Tag Archives: Peter Clarke

A step closer to artificial synapses courtesy of memristors

Researchers from HRL Laboratories and the University of Michigan have built what they claim is a type of artificial synapse by using memristors. From the March 29, 2012 news item on Nanowerk,

In a step toward computers that mimic the parallel processing of complex biological brains, researchers from HRL Laboratories, LLC, and the University of Michigan have built a type of artificial synapse.

They have demonstrated the first functioning “memristor” array stacked on a conventional complementary metal-oxide semiconductor (CMOS) circuit. Memristors combine the functions of memory and logic like the synapses of biological brains.

The researchers developed a vertically integrated hybrid electronic circuit by combining the novel memristor developed at the University of Michigan with wafer scale heterogeneous process integration methodology and CMOS read/write circuitry developed at HRL. “This hybrid circuit is a critical advance in developing intelligent machines,” said HRL SyNAPSE program manager and principal investigator Narayan Srinivasa. “We have created a multi-bit fully addressable memory storage capability with a density of up to 30 Gbits/cm², which is unprecedented in microelectronics.”

Industry is seeking hybrid systems such as this one, the researchers say. Dubbed “R-RAM,” they could shatter the looming limits of Moore’s Law, which predicts a doubling of transistor density and therefore chip speed every two years.

“We’re reaching the fundamental limits of transistor scaling. This hybrid integration opens many opportunities for greater memory capacity and higher performance of conventional computers.  It has great potential in future non-volatile memory that would improve upon today’s Flash, as well as reconfigurable circuits,” said Wei Lu, an associate professor at the U-M Department of Electrical Engineering and Computer Science whose group developed the memristor array.

This work is being done as part of a DARPA (Defense Advanced Research Projects Agency) project titled, SyNAPSE, from the news item,

The work is part of the Defense Advanced Research Projects Agency’s (DARPA) SyNAPSE Program, or Systems of Neuromorphic Adaptive Plastic Scalable Electronics. Since 2008, the HRL-led SyNAPSE team has been developing a new paradigm for “neuromorphic computing” modeled after biology.

While I haven’t come across HRL Laboratories before, I have mentioned Dr. Wei Lu and his work with memristors in my April 15, 2010 posting. As for HRL Laboratories, they were founded in 1948 by Howard Hughes as the Hughes Research Laboratories (from the company’s History page),

HRL Laboratories continues the legacy of technology advances that began at Hughes Research Laboratories, established by Howard Hughes in 1948. HRL Laboratories, LLC, was organized as a limited liability company (LLC) on December 17, 1997 and received its first patent on September 12, 2000. With more than 750 patents to our name since then and counting, we’re proud of our talented group of researchers, who continue the long tradition of technical excellence in innovation.

First Laser
One of Hughes’ most notable achievements came in 1960 with the demonstration of the world’s first laser which used a synthetic ruby crystal. The ruby laser became the basis of a multibillion-dollar laser range finder business for Hughes. In 2010 during the 50th anniversary of the laser, HRL was designated a Physics Historic Site by the American Physical Society and was selected an IEEE Milestones location as the facility where the first working laser was demonstrated.

HRL has organized its researchers in a number of teams, the one of most interest in this context is the Center for Neural and Emergent Systems,

Part of HRL’s Information and Systems Sciences Laboratory, the Center for Neural and Emergent Systems (CNES) is dedicated to exploring and developing an innovative neural & emergent computing paradigm for creating intelligent, efficient machines that can interact with, react and adapt to, evolve, and learn from their environments.

CNES was founded on the principle that all intelligent systems are open thermodynamic systems capable of self-organization, whereby structural order emerges from disorder as a natural consequence of exchanging energy, matter or entropy with their environments.

These systems exist in a state far from equilibrium where the evolution of complex behaviors cannot be readily predicted from purely local interactions between the system’s parts. Rather, the emergent order and structure of the system arises from manifold interactions of its parts. These emergent systems contain amplifying-damping loops as a result of which very small perturbations can cause large effects or no effect at all. They become adaptive when the component relationships within the system become tuned for a particular set of tasks.

CNES promotes the idea that the neural system in the brain is an example of such a complex adaptive system. A key goal of CNES is to explain how computations in the brain can help explain the realization of complex behaviors such as perception, planning, decision making and navigation due to brain-body-environment interactions.

This has reminded me of HP Labs and their work with memristors (I have many postings, too many to list here) and understand that they will be rolling out ‘memristor-based’ products in 2013. From the  Oct. 8, 2011 article by Peter Clarke for EE Times,

The ‘memristor’ two-terminal non-volatile memory technology, in development at Hewlett Packard Co. since 2008, is on track to be in the market and taking share from flash memory within 18 months, according to Stan Williams, senior fellow at HP Labs.

“We have a lot of big plans for it and we’re working with Hynix Semiconductor to launch a replacement for flash in the summer of 2013 and also to address the solid-state drive market,” Williams told the audience of the International Electronics Forum, being held here [Seville, Spain].

ETA June 11, 2012: New artificial synapse development is mentioned in George Dvorsky’s June 11, 2012 posting (on the IO9.com website) about a nanoscale electrochemical switch developed by researchers in a Japan.

To be or not to be the memristor?

The memristor (aka, memresistor), for anyone not familiar with it, is a contested ‘new’ circuit element. In my April 5, 2010 posting I gave a brief overview of the history as I understood it (the memristor was a new addition to the traditional circuit elements [the capacitor, the resistor, and the inductor]) and in my April 7, 2010 posting I conducted an interview with Forrest H Bennett III who presented an alternative view to the memristor as ‘new’ circuit element discussion.

Discussion has continued on and off since then but in the last few weeks it has become more topical with the publication of a paper (Memresistors and non-memristive zero-crossing hysteresis curves) by Blaise Mouttet at arXiv.org on Jan. 12, 2012.

I don’t feel competent to summarize the gist of Blaise’s paper so I’m excerpting a passage *from* Peter Clarke’s Jan. 18, 2012 article for EE (Electronic Engineering) Times,

Blaise Mouttet argues that the interpretation of the memristor as a fourth fundamental circuit element – after the resistor, capacitor and inductor – was incorrect and that the memory device under development at HP Labs is not actually a memristor but part of a broader class of variable resistance systems.

Since publishing his arXiv paper Mouttet has also been in discussion with an e-mailing list of researchers into non-volatile memory device physics.

Some e-mail correspondents have come out in favor of Mouttet’s position stating that trying to define any two-terminal device in which the resistance can be altered by the current passed through the device as a memristor, adds nothing to the understanding of a complex field in which there are many types of device.

The article and the comments that follow (quite interesting and technical) are worth reviewing if this area of nanoelectronics interests you.

HP Labs has responded to Blaise’s paper and subsequent debate, and before included an excerpt from the response, I want to include a few passages from Blaise’s paper,

The “memristor” was originally proposed in 1971by Leon Chua as a missing fourth fundamental circuit element linking magnetic flux and electric charge. In 2008 a group of scientists from HP led by Stan Williams claimed to have discovered this missing memristor . It is my position that HP’s “memristor” claim lacks any scientific merit. My position is not that the HP researchers have presented an incorrect model of a memristor or even an incorrect model of resistance memory. If this were the case it would not be so bad because an incorrect model could at least be proven incorrect and possibly corrected to produce a better model. My position is that the HP researchers have avoided presenting any scientifically testable model at all by hiding behind the reputation of Leon Chua and the mythology of the memristor. They have thus attempted to bypass the principle of the scientific method.

If the HP researchers had developed a realistic model for resistive memory (whether it is called “memristor” or by some other name) it could be vetted by other researchers, compared to experimental data, and determined to be true or false. If necessary the model could be modified or corrected and an improved version of the model could be produced.

This is not what has happened. (p. 1 PDF)

Here’s my excerpt of HP’s response (from Peter Clarke’s Jan. 20, 2012 article for EE Times),

The spokesperson said in email: “HP is proud of the research it has undertaken into memristor technology and the recognition this has received in the scientific community. In a little over three years, our papers, which were subject to rigorous peer review before being published in leading scientific journals, have been cited more than 1,000 times by other researchers in the field. We continue this research and collaboration with the electronics industry to bring this important technology to market.”

Deciding what something is and how fits into our understanding of how the world operates, in this case, a new circuit element, or not, has consequences beyond the actual discussion. If science is the process of posing questions, we need to test the assumptions we make (in this case, whether or not the memristor is a fourth circuit element or part of a larger system of variable resistance systems) as they can define the questions we’ll ask in the future.

As I noted earlier, I’m not competent to draw any conclusions as to which party may have the right approach but I am glad to see the discussion taking place.

*’from’ added on Sept. 27, 2016.