP, and analytical and probabilistic methods in the vCSP. The starting point is to generalize and improve the Absorption Theory from finite domains.
Max ERC Funding
1 211 375 €
Duration
Start date: 2018-02-01, End date: 2023-01-31
Project acronym CohesinLooping
Project Cohesin-mediated chromosomal looping: From linear paths to 3D effects
Researcher (PI) Benjamin Rowland
Host Institution (HI) STICHTING HET NEDERLANDS KANKER INSTITUUT-ANTONI VAN LEEUWENHOEK ZIEKENHUIS
Call Details Consolidator Grant (CoG), LS2, ERC-2017-COG
Summary The 3D organization of chromosomes within the nucleus is of great importance to control gene expression. The cohesin complex plays a key role in such higher-order chromosome organization by looping together regulatory elements in cis. How these often megabase-sized looped structures are formed is one of the main open questions in chromosome biology. Cohesin is a ring-shaped complex that can entrap DNA inside its lumen. However, cohesin’s default behaviour is that it only transiently entraps and then releases DNA. Our recent findings indicate that chromosomes are structured through the processive enlargement of chromatin loops, and that the duration with which cohesin embraces DNA determines the degree to which loops are enlarged. The goal of this proposal is two-fold. First, we plan to investigate the mechanism by which chromatin loops are formed, and secondly we wish to dissect how looped structures are maintained. We will use a multi-disciplinary approach that includes refined genetic screens in haploid human cells, chromosome conformation capture techniques, the tracing in vivo of cohesin on individual DNA molecules, and visualization of chromosome organization by super-resolution imaging. With unbiased genetic screens, we have identified chromatin regulators involved in the formation of chromosomal loops. We will investigate how they drive loop formation, and also whether cohesin’s own enzymatic activity plays a role in the enlargement of loops. We will study whether and how these factors control the movement of cohesin along individual DNA molecules, and whether chromatin loops pass through cohesin rings during their formation. Ultimately, we plan to couple cohesin’s linear trajectory along chromatin to the 3D consequences for chromosomal architecture. Together our experiments will provide vital insight into how cohesin structures chromosomes.
Summary
The 3D organization of chromosomes within the nucleus is of great importance to control gene expression. The cohesin complex plays a key role in such higher-order chromosome organization by looping together regulatory elements in cis. How these often megabase-sized looped structures are formed is one of the main open questions in chromosome biology. Cohesin is a ring-shaped complex that can entrap DNA inside its lumen. However, cohesin’s default behaviour is that it only transiently entraps and then releases DNA. Our recent findings indicate that chromosomes are structured through the processive enlargement of chromatin loops, and that the duration with which cohesin embraces DNA determines the degree to which loops are enlarged. The goal of this proposal is two-fold. First, we plan to investigate the mechanism by which chromatin loops are formed, and secondly we wish to dissect how looped structures are maintained. We will use a multi-disciplinary approach that includes refined genetic screens in haploid human cells, chromosome conformation capture techniques, the tracing in vivo of cohesin on individual DNA molecules, and visualization of chromosome organization by super-resolution imaging. With unbiased genetic screens, we have identified chromatin regulators involved in the formation of chromosomal loops. We will investigate how they drive loop formation, and also whether cohesin’s own enzymatic activity plays a role in the enlargement of loops. We will study whether and how these factors control the movement of cohesin along individual DNA molecules, and whether chromatin loops pass through cohesin rings during their formation. Ultimately, we plan to couple cohesin’s linear trajectory along chromatin to the 3D consequences for chromosomal architecture. Together our experiments will provide vital insight into how cohesin structures chromosomes.
Max ERC Funding
1 998 375 €
Duration
Start date: 2018-04-01, End date: 2023-03-31
Project acronym D-SynMA
Project Distributed Synthesis: from Single to Multiple Agents
Researcher (PI) Nir PITERMAN
Host Institution (HI) GOETEBORGS UNIVERSITET
Call Details Consolidator Grant (CoG), PE6, ERC-2017-COG
Summary Computing is changing from living on our desktops and in dedicated devices to being everywhere. In phones, sensors, appliances, and robots – computers (from now on devices) are everywhere and affecting all aspects of our lives. The techniques to make them safe and reliable are investigated and are starting to emerge and consolidate. However, these techniques enable devices to work in isolation or co-exist. We currently do not have techniques that enable development of real autonomous collaboration between devices. Such techniques will revolutionize all usage of devices and, as consequence, our lives. Manufacturing, supply chain, transportation, infrastructures, and earth- and space exploration would all transform using techniques that enable development of collaborating devices.
When considering isolated (and co-existing) devices, reactive synthesis – automatic production of plans from high level specification – is emerging as a viable tool for the development of robots and reactive software. This is especially important in the context of safety-critical systems, where assurances are required and systems need to have guarantees on performance. The techniques that are developed today to support robust, assured, reliable, and adaptive devices rely on a major change in focus of reactive synthesis. The revolution of correct-by-construction systems from specifications is occurring and is being pushed forward.
However, to take this approach forward to work also for real collaboration between devices the theoretical frameworks that will enable distributed synthesis are required. Such foundations will enable the correct-by-construction revolution to unleash its potential and allow a multiplicative increase of utility by cooperative computation.
d-SynMA will take distributed synthesis to this new frontier by considering novel interaction and communication concepts that would create an adaptable framework of correct-by-construction application of collaborating devices.
Summary
Computing is changing from living on our desktops and in dedicated devices to being everywhere. In phones, sensors, appliances, and robots – computers (from now on devices) are everywhere and affecting all aspects of our lives. The techniques to make them safe and reliable are investigated and are starting to emerge and consolidate. However, these techniques enable devices to work in isolation or co-exist. We currently do not have techniques that enable development of real autonomous collaboration between devices. Such techniques will revolutionize all usage of devices and, as consequence, our lives. Manufacturing, supply chain, transportation, infrastructures, and earth- and space exploration would all transform using techniques that enable development of collaborating devices.
When considering isolated (and co-existing) devices, reactive synthesis – automatic production of plans from high level specification – is emerging as a viable tool for the development of robots and reactive software. This is especially important in the context of safety-critical systems, where assurances are required and systems need to have guarantees on performance. The techniques that are developed today to support robust, assured, reliable, and adaptive devices rely on a major change in focus of reactive synthesis. The revolution of correct-by-construction systems from specifications is occurring and is being pushed forward.
However, to take this approach forward to work also for real collaboration between devices the theoretical frameworks that will enable distributed synthesis are required. Such foundations will enable the correct-by-construction revolution to unleash its potential and allow a multiplicative increase of utility by cooperative computation.
d-SynMA will take distributed synthesis to this new frontier by considering novel interaction and communication concepts that would create an adaptable framework of correct-by-construction application of collaborating devices.
Max ERC Funding
1 871 272 €
Duration
Start date: 2018-05-01, End date: 2023-04-30
Project acronym DeciGUT
Project A Grand Unified Theory of Decidability in Logic-Based Knowledge Representation
Researcher (PI) Sebastian Rudolph
Host Institution (HI) TECHNISCHE UNIVERSITAET DRESDEN
Call Details Consolidator Grant (CoG), PE6, ERC-2017-COG
Summary "Logic-based knowledge representation (KR) constitutes a vital area of IT. The field inspires and guides scientific and technological developments enabling intelligent management of large and complex knowledge resources. Elaborate languages for specifying knowledge (so-called ontology languages) and querying it have been defined and standardized. Algorithms for automated reasoning and intelligent querying over knowledge resources are being developed, implemented and practically deployed on a wide scale.
Thereby, decidability investigations play a pivotal role to characterize what reasoning or querying tasks are at all computationally solvable.
Past decades have seen a proliferation of new decidable formalisms for KR, dominated by two major paradigms: description logics and rule-based approaches, most notably existential rules. Recently, these research lines have started to converge and first progress has been made toward identifying commonalities among the various formalisms. Still, the underlying principles for establishing their decidability remain disparate, ranging from proof-theoretic notions to model-theoretic ones.
DeciGUT will accomplish a major breakthrough in the field by establishing a ""Grand Unified Theory"" of decidability. We will provide a novel, powerful model-theoretic criterion inspired by advanced graph-theoretic notions. We will prove that the criterion indeed ensures decidability and that it subsumes most of (if not all) currently known decidable formalisms in the KR field.
We will exploit our results toward the definition of novel decidable KR languages of unprecedented expressivity. We will ultimately extend our framework to encompass more advanced KR features beyond standard first order logic such as counting and non-monotonic aspects.
Our research will draw from and significantly impact the scientific fields of AI, Database Theory and Logic, but also give rise to drastically improved practical information management technology."
Summary
"Logic-based knowledge representation (KR) constitutes a vital area of IT. The field inspires and guides scientific and technological developments enabling intelligent management of large and complex knowledge resources. Elaborate languages for specifying knowledge (so-called ontology languages) and querying it have been defined and standardized. Algorithms for automated reasoning and intelligent querying over knowledge resources are being developed, implemented and practically deployed on a wide scale.
Thereby, decidability investigations play a pivotal role to characterize what reasoning or querying tasks are at all computationally solvable.
Past decades have seen a proliferation of new decidable formalisms for KR, dominated by two major paradigms: description logics and rule-based approaches, most notably existential rules. Recently, these research lines have started to converge and first progress has been made toward identifying commonalities among the various formalisms. Still, the underlying principles for establishing their decidability remain disparate, ranging from proof-theoretic notions to model-theoretic ones.
DeciGUT will accomplish a major breakthrough in the field by establishing a ""Grand Unified Theory"" of decidability. We will provide a novel, powerful model-theoretic criterion inspired by advanced graph-theoretic notions. We will prove that the criterion indeed ensures decidability and that it subsumes most of (if not all) currently known decidable formalisms in the KR field.
We will exploit our results toward the definition of novel decidable KR languages of unprecedented expressivity. We will ultimately extend our framework to encompass more advanced KR features beyond standard first order logic such as counting and non-monotonic aspects.
Our research will draw from and significantly impact the scientific fields of AI, Database Theory and Logic, but also give rise to drastically improved practical information management technology."
Max ERC Funding
1 814 937 €
Duration
Start date: 2018-10-01, End date: 2023-09-30
Project acronym DeeViSe
Project Deep Learning for Dynamic 3D Visual Scene Understanding
Researcher (PI) Bastian LEIBE
Host Institution (HI) RHEINISCH-WESTFAELISCHE TECHNISCHE HOCHSCHULE AACHEN
Call Details Consolidator Grant (CoG), PE6, ERC-2017-COG
Summary Over the past 5 years, deep learning has exercised a tremendous and transformational effect on the field of computer vision. However, deep neural networks (DNNs) can only realize their full potential when applied in an end-to-end manner, i.e., when every stage of the processing pipeline is differentiable with respect to the network’s parameters, such that all of those parameters can be optimized together. Such end-to-end learning solutions are still rare for computer vision problems, in particular for dynamic visual scene understanding tasks. Moreover, feed-forward processing, as done in most DNN-based vision approaches, is only a tiny fraction of what the human brain can do. Feedback processes, temporal information processing, and memory mechanisms form an important part of our human scene understanding capabilities. Those mechanisms are currently underexplored in computer vision.
The goal of this proposal is to remove this bottleneck and to design end-to-end deep learning approaches that can realize the full potential of DNNs for dynamic visual scene understanding. We will make use of the positive interactions and feedback processes between multiple vision modalities and combine them to work towards a common goal. In addition, we will impart deep learning approaches with a notion of what it means to move through a 3D world by incorporating temporal continuity constraints, as well as by developing novel deep associative and spatial memory mechanisms.
The results of this research will enable deep neural networks to reach significantly improved dynamic scene understanding capabilities compared to today’s methods. This will have an immediate positive effect for applications in need for such capabilities, most notably for mobile robotics and intelligent vehicles.
Summary
Over the past 5 years, deep learning has exercised a tremendous and transformational effect on the field of computer vision. However, deep neural networks (DNNs) can only realize their full potential when applied in an end-to-end manner, i.e., when every stage of the processing pipeline is differentiable with respect to the network’s parameters, such that all of those parameters can be optimized together. Such end-to-end learning solutions are still rare for computer vision problems, in particular for dynamic visual scene understanding tasks. Moreover, feed-forward processing, as done in most DNN-based vision approaches, is only a tiny fraction of what the human brain can do. Feedback processes, temporal information processing, and memory mechanisms form an important part of our human scene understanding capabilities. Those mechanisms are currently underexplored in computer vision.
The goal of this proposal is to remove this bottleneck and to design end-to-end deep learning approaches that can realize the full potential of DNNs for dynamic visual scene understanding. We will make use of the positive interactions and feedback processes between multiple vision modalities and combine them to work towards a common goal. In addition, we will impart deep learning approaches with a notion of what it means to move through a 3D world by incorporating temporal continuity constraints, as well as by developing novel deep associative and spatial memory mechanisms.
The results of this research will enable deep neural networks to reach significantly improved dynamic scene understanding capabilities compared to today’s methods. This will have an immediate positive effect for applications in need for such capabilities, most notably for mobile robotics and intelligent vehicles.
Max ERC Funding
2 000 000 €
Duration
Start date: 2018-04-01, End date: 2023-03-31
