Skip to content

Brain-inspired computing needs a master plan

If neural computing is needed, how can it be achieved? First, the technical requirements. Bringing together diverse research communities is necessary but not sufficient. Incentives, opportunities and infrastructure are needed. The Neural community is a disparate community that lacks a focus on quantum computing, or a clear roadmap for the semiconductor industry. Initiatives around the world are beginning to gather needed expertise, and momentum in the early stages is growing. To reinforce this, financing is key. Investing in neuromorphology research is nowhere near as large as that in digital artificial intelligence or quantum technologies (Box 2). Although this is not surprising given the maturity of digital semiconductor technology, it is a missed opportunity. There are some examples of mid-size investment in research and development in neuromorphology, such as the IBM AI Hardware Center suite of brain-inspired projects (including the TrueNorth chip), Intel’s development of the Loihi processor, and the US Brain Initiative project, but the amounts committed are lower. Much more than technology should give to disrupt digital AI.

The neurological community is a large and growing community, but it lacks focus. Although there are many conferences, symposia and journals emerging in this field, there is still much work to be done to bring disparate communities together and unite their efforts to convince funding bodies and governments of the importance of this field.

It is time for bold initiatives. At the national level, governments need to work with academic researchers and industry to establish mission-oriented research centers to accelerate the development of neuromorphology technologies. This has worked well in areas such as quantum technologies and nanotechnology – the US National Nano Initiative illustrates this well10Provides focus and motivation. These centers may be physical or virtual but should bring together the best researchers in various fields. Their approach should be different from that of traditional electronic technologies where each level of abstraction (materials, devices, circuits, systems, algorithms, applications) belongs to a different community. We need comprehensive and synchronized design across the entire suite. It is not enough for circuit designers to consult computational neuroscientists before designing systems; Engineers and neuroscientists must work together throughout the process to ensure that bio-inspired principles are as fully integrated into the devices as possible. Interdisciplinary co-creation must be at the heart of our approach. Research centers should include a wide range of researchers.

Besides the required physical and financial infrastructure, we need a trained workforce. Electronics engineers are rarely exposed to ideas from neuroscience, and vice versa. Circuit designers and physicists may have a passing knowledge of neurons and synapses but are unlikely to be familiar with the latest in computational neuroscience. There is a strong argument for preparing master’s courses and doctoral training programs for the development of neuromorphic engineers. UK Research Councils sponsor Doctoral Training Centers (CDTs), which are focused programs that support areas of specific need for trained researchers. CDTs can be a single or multiple organization; There are significant benefits for organizations collaborating on these programs by creating complementary teams across institutional boundaries. Programs generally work closely with industry and build pools of highly skilled researchers in ways that traditional doctoral programs often do not. There is a good case to be made to develop something similar, to stimulate interaction between nascent neuroengineering communities and provide the next generation of researchers and research leaders. Leading examples include the Groningen Cognitive Systems and Research Materials Program, which aims to train dozens of doctoral students specifically in materials for cognitive systems (AI)11Master’s Program in Neuroengineering at the Technical University of Munich12; ETH courses in Zurich in Analog Circuit Design for Neuromorphology13; Large-scale neural modeling at Stanford University14; and development of optical neural systems at the Institute of Microelectronics in Seville15th. There is room to do more.

Similar approaches can work at the transnational level. As always in research, collaboration is most successful when the best work with the best, regardless of boundaries. In such an interdisciplinary endeavour as neural computing, this is critical, so there is no doubt that international research networks and projects have a role to play. Early examples include the European Union for Neurotechnology16with a focus on Neural Computing Technologies, as well as the Chua Memristor Center at the University of Dresden17, which brings together many of the leading researchers in the field of memristor across materials, devices, and algorithms. Again, more can and must be done.

How can this be made attractive to governments? The government’s commitment to more energy-efficient, bio-inspired computing could be part of a broader push to decarbonise. This will not only tackle climate change, but also accelerate the emergence of new, low-carbon industries around big data, the Internet of Things, healthcare analytics, drug and vaccine discovery modeling, and robotics, among others. If current industries rely on traditional digital data analysis on a large scale, they are increasing the cost of energy while delivering suboptimal performance. We can instead create a virtuous circle in which we reduce the carbon footprint of the knowledge technologies that will drive the next generation of disruptive industries, and in doing so create a range of new neuro industries.

If that sounds difficult, consider quantum technologies. In the UK, the government has so far allocated around £1 billion to a range of quantum initiatives, largely under the umbrella of the National Quantum Technologies Programme. Bringing industry and academia together, a series of research centers translate quantum science into technologies targeting sensors, measurement, imaging, communications and computing. The National Center for Discrete Quantum Computing draws on the work of hubs and other researchers to provide demonstration hardware and software for the development of a general-purpose quantum computer. China has established a multi-billion dollar Chinese national laboratory for quantum information science, and the United States of America in 2018 commissioned a national strategic overview of quantum information science.18resulting in a five-year investment of $1.2 billion, as well as supporting a group of national quantum research centers19. Thanks to this research work, there has been a global rush to create quantum technology companies. One analysis found that in 2017 and 2018, private corporate funding amounted to $450 million20. There is no such joint support for neural computing, even though the technology is more well-established than quantum, and despite its ability to disrupt current AI technologies in a much shorter time horizon. Among the three strands of future computing in our vision, the neural form is woefully underinvested.

Finally, a few words about what the COVID-19 pandemic may hold for our arguments. There is a growing consensus that the crisis has precipitated many of the developments already underway: for example, the transition to more homework. Although reducing commuting and travel has direct benefits – some estimates suggest a reduction in global carbon dioxide2 As a result of the crisis by up to 17%21New ways of working have a cost. To what extent will the carbon savings from reduced travel be offset by increased data center emissions? If anything, the COVID pandemic further underscores the need to develop low-carbon computing technologies such as neural systems.

Our message on how to realize the potential of neural systems is clear: provide targeted support for collaborative research through the establishment of research centers of excellence; Provide agile financing mechanisms to enable rapid progress; Provide mechanisms for close cooperation with industry to bring in commercial financing and generate new spin-offs and start-ups, similar to the schemes already in place for quantum technology; Develop training programs for the next generation of neuroscientists and entrepreneurs; And we do all of this quickly and on a large scale.

Neural computing has the potential to transform our approach to artificial intelligence. Thanks to the coupling of new technologies and the huge and growing demand for effective AI, we have an opportunity. Bold thinking and bold initiatives are needed to support this thinking. Shall we seize the opportunity?

Source link