IBM Scientists Show Blueprints for Brainlike Computing - MIT Technology Review

IBM Scientists Show Blueprints for Brainlike Computing - MIT Technology Review: “Programs” are written using special blueprints called corelets. Each corelet specifies the basic functioning of a network of neurosynaptic cores. Individual corelets can be linked into more and more complex structures—nested, Modha says, “like Russian dolls.”

The Flaw Lurking In Every Deep Neural Net

The Flaw Lurking In Every Deep Neural Net: "For all the networks we studied, for each sample, we always manage to generate very close, visually indistinguishable, adversarial examples that are misclassified by the original network."
To be clear, the adversarial examples looked to a human like the original, but the network misclassified them. You can have two photos that look not only like a cat but the same cat, indeed the same photo, to a human, but the machine gets one right and the other wrong.

This Could Be the First Animal to Live Entirely Inside a Computer

This Could Be the First Animal to Live Entirely Inside a Computer: "Our project is to simulate as much of the important physics — or biophysics — of the C. elegans as we can, and then compare against measurements from real worms. When we say simulation, we are specifically referring to writing computer programs that use equations from physics that are applied to what we know about the worm..."

"We are currently addressing the challenge of closing the 'brain-behavior loop' in C. elegans," he says. "In other words, through this simulation we want to understand how its proto-brain controls its muscles to move its body around an environment, and then how the environment is interpreted by the proto-brain. That means leaving aside reproduction or digestion or other internal functions for now until that first part is complete. Once we get there, we will move on to these other aspects.

Zuckerberg, Musk, and Kutcher Want to Build You a New Brain | Enterprise | WIRED

Zuckerberg, Musk, and Kutcher Want to Build You a New Brain | Enterprise | WIRED: ...a $40 million investment in a new kind of artificial intelligence called Vicarious...

Vicarious co-founder Dileep George previously built a similar company called Numenta...

This is the second big funding round for Vicarious, and it was lead by venture capital outfit Formation 8. The first $15 million round included investments by Facebook and Asana co-founder Dustin Moskovitz; former Facebook CTO and Quora founder Adam D’Angelo; PayPal and Palantir co-founder Peter Thiel; and Palantir co-founder Joe Lonsdale.

IBM unveils concept for a future brain-inspired 3D computer | KurzweilAI

IBM unveils concept for a future brain-inspired 3D computer | KurzweilAI: IBM has unveiled a prototype of a new brain-inspired computer powered by what it calls “electronic blood,” BBC News reports.

The firm says it is learning from nature by building computers fueled and cooled by a liquid, like our minds...

Its new “redox flow” system pumps an electrolyte “blood” through a computer, carrying power in and taking heat out.

Stem cells mimic human brain : Nature News & Comment

Stem cells mimic human brain : Nature News & Comment: ...in the latest advance, scientists developed bigger and more complex neural-tissue clumps by first growing the stem cells on a synthetic gel that resembled natural connective tissues found in the brain and elsewhere in the body. Then, they plopped the nascent clumps into a spinning bath to infuse the tissue with nutrients and oxygen...

Under a microscope, researchers saw discrete brain regions that seemed to interact with one another. But the overall arrangement of the different proto-brain areas varied randomly across tissue samples — amounting to no recognizable physiological structure.

“The entire structure is not like one brain,” says Knoblich, adding that normal brain maturation in an intact embryo is probably guided by growth signals from other parts of the body. The tissue balls also lacked blood vessels, which could be one reason that their size was limited to 3–4 millimetres in diameter, even after growing for 10 months or more.

IBM Scientists Show Blueprints for Brainlike Computing | MIT Technology Review

IBM Scientists Show Blueprints for Brainlike Computing | MIT Technology Review: Modha’s team has also developed software that runs on a conventional supercomputer but simulates the functioning of a massive network of neurosynaptic cores—with 100 trillion virtual synapses and two billion neurosynaptic cores.

Each core of the simulated neurosynaptic computer contains its own network of 256 “neurons,” which operate using a new mathematical model. In this model, the digital neurons mimic the independent nature of biological neurons, developing different response times and firing patterns in response to input from neighboring neurons.

“Programs” are written using special blueprints called corelets. Each corelet specifies the basic functioning of a network of neurosynaptic cores. Individual corelets can be linked into more and more complex structures—nested, Modha says, “like Russian dolls.”

Slime mould could make memristors for biocomputers

Slime mould could make memristors for biocomputers: The feeding fronds of the slime mould Physarum polycephalum turn out to have memory resistance – or memristance...

..."Slime mould can be used to perform all the logic functions that conventional computer hardware components can do," says Gale.

Her team is also exploring whether, in addition to number-crunching, slime mould's knack for finding the shortest path to nutrients can be used to design the most efficient circuit patterns for biocomputers.

nsf.gov - National Science Foundation (NSF) Discoveries - Energy Efficient Brain Simulator Outperforms Supercomputers - US National Science Foundation (NSF)

nsf.gov - National Science Foundation (NSF) Discoveries - Energy Efficient Brain Simulator Outperforms Supercomputers - US National Science Foundation (NSF): Each of Neurogrid's 16 chips contains more than 65,000 silicon "neurons" whose activity can be programmed according to nearly 80 parameters, allowing the researchers to replicate the unique characteristics of different types of neurons. Soft-wired "synapses" crisscross the board, shuttling signals between every simulated neuron and the thousands of neurons it is networked with, effectively replicating the electrical chatter that constitutes communication in the brain.

Precisely engineering 3-D brain tissues

Precisely engineering 3-D brain tissues: To mimic this architectural complexity in their engineered tissues, the researchers embedded a mixture of brain cells taken from the primary cortex of rats into sheets of hydrogel. They also included components of the extracellular matrix, which provides structural support and helps regulate cell behavior.

Those sheets were then stacked in layers, which can be sealed together using light to crosslink hydrogels. By covering layers of gels with plastic photomasks of varying shapes, the researchers could control how much of the gel was exposed to light, thus controlling the 3-D shape of the multilayer tissue construct.

This type of photolithography is also used to build integrated circuits onto semiconductors — a process that requires a photomask aligner machine, which costs tens of thousands of dollars. However, the team developed a much less expensive way to assemble tissues using masks made from sheets of plastic, similar to overhead transparencies, held in place with alignment pins.

The tissue cubes can be made with a precision of 10 microns, comparable to the size of a single cell body. At the other end of the spectrum, the researchers are aiming to create a cubic millimeter of brain tissue with 100,000 cells and 900 million connections.

Simulated brain scores top test marks : Nature News & Comment

Simulated brain scores top test marks : Nature News:  It stands apart from other attempts to simulate a brain, such as the ambitious Blue Brain Project (see 'Brain in a box'), because it produces complex behaviours with fewer neurons...


A pure computer simulation, Spaun simulates the physiology of each of its neurons, from spikes of electricity that flow through them to neurotransmitters that cross between them. The computing cells are divided into groups, corresponding to specific parts of the brain that process images, control movements and store short-term memories. These regions are wired together in a realistic way, and even respond to inputs that mimic the action of neurotransmitters.

As Spaun sees a stream of numbers, it extracts visual features so that it can recognize the digits. It can then perform at least eight different tasks, from simple ones like copying an image, to more complex ones similar to those found on IQ tests, such as finding the next number in a series. When finished, it writes out its answer with a physically modelled arm.

IBM simulates 530 billon neurons, 100 trillion synapses on supercomputer | KurzweilAI

IBM simulates 530 billon neurons, 100 trillion synapses on supercomputer | KurzweilAI: IBM and LBNL achieved an unprecedented scale of 2.084 billion neurosynaptic cores* containing 53×1010  (530 billion) neurons and 1.37×1014 (100 trillion) synapses running only 1542 times slower than real time.

“We have not built a biologically realistic simulation of the complete human brain,” explains an abstract....  “Computation (‘neurons’), memory (‘synapses’), and communication (‘axons,’ ‘dendrites’) are mathematically abstracted away from biological detail toward engineering goals of maximizing function (utility, applications) and minimizing cost (power, area, delay) and design complexity of hardware implementation.”

New Scientist TV: Frankenoctopus unveils novel shape-shifting arms

New Scientist TV: Frankenoctopus unveils novel shape-shifting arms: The first prototype, currently on display at the Science Museum in London, has six silicone legs designed for locomotion, while two specialised arms use artificial muscles, motors and sensors to detect and grasp objects. A spring-like structure inside these tentacles, made from a shape-memory alloy, can expand, contract or bend in any direction with changes in temperature.

Cecilia Laschi of Sant'Anna School of Advanced Studies in Pisa, Italy, and colleagues control their robot's behaviour via a centralised unit that mimics the central nervous system of an octopus.

Google Puts Its Virtual Brain Technology to Work

Google Puts Its Virtual Brain Technology to Work: ...Google engineers published results of an experiment that threw 10 million images grabbed from YouTube videos at their simulated brain cells, running 16,000 processors across a thousand computers for 10 days without pause.

"Most people keep their model in a single machine, but we wanted to experiment with very large neural networks," says Jeff Dean, an engineer helping lead the research at Google. "If you scale up both the size of the model and the amount of data you train it with, you can learn finer distinctions or more complex features."

Ferroelectric memristors may lead to brain-like computers

Ferroelectric memristors may lead to brain-like computers: In a new study, a team of researchers from France, the UK, and Japan has demonstrated that a device called a ferroelectric tunnel junction (FTJ) that experiences voltage-controlled resistance variation represents a new class of memristor. Due to the FTJ's quasi-continuous resistance variations exceeding two orders of magnitude, along with its rapid 10-ns operation speed, the device could one day serve as the basic hardware of neuromorphic computational architectures, or computers that function like brains...

"We have conceptualized, designed and realized a completely new type of memristor that performs as well as classical ionic memristors, but operates through an electronic mechanism," coauthor Manuel Bibes, a CNRS research scientist, told Phys.org. "While this should have clear advantages in terms of reproducibility, the key breakthrough is that our ferroelectric memristor behaves according to well-established physical models. This allows a precise understanding of the memristive response, and also opens the door for engineering memristive functions on-demand."

‘Green Brain’ project to create an autonomous flying robot with a honey bee brain - News releases - News - The University of Sheffield

‘Green Brain’ project to create an autonomous flying robot with a honey bee brain - News releases - News - The University of Sheffield: The team will build models of the systems in the brain that govern a honey bee's vision and sense of smell. Using this information, the researchers aim to create the first flying robot able to sense and act as autonomously as a bee, rather than just carry out a pre-programmed set of instructions.

A Startup Tries to Make a Better Artificial Brain - Technology Review

A Startup Tries to Make a Better Artificial Brain - Technology Review: Vicarious, George says, is using a more sophisticated architecture and training its system with a video stream that varies over time...

Blue Brain project accurately predicts connections between neurons

Blue Brain project accurately predicts connections between neurons: To their great surprise, they found that the locations on the model matched that of synapses found in the equivalent real-brain circuit with an accuracy ranging from 75 percent to 95 percent...

This means that neurons grow as independently of each other as physically possible and mostly form synapses at the locations where they randomly bump into each other.
A few exceptions were also discovered, pointing out special cases where signals are used by neurons to change the statistical connectivity. By taking these exceptions into account, the Blue Brain team can now make a near perfect prediction of the locations of all the synapses formed inside the circuit.

A Brain Implant that Thinks

A Brain Implant that Thinks: The researchers used an array of electrodes to record the electrical activity of neurons in the prefrontal cortex of monkeys while they performed a memory task...
The five monkeys in the study were trained to play a matching game in which they were shown an image on a screen and then had to use hand movements to steer a cursor to that same image...
From their recordings in the prefrontal cortex, the research team extrapolated a mathematical model of the electrical activity of neurons involved in the movement decision...
In the new study, the model took multiple signals produced by the brain layer that integrates sensory information related to the task. It then extracted the relevant information to choose a particular movement. The implant can stimulate neurons in order to influence the decision to move the hand to select the correct image.

Artificial cerebellum enables human-like object handling by robots

Artificial cerebellum enables human-like object handling by robots: To solve this, University of Granada researchers have implemented a biologically inspired adaptive microcircuit based on a new cerebellar spiking model that adapts to corrections and stores sensory effects.

It also records motor commands to predict the action or movement to be performed by the robotic arm.

The biologically inspired architectures used in this model combine the error training approach with predictive adaptive control. The robot performs automatic learning and two control systems enable accurate and robust control of the robotic arm during object handling.