Periodic Reporting for period 1 - POST-DIGITAL (European Training Network on Post-Digital Computing (Post-Digital))
Período documentado: 2020-04-01 hasta 2022-03-31
The future economic growth in Europe requires engineers and researchers capable to design, develop and implement new information technologies to support explosive data-driven transformation of economy, public and government activities. Current information, computing and processing technologies strongly rely on the classical digital approaches and architectures developed by von Neumann. It is well understood nowadays that brain- or nature-inspired (neuromorphic) technologies can offer substantial advantage in terms of processing capabilities and power efficiency. The development of analog hardware platforms will allow us to achieve significantly higher bandwidth efficiency, faster processing and higher adaptability through integration of self-learning systems.
European Training Network on Post-Digital Computing – POST-DIGITAL is an interdisciplinary training network comprising internationally leading teams from academia, research centres and industry, including IBM, Thales, VLC Photonics and two highly reputed SMEs. POST-DIGITAL provides a unique training opportunity to a cohort of 15 early stage researchers (ESRs) in the inter-disciplinary fields of emerging disruptive neuromorphic computational technologies and their applications. The strong industrial presence in the network will bridge the gap between early stage innovation and utilization, providing ESRs with the experience of practical applications and solutions beyond traditional digital methods. POST-DIGITAL has the ambition and the vision to create a new generation of scientific and industrial leaders that will greatly contribute to strengthening Europe’s human resources and industry competitiveness in future digital and post digital economy and technology.
European Training Network on Post-Digital Computing – POST-DIGITAL is an interdisciplinary training network comprising internationally leading teams from academia, research centres and industry, including IBM, Thales, VLC Photonics and two highly reputed SMEs. POST-DIGITAL provides a unique training opportunity to a cohort of 15 early stage researchers (ESRs) in the inter-disciplinary fields of emerging disruptive neuromorphic computational technologies and their applications. The strong industrial presence in the network will bridge the gap between early stage innovation and utilization, providing ESRs with the experience of practical applications and solutions beyond traditional digital methods. POST-DIGITAL has the ambition and the vision to create a new generation of scientific and industrial leaders that will greatly contribute to strengthening Europe’s human resources and industry competitiveness in future digital and post digital economy and technology.
All WPs have made excellent progress in the 24 months, as detailed below:
WP1: New Concepts and Theory (GRONINGEN)
There is an urgent need for a unified theory development in non-digital computing, in order to integrate the multiplicity of as yet largely disconnected researches in different traditional disciplines. WP1 has taken up this challenge. A detailed pathway toward a unified theory has been established. The ESRs contributing to WP1, who come from a variety of different disciplines (physics, engineering, computer science, mathematics) have received an intense interdisciplinary training that enables them to link their individual projects into a general unifying frame.
At the end of the project this will lead to a substantial solidification of the current worldwide effort to establish non-digital computing as a scientific engineering discipline which can stand side by side with digital computing. It will help to overcome the technological limits that are currently discussed under the headline "End of Moore's Law", and it will offer extremely energy-efficient solutions that can make information and communication technologies societally and ecologically sustainable.
WP2: Implementation and characterization (CSIC)
Photonics is one of the main drivers for unconventional computing implementations due to potential advantages in speed and power consumption. In particular, photonic substrates are being proposed for the implementation of artificial neural networks. The ESRs of Post-Digital are exploring different alternatives for the implementation of scalable photonic neural networks. These proposals include systems based on large-area vertical-cavity surface-emitting lasers, optoelectronic time-multiplexed delay dynamical systems, multi-mode fibers, or frequency-multiplexed fiber setups. The characterization of the hardware implementations is targeting the demonstration of analogue computational concepts for applications such as the recovery of optical communication signals.
WP3: Integrated systems (IMEC)
Photonic reservoirs boast many advantages that render them industrially promising processing approaches. As such, the design and fabrication of on-chip photonic reservoir computing architectures with integrated optical weights and readouts are a main focus in this work package. Furthermore, the work package looks at applications where such solutions can demonstrate low power and high bandwidth. To this end, ESR 9 has shown through simulations that a passive photonic reservoir can be used successfully in nonlinear equalization where legacy electronics struggle. She will then demonstrate these advantages experimentally using the fabricated on-chip reservoir. ESR 12 has also shown that a photonic reservoir computer based on frequency multiplexing was successful on benchmark tests. ESR 11 and 12 are now designing a photonic integrated implementation of this reservoir with a novel optical output layer. ESR 15 is working on a photonic analog crossbar array, where he has done extensive experimental measurements on a prototype circuit. He will later design and fabricate new circuits based on the understanding obtained from the characterization of the prototype.
WP4: Benchmarking and applications (ULB)
In the first two years of Post-Digital most of the work in WP4 has focused on decreasing errors in optical communication systems. This goal is particularly suited to photonic reservoir computers, as the signals to be processed are optical, and the high speed possible in optical information processing systems is adapted to the high speeds of optical communications. In addition progress has been made on an experimental photonic crossbar array. Once it is operational it will be possible to benchmark it against other implementations of crossbar arrays. Finally preliminary experimental results have been obtained on a new task: the recognition of human actions in videos.
WP5-7: POST-DIGITAL is keenly aware of the importance of disseminating results both to fellow professionals and to the general public. ESRs have published their research, presented it at conferences and have been active in engaging with the general public in order to promote their work.
WP8: Ethics (ASTON)
The POST-DIGITAL consortium appointed an external Ethics Advisor. The Ethics Advisor provided a training on ethical considerations in research for the ESRs and defined key ethical issues concerning the research activities related to the first 24 months of the project.
WP1: New Concepts and Theory (GRONINGEN)
There is an urgent need for a unified theory development in non-digital computing, in order to integrate the multiplicity of as yet largely disconnected researches in different traditional disciplines. WP1 has taken up this challenge. A detailed pathway toward a unified theory has been established. The ESRs contributing to WP1, who come from a variety of different disciplines (physics, engineering, computer science, mathematics) have received an intense interdisciplinary training that enables them to link their individual projects into a general unifying frame.
At the end of the project this will lead to a substantial solidification of the current worldwide effort to establish non-digital computing as a scientific engineering discipline which can stand side by side with digital computing. It will help to overcome the technological limits that are currently discussed under the headline "End of Moore's Law", and it will offer extremely energy-efficient solutions that can make information and communication technologies societally and ecologically sustainable.
WP2: Implementation and characterization (CSIC)
Photonics is one of the main drivers for unconventional computing implementations due to potential advantages in speed and power consumption. In particular, photonic substrates are being proposed for the implementation of artificial neural networks. The ESRs of Post-Digital are exploring different alternatives for the implementation of scalable photonic neural networks. These proposals include systems based on large-area vertical-cavity surface-emitting lasers, optoelectronic time-multiplexed delay dynamical systems, multi-mode fibers, or frequency-multiplexed fiber setups. The characterization of the hardware implementations is targeting the demonstration of analogue computational concepts for applications such as the recovery of optical communication signals.
WP3: Integrated systems (IMEC)
Photonic reservoirs boast many advantages that render them industrially promising processing approaches. As such, the design and fabrication of on-chip photonic reservoir computing architectures with integrated optical weights and readouts are a main focus in this work package. Furthermore, the work package looks at applications where such solutions can demonstrate low power and high bandwidth. To this end, ESR 9 has shown through simulations that a passive photonic reservoir can be used successfully in nonlinear equalization where legacy electronics struggle. She will then demonstrate these advantages experimentally using the fabricated on-chip reservoir. ESR 12 has also shown that a photonic reservoir computer based on frequency multiplexing was successful on benchmark tests. ESR 11 and 12 are now designing a photonic integrated implementation of this reservoir with a novel optical output layer. ESR 15 is working on a photonic analog crossbar array, where he has done extensive experimental measurements on a prototype circuit. He will later design and fabricate new circuits based on the understanding obtained from the characterization of the prototype.
WP4: Benchmarking and applications (ULB)
In the first two years of Post-Digital most of the work in WP4 has focused on decreasing errors in optical communication systems. This goal is particularly suited to photonic reservoir computers, as the signals to be processed are optical, and the high speed possible in optical information processing systems is adapted to the high speeds of optical communications. In addition progress has been made on an experimental photonic crossbar array. Once it is operational it will be possible to benchmark it against other implementations of crossbar arrays. Finally preliminary experimental results have been obtained on a new task: the recognition of human actions in videos.
WP5-7: POST-DIGITAL is keenly aware of the importance of disseminating results both to fellow professionals and to the general public. ESRs have published their research, presented it at conferences and have been active in engaging with the general public in order to promote their work.
WP8: Ethics (ASTON)
The POST-DIGITAL consortium appointed an external Ethics Advisor. The Ethics Advisor provided a training on ethical considerations in research for the ESRs and defined key ethical issues concerning the research activities related to the first 24 months of the project.
The main progress beyond the state of the art during the first two years of the project consists in:
- outlining a unified theory of non-digital computing in order to integrate the largely disconnected research efforts carried out in different disciplines;
-developing photonic neural networks based on frequency multiplexing.
Expected results until the end of the project are:
-further strengthening the theoretical basis for non-digital computing, including understanding and quantifying the impact of noise on the performance of non-digital computing;
-developing and testing a variety of photonic neural networks using different implementation principles;
-developing and testing several integrated photonic neural networks, including systems (or parts of systems) based on passive multimode interferometers, frequency multiplexing, photonic crossbar arrays;
-applying the methods developed in the project to decrease errors in high-speed telecommunication.
The main progress so far consists in basic research or experimental system at low Technological Readiness Levels (TRL).
- outlining a unified theory of non-digital computing in order to integrate the largely disconnected research efforts carried out in different disciplines;
-developing photonic neural networks based on frequency multiplexing.
Expected results until the end of the project are:
-further strengthening the theoretical basis for non-digital computing, including understanding and quantifying the impact of noise on the performance of non-digital computing;
-developing and testing a variety of photonic neural networks using different implementation principles;
-developing and testing several integrated photonic neural networks, including systems (or parts of systems) based on passive multimode interferometers, frequency multiplexing, photonic crossbar arrays;
-applying the methods developed in the project to decrease errors in high-speed telecommunication.
The main progress so far consists in basic research or experimental system at low Technological Readiness Levels (TRL).