Project acronym DISCOVERER
Project A novel chemical discovery platform enabled by machine learning
Host Institution (HI) UNIVERSITE DU LUXEMBOURG
Country Luxembourg
Call Details Proof of Concept (PoC), ERC-2020-PoC
Summary Computational design and discovery of molecules and materials relies on the exploration of increasingly growing chemical spaces. The discovery and formulation of new drugs, antivirals, antibiotics, catalysts, battery materials, and in general chemicals with tailored properties, require a fundamental paradigm shift to search in unchartered swaths of the vast chemical space. This is in stark contrast to current approaches, which start from (commercially available) libraries of compounds from various suppliers. Within the ERC Consolidator grant BeStMo (grant agreement ID 725291) we aimed to substantially advance our ability to model and understand the behaviour of molecules in complex environments. As a result, we successfully developed a set of machine learning and physics-based methods for covalent and non-covalent interactions that now allow an accurate and efficient modelling of molecules of increasing size (from 10 to 1000 atoms). These methods now enable routine calculations of quantum-mechanical properties of molecules throughout chemical compound space, provided that enough reference data is produced as a starting point for training. Within DISCOVERER, we aim to promote a paradigm shift in chemical discovery by inverting the selection pyramid by starting with pre-defined parameters from which new chemical entities are designed through machine learning and AI-enabled algorithms. We can do so by integrating these modules into a commercial platform: “Chemical Space Machine”. DISCOVERER’s main goal is to finalize the development of a commercial alpha version of “Chemical Space Machine” and setting up its commercialisation strategy.
Summary
Computational design and discovery of molecules and materials relies on the exploration of increasingly growing chemical spaces. The discovery and formulation of new drugs, antivirals, antibiotics, catalysts, battery materials, and in general chemicals with tailored properties, require a fundamental paradigm shift to search in unchartered swaths of the vast chemical space. This is in stark contrast to current approaches, which start from (commercially available) libraries of compounds from various suppliers. Within the ERC Consolidator grant BeStMo (grant agreement ID 725291) we aimed to substantially advance our ability to model and understand the behaviour of molecules in complex environments. As a result, we successfully developed a set of machine learning and physics-based methods for covalent and non-covalent interactions that now allow an accurate and efficient modelling of molecules of increasing size (from 10 to 1000 atoms). These methods now enable routine calculations of quantum-mechanical properties of molecules throughout chemical compound space, provided that enough reference data is produced as a starting point for training. Within DISCOVERER, we aim to promote a paradigm shift in chemical discovery by inverting the selection pyramid by starting with pre-defined parameters from which new chemical entities are designed through machine learning and AI-enabled algorithms. We can do so by integrating these modules into a commercial platform: “Chemical Space Machine”. DISCOVERER’s main goal is to finalize the development of a commercial alpha version of “Chemical Space Machine” and setting up its commercialisation strategy.
Max ERC Funding
150 000 €
Duration
Start date: 2021-09-01, End date: 2023-02-28
Project acronym DREAM
Project Demonstration of a Radar Enabled weArable platforM
Researcher (PI) Bjoern Ottersten
Host Institution (HI) UNIVERSITE DU LUXEMBOURG
Country Luxembourg
Call Details Proof of Concept (PoC), ERC-2020-PoC
Summary Sport research and procedures that aim to increase performance or assist recovery after an injury often require capture of athlete’s motion. However, implementing a motion capture system in routine practice takes significant effort. Capture volume, weather conditions, motion dynamics and athletes' timing (where timing defines coincide movements in relation to external factors) are the key parameters for outdoor sport activities (e.g. track and field or football), yet no available motion capture system is suitable for much needed high accuracy absolute position measurements of body segments in outdoor conditions with minimal setup effort for large capture volumes, which is capable of measuring the athletes' timing. Therefore, there is a significant unmet need for a portable motion capture system that can perform accurate infield measurements on a large volume, even in diverse weather conditions (e.g. sunlight, fog, rain) with the capability of the athletes' timing analysis, which is a combination of reaction time, decision -making and co-ordination in relation to external factors.
DREAM aims to generate a motion capture and mapping system (minimum viable product) that creates a complete orientation and absolute position measurement with no external position reference, provides reliable orientation of body segments even for high dynamic motions, and has a small form factor, which is minimally invasive for the athlete. Further, the offered solution is complementary to existing solutions and operates in real time. DREAM’s value proposition is a cost-effective, robust and accurate infield motion capture and mapping system with minimal setup effort, for large field coverage in any place and in any condition. The uniqueness of the product stems from the complementarity between radar and IMU sensors, enhanced athletes timing analysis, all enabled through an innovative
algorithm developed in the AGNOSTIC ERC Advanced grant (ID 742648).
Summary
Sport research and procedures that aim to increase performance or assist recovery after an injury often require capture of athlete’s motion. However, implementing a motion capture system in routine practice takes significant effort. Capture volume, weather conditions, motion dynamics and athletes' timing (where timing defines coincide movements in relation to external factors) are the key parameters for outdoor sport activities (e.g. track and field or football), yet no available motion capture system is suitable for much needed high accuracy absolute position measurements of body segments in outdoor conditions with minimal setup effort for large capture volumes, which is capable of measuring the athletes' timing. Therefore, there is a significant unmet need for a portable motion capture system that can perform accurate infield measurements on a large volume, even in diverse weather conditions (e.g. sunlight, fog, rain) with the capability of the athletes' timing analysis, which is a combination of reaction time, decision -making and co-ordination in relation to external factors.
DREAM aims to generate a motion capture and mapping system (minimum viable product) that creates a complete orientation and absolute position measurement with no external position reference, provides reliable orientation of body segments even for high dynamic motions, and has a small form factor, which is minimally invasive for the athlete. Further, the offered solution is complementary to existing solutions and operates in real time. DREAM’s value proposition is a cost-effective, robust and accurate infield motion capture and mapping system with minimal setup effort, for large field coverage in any place and in any condition. The uniqueness of the product stems from the complementarity between radar and IMU sensors, enhanced athletes timing analysis, all enabled through an innovative
algorithm developed in the AGNOSTIC ERC Advanced grant (ID 742648).
Max ERC Funding
150 000 €
Duration
Start date: 2020-10-01, End date: 2022-03-31
Project acronym NATURAL
Project Natural Program Repair
Researcher (PI) Tegawende F. Bissyande
Host Institution (HI) UNIVERSITE DU LUXEMBOURG
Country Luxembourg
Call Details Starting Grant (StG), PE6, ERC-2020-STG
Summary Automatic bug fixing, i.e., the idea of having programs that fix other programs, is a long-standing dream that is increasingly embraced by the software engineering community. Indeed, despite the significant effort that humans put into reviewing code and running software test campaigns, programming mistakes slip by, with severe consequences. Fixing those mistakes automatically has recently been the focus of a number of potentially promising techniques. Proposed approaches are however recurrently criticized as being shallow (i.e., they mostly address unit test failures, which are often neither hard nor important problems).
Initial successes in automatic bug fixing are based on scenarios such as the following: when a bug is localized, patches are generated repetitively and automatically, through trial and error, until a valid patch is produced. The produced patch could then be later revised by developers. While the reported achievements are certainly worthwhile, they do not address what we believe is a more comprehensive challenge of software engineering: to systematically fix features of a software system based on end-user requirements.
The ambition of NATURAL is to develop a methodology for yielding an intelligent agent that is capable of receiving a natural language description of a problem that a user faces with a software feature, and then synthesizing code to address this problem so that it meets the user's expectations. Such a repair bot would be a trustworthy software contributor that is 1) first, targeting real bugs in production via exploiting bug reports, which remain largely under-explored, 2) second, aligning with the conversational needs of collaborative work via generating explanations for patch suggestions, 3) third, shifting the repair paradigm towards the design of self-improving systems via yielding novel algorithms that iteratively integrate feedback from humans. Ultimately, NATURAL will be transformative in the practice of software engineering.
Summary
Automatic bug fixing, i.e., the idea of having programs that fix other programs, is a long-standing dream that is increasingly embraced by the software engineering community. Indeed, despite the significant effort that humans put into reviewing code and running software test campaigns, programming mistakes slip by, with severe consequences. Fixing those mistakes automatically has recently been the focus of a number of potentially promising techniques. Proposed approaches are however recurrently criticized as being shallow (i.e., they mostly address unit test failures, which are often neither hard nor important problems).
Initial successes in automatic bug fixing are based on scenarios such as the following: when a bug is localized, patches are generated repetitively and automatically, through trial and error, until a valid patch is produced. The produced patch could then be later revised by developers. While the reported achievements are certainly worthwhile, they do not address what we believe is a more comprehensive challenge of software engineering: to systematically fix features of a software system based on end-user requirements.
The ambition of NATURAL is to develop a methodology for yielding an intelligent agent that is capable of receiving a natural language description of a problem that a user faces with a software feature, and then synthesizing code to address this problem so that it meets the user's expectations. Such a repair bot would be a trustworthy software contributor that is 1) first, targeting real bugs in production via exploiting bug reports, which remain largely under-explored, 2) second, aligning with the conversational needs of collaborative work via generating explanations for patch suggestions, 3) third, shifting the repair paradigm towards the design of self-improving systems via yielding novel algorithms that iteratively integrate feedback from humans. Ultimately, NATURAL will be transformative in the practice of software engineering.
Max ERC Funding
1 495 988 €
Duration
Start date: 2021-02-01, End date: 2026-01-31
