 Ruh-roh, here is another science news item
sure to memristor denier and über nimrod Tim H.,
who for years sent me harassing e-mails to inform me of how memristors are only
theoretical entities. He got downright abusive and threatening. Maybe someday I'll
reveal his identity ... but I digress. This article from the SciTechDaily website
reports on work done at the University of Southern California. It begins: "Researchers
have made
significant advances in memristor technology, enhancing its precision and efficiency.
This innovation promises to bridge the gap between analog and digital computing,
offering faster, more energy-efficient processing suitable for AI, machine learning,
and beyond..."
 Considering medical-diagnosis and other
safety-critical, sensory-processing applications that require accurate decisions
based on a small amount of noisy input data, the study notes that while Bayesian
neural networks excel at such tasks because they provide predictive uncertainty
assessment, their probabilistic nature requires increased use of energy and computation.
The increase is caused by the fact that implementing the networks in hardware requires
a random number generator to store the probability distributions, i.e. synaptic
weights. "Our paper presents, for the first time, a complete hardware implementation
of a Bayesian neural network utilizing the
intrinsic variability of memristors to store these probability distributions,"
said Elisa Vianello, CEA-Leti chief scientist. "We exploited the intrinsic variability
of memristors to store these probability distributions, instead of using random
number generators." A team comprising CEA-Leti, CEA-List and two CNRS laboratories...
"In the latest episode of Brains and Machines, EE Times regular
Dr. Sunny Bains talks to Professor Melika Payvand, who designs neural systems from
the circuit-level up at the Institute of Neuroinformatics in Zurich. You’ll find
out the role that memristors are playing in the systems she designs, why neural
circuits need to operate at different timescales, and why copying some features
of biological dendrites could add computational power to silicon brains. Discussion
follows with Dr. Giulia D'Angelo from the Italian Institute of Technology and Professor
Ralph Etienne-Cummings from Johns Hopkins University. Welcome to Brains and Machines,
a deep dive into neuromorphic engineering and biologically inspired technology.
In this episode, EE Times regular Sunny Bains talks to Professor Melika Payvand,
who designs neural systems from the circuit-level up at the Institute of Neuroinformatics
in Zurich. You'll find out the role that memristors are playing in the systems she
designs..."
 Uh-oh, I wonder if the dummkopf who
keeps writing to call me an idiot for claiming
memristors are real will crawl out of his hole again? "The world's first fully
system-integrated memristor chip has been unveiled by a team of Chinese scientists
who believe it could not only make artificial intelligence smarter, but also more
time and energy efficient. While the semiconductor has yet to leave the lab setting,
it could allow for the development of AI that is capable of more human-like learning,
which could have implications for the way smart devices and autonomous driving work,
according to the researchers. 'Learning is highly important,' for edge intelligence
devices, the research team from Tsinghua University said in their study released
in the journal Science on September 15, referencing devices that process data internally
with technology like AI..."
 "Researchers at Hewlett Packard Labs, where
the first practical memristor was created, have invented a new variation on the
device - a
memristor laser..." Also, "From
Transistor to Memristor: Switching Technologies for the Future."
 Here's another news tidbit to tweak the
idiot who keeps harassing me for daring to claim
memristors are real
entities. "These brain-mimicking devices boast tiny energy budgets and hardened
circuits. Memristive devices that mimic neuron-connecting synapses could serve as
the hardware for neural networks that copy the way the brain learns. Now two new
studies may help solve key problems these components face not just with yields and
reliability, but with finding applications beyond neural nets. Memristors, or memory
resistors, are essentially switches that can remember which electric state they
were toggled to after their power is turned off. Scientists worldwide aim to use
memristors and similar components to build electronics that, like neurons, can both
compute and store data. Such brain-inspired neuromorphic hardware may also prove
ideal for implementing neural networks - AI systems increasingly finding use in
applications such as analyzing medical scans and empowering autonomous vehicles.
However, current memristive devices typically rely on emerging technologies with
low production yields and unreliable electronic performance..."
 The
term memristor - a portmanteau
of "memory" and "resistor" - is the fourth fundamental electronic component,
along with the resistor,
capacitor, and
inductor. The name was coined in 1971, which sounds like yesterday to someone
like me (born in 1958), but incredibly that is now half a century ago.
Until fairly recently, the memristor was merely a theoretical curiosity existing
in academic papers. In April of 2008, HP Labs (Hewlett-Packard) reported on successfully
building a nanoscale
memristor in their R&D lab. As with all new technologies, since that time
much progress has been made.
 To the left is a conceptual diagram illustrating
the symmetry of the four basic circuit components - the resistor, the capacitor,
the inductor, and the memristor. Per
Wikipedia: "Chua in his 1971 paper identified a theoretical symmetry between
the non-linear resistor (voltage vs. current), non-linear capacitor (voltage vs.
charge), and non-linear inductor (magnetic flux linkage vs. current). From this
symmetry he inferred the characteristics of a fourth fundamental non-linear circuit
element, linking magnetic flux and charge, which he called the memristor. In contrast
to a linear (or non-linear) resistor, the memristor has a dynamic relationship between
current and voltage, including a memory of past voltages or currents..."
A nimrod named Tim H. kept contacting
me a few years ago to say there is no such thing as a real memristor and called me bad names while
doing so. Maybe someday I'll publish his messages along with his name, e-mail address
and things I've discovered about him on the Internet. Funny, I haven't heard
from the dope in quite a while. Tim, if you're out there, let me know if you
want to be famous.
 "Over the past few decades, the performance
of machine learning models on various real-world tasks has improved significantly.
Training and implementing most of these models, however, still requires vast amounts
of energy and computational power. Engineers worldwide have thus been trying to
develop alternative hardware solutions that can run artificial intelligence models
more efficiently, as this could promote their widespread use and increase their
sustainability. Some of these solutions are based on
memristors, memory devices that can store information without consuming energy.
Researchers at Université Paris-Saclay- CNRS, Université Grenoble-Alpes-CEA-LETI,
HawAI.tech, Sorbonne Université, and Aix-Marseille Université-CNRS have recently
created a so-called Bayesian machine (i.e., an AI approach that performs computations
based on Bayes' theorem), using memristors. Their proposed system, introduced in
a paper published in Nature Electronics, was found to be significantly more energy-efficient
than currently employed hardware solutions. "Artificial intelligence is making major
progress..."
 "A chip consisting of
memristor crossbars was trained using a local on-chip learning algorithm. The
team demonstrated that their approach could accurately reconstruct Braille representations
of nine famous computer scientists from highly distorted inputs. Deep-learning models
have proven to be highly valuable tools for making predictions and solving real-world
tasks that involve the analysis of data. Despite their advantages, before they are
deployed in real software and devices such as cell phones, these models require
extensive training in physical data centers, which can be both time and energy consuming.
Researchers at Texas A&M University, Rain Neuromorphics and Sandia National
Laboratories have recently devised a new system for training deep learning models
more efficiently and on a larger scale. This system, introduced in a paper published
in Nature Electronics, relies on the use of new training algorithms and memristor
crossbar hardware, that can carry out multiple operations at once..."
 "Reservoir computing (RC) is an approach
for building computer systems inspired by current knowledge of the human brain.
Neuromorphic computing architectures based on this approach are comprised of dynamic
physical nodes, which combined can process spatiotemporal signals. Researchers at
Tsinghua University in China have recently created a new RC system based on
memristors, electrical components that regulate the flow of electrical current
in a circuit, while also recording the amount of charge that previously flowed through
it..."
Last updated January 13, 2024
|
