Technology Review – The properties of memelements that make them so good at biological computing has been hard to pin down. Which is where Di Ventra and Pershin come in. They have distilled the essential properties that ought to allow memelements to match the brain’s performance.
They say these properties include the ability to store information over long periods; the ability to act collectively so that the state of a memdevice as a whole depends on the states of all its memelements; a robustness against noise and small imperfections; and so on.
Perhaps the most important, however, is the ability to store and process information at the same time, a property that is entirely alien in the conventional computing world
This is an interesting approach that attempts to crystallise the best way to approach memcomputing. And it has huge potential. Memcapacitors and meminductors essentially consume no energy and so ought to allow very low energy applications. That should make it possible for them to approach the energy efficiency of natural systems for the first time.
“An important milestone in this field would be the demonstration of a memcomputing device with computing capabilities and power consumption comparable to (or better than) those of the human brain,” say Di Ventra and Pershin.
In present day technology, storing and processing of information occur on physically distinct regions of space. Not only does this result in space limitations; it also translates into unwanted delays in retrieving and processing of relevant information. There is, however, a class of two-terminal passive circuit elements with memory, memristive, memcapacitive and meminductive systems – collectively called memelements – that perform both information processing and storing of the initial, intermediate and ﬁnal computational data on the same physical platform. Importantly, the states of these memelements adjust to input signals and provide analog capabilities unavailable in standard circuit elements, resulting in adaptive circuitry, and providing analog massively-parallel computation. All these features are tantalizingly similar to those encountered in the biological realm, thus oﬀering new opportunities for biologically-inspired computation. Of particular importance is the fact that these memelements emerge naturally in nanoscale systems, and are therefore a consequence and a natural by-product of the continued miniaturization of electronic devices. We will discuss the various possibilities oﬀered by memcomputing, discuss the criteria that need to be satisﬁed to realize this paradigm, and provide an example showing the solution of the shortest-path problem and demonstrate the healing property of the solution path.
While quantum computing relies on the superposition of states, memcomputing utilizes the collective dynamics of a large number of (essentially classical) systems. Its speciﬁc criteria are then as follows.
1. Scalable massively-parallel architecture with combined information processing and storage
2. Suﬃciently long information storage times
3. The ability to initialize memory states
4. Mechanism(s) of collective dynamics, strong “memory content”
5. The ability to read the ﬁnal result (from relevant memelements)
6. Robustness against small imperfections and noise
There are several schemes that have been recently suggested that satisfy all or some of the above criteria. These schemes include neuromorphic computing with memristive synapses, massively-parallel computing with memristive networks, logic with memory circuit elements, and memristive cellularautomata. For instance, in the work of Ref. 11 a memristive network has been used to solve a popular optimization problem, namely the maze problem.
Criterion 1 is fully satisﬁed by that network: the solution of the maze is done in an analog massively-parallel fashion, and it is locally stored in the system for essentially an unlimited time (criterion 2). The network can also be initialized easily as explained in Ref. 11, thus satisfying criterion 3. The dynamics of the system is collective, and the diﬀerence between the low-resistance state and the high-resistance state was chosen in order to easily perturb all memelements in the system (criterion 4). Although not explicitly discussed in that work, criterion 5
can be easily accomplished with an appropriate choice of memelements. Finally, criterion 6 was naturally built in the problem: any change of topology of the network–and consequent emergence of new maze solution(s)–could be handled eﬀortlessly. Similar considerations would apply to the other schemes proposed in the literature.
We note at this point that although they are not extensively studied yet, memcapacitors and meminductors can also be used in the above memcomputing schemes by replacing memristors, albeit in a modiﬁed form. Since memcapacitors and meminductors may in principle be constructed to consume little or virtually no energy, their use in memcomputing is potentially energetically more eﬃcient than the use of memristors. An important milestone in this ﬁeld would be the demonstration of a memcomputing device with computing capabilities and power consumption comparable to (or better than) those of the human brain.
In conclusion, we have discussed the concept of memcomputing: storing and processing of information on the same physical platform. In particular, we have outlined the main criteria that need to be satisﬁed in order to realize such a paradigm and analyzed a speciﬁc example to show the healing properties of the solution. Unlike other promising but more speculative proposals, like quantum computing, memcomputing is already a practical reality, at least in regard to some applications, such as digital logic. It bypasses several of the bottlenecks of present day computing architectures and its constitutive units – memristors, memcapacitors, and meminductors – are already widely available. Indeed, these elements emerge quite naturally with increasing miniaturization of electronic devices. The computational possibilities oﬀered by this paradigm are varied, and due to its tantalizing similarities both with some features of the brain as well as with the collective properties of colonies of living organisms, it promises to open new directions in neuromorphic architectures and biological studies.
Brian Wang is a Futurist Thought Leader and a popular Science blogger with 1 million readers per month. His blog Nextbigfuture.com is ranked #1 Science News Blog. It covers many disruptive technology and trends including Space, Robotics, Artificial Intelligence, Medicine, Anti-aging Biotechnology, and Nanotechnology.
Known for identifying cutting edge technologies, he is currently a Co-Founder of a startup and fundraiser for high potential early-stage companies. He is the Head of Research for Allocations for deep technology investments and an Angel Investor at Space Angels.
A frequent speaker at corporations, he has been a TEDx speaker, a Singularity University speaker and guest at numerous interviews for radio and podcasts. He is open to public speaking and advising engagements.