Nav: Home

Neural networks everywhere

February 14, 2018

CAMBRIDGE, MASS.--Most recent advances in artificial-intelligence systems such as speech- or face-recognition programs have come courtesy of neural networks, densely interconnected meshes of simple information processors that learn to perform tasks by analyzing huge sets of training data.

But neural nets are large, and their computations are energy intensive, so they're not very practical for handheld devices. Most smartphone apps that rely on neural nets simply upload data to internet servers, which process it and send the results back to the phone.

Now, MIT researchers have developed a special-purpose chip that increases the speed of neural-network computations by three to seven times over its predecessors, while reducing power consumption 94 to 95 percent. That could make it practical to run neural networks locally on smartphones or even to embed them in household appliances.

"The general processor model is that there is a memory in some part of the chip, and there is a processor in another part of the chip, and you move the data back and forth between them when you do these computations," says Avishek Biswas, an MIT graduate student in electrical engineering and computer science, who led the new chip's development.

"Since these machine-learning algorithms need so many computations, this transferring back and forth of data is the dominant portion of the energy consumption. But the computation these algorithms do can be simplified to one specific operation, called the dot product. Our approach was, can we implement this dot-product functionality inside the memory so that you don't need to transfer this data back and forth?"

Biswas and his thesis advisor, Anantha Chandrakasan, dean of MIT's School of Engineering and the Vannevar Bush Professor of Electrical Engineering and Computer Science, describe the new chip in a paper that Biswas is presenting this week at the International Solid State Circuits Conference.

Back to analog

Neural networks are typically arranged into layers. A single processing node in one layer of the network will generally receive data from several nodes in the layer below and pass data to several nodes in the layer above. Each connection between nodes has its own "weight," which indicates how large a role the output of one node will play in the computation performed by the next. Training the network is a matter of setting those weights.

A node receiving data from multiple nodes in the layer below will multiply each input by the weight of the corresponding connection and sum the results. That operation - the summation of multiplications - is the definition of a dot product. If the dot product exceeds some threshold value, the node will transmit it to nodes in the next layer, over connections with their own weights.

A neural net is an abstraction: The "nodes" are just weights stored in a computer's memory. Calculating a dot product usually involves fetching a weight from memory, fetching the associated data item, multiplying the two, storing the result somewhere, and then repeating the operation for every input to a node. Given that a neural net will have thousands or even millions of nodes, that's a lot of data to move around.

But that sequence of operations is just a digital approximation of what happens in the brain, where signals traveling along multiple neurons meet at a "synapse," or a gap between bundles of neurons. The neurons' firing rates and the electrochemical signals that cross the synapse correspond to the data values and weights. The MIT researchers' new chip improves efficiency by replicating the brain more faithfully.

In the chip, a node's input values are converted into electrical voltages and then multiplied by the appropriate weights. Only the combined voltages are converted back into a digital representation and stored for further processing.

The chip can thus calculate dot products for multiple nodes - 16 at a time, in the prototype - in a single step, instead of shuttling between a processor and memory for every computation.

All or nothing

One of the keys to the system is that all the weights are either 1 or -1. That means that they can be implemented within the memory itself as simple switches that either close a circuit or leave it open. Recent theoretical work suggests that neural nets trained with only two weights should lose little accuracy - somewhere between 1 and 2 percent.

Biswas and Chandrakasan's research bears that prediction out. In experiments, they ran the full implementation of a neural network on a conventional computer and the binary-weight equivalent on their chip. Their chip's results were generally within 2 to 3 percent of the conventional network's.
-end-


Massachusetts Institute of Technology

Related Memory Articles:

Taking photos of experiences boosts visual memory, impairs auditory memory
A quick glance at any social media platform will tell you that people love taking photos of their experiences -- whether they're lying on the beach, touring a museum, or just waiting in line at the grocery store.
Think you know how to improve your memory? Think again
Research from Katherine Duncan at the University of Toronto suggests we may have to rethink how we improve memory.
Improving memory with magnets
The ability to remember sounds, and manipulate them in our minds, is incredibly important to our daily lives -- without it we would not be able to understand a sentence, or do simple arithmetic.
Who has the better memory -- men or women?
In the battle of the sexes, women have long claimed that they can remember things better and longer than men can.
New study of the memory through optogenetics
A collaboration between Universitat Autònoma de Barcelona and Harvard University pioneers the increase of memory using optogenetics in mice in Spain.
Peppermint tea can help improve your memory
Peppermint tea can improve long-term and working memory and in healthy adults.
A new glimpse into working memory
MIT study finds bursts of neural activity as the brain holds information in mind, overturns a long-held model.
Memory ensembles
For over forty years, neuro-scientists have been interested in the biological mechanisms underlying the storage of the information that our brain records every day.
What is your memory style?
Why is it that some people have richly detailed recollection of past experiences (episodic memory), while others tend to remember just the facts without details (semantic memory)?
Watching a memory form
Neuroscientists at Rosalind Franklin University of Medicine and Science have discovered a novel mechanism for memory formation.

Related Memory Reading:

Best Science Podcasts 2019

We have hand picked the best science podcasts for 2019. Sit back and enjoy new science podcasts updated daily from your favorite science news services and scientists.
Now Playing: TED Radio Hour

Changing The World
What does it take to change the world for the better? This hour, TED speakers explore ideas on activism—what motivates it, why it matters, and how each of us can make a difference. Guests include civil rights activist Ruby Sales, labor leader and civil rights activist Dolores Huerta, author Jeremy Heimans, "craftivist" Sarah Corbett, and designer and futurist Angela Oguntala.
Now Playing: Science for the People

#521 The Curious Life of Krill
Krill may be one of the most abundant forms of life on our planet... but it turns out we don't know that much about them. For a create that underpins a massive ocean ecosystem and lives in our oceans in massive numbers, they're surprisingly difficult to study. We sit down and shine some light on these underappreciated crustaceans with Stephen Nicol, Adjunct Professor at the University of Tasmania, Scientific Advisor to the Association of Responsible Krill Harvesting Companies, and author of the book "The Curious Life of Krill: A Conservation Story from the Bottom of the World".