THINKING LIKE A HUMAN WITH MEMRISTORS

Muhammad Nauman EE 01083 060

Bio Medical Instrumentati on

HARDWARE VS. WETWARE Is there any difference between human brain and computer ?
Computer: Memory and processor are physically separated a physical distance exists. Steps to model a single synapse: Synapses state is located in the main memory. Signal originates in processor and packed to travel on bus for about 2 to 10 centimeters. Reaches memory and unpacked to actually access the memory. Such sequence is multiplied with 8000 to build a single neuron of rat. Brain: Storage and computation happen at the same time and in the same place. Importance of information is evaluated by contrasting it with previous state of synapse between Neurons. Computation takes during the information transfer.

THE GREAT BRAIN RACE - HISTORY

BLUE BRAIN: In 2005, Henry Markram along with his team used an IBM supercomputer to simulate one square centimeter of cerebral cor tex. C2: In 2009, IBM Almaden, the C2 simulator re-creates 1 billion neurons connected by 10 trillion individual synapses, or about the amount found in a small mammal. NEUROGRID: Kwabena Boahen at Stanford is developing a silicon chip. One of the goals of this research is to build ar tificial retinas to be used as medical implants for the blind. IFAT 4G: At Johns Hopkins University, implemented a visual cortex model for object recognition. BRAINSCALES: In the European Union's neuromorphic chip program, Started in January 2011 , the nonvon Neumann hardware included a complex neuron model with up to 16 000 synaptic inputs per neuron.

MAJOR PROBLEMS WITH COMPUTER TO BE USED AS MAMMALIAN BRAIN

On a standard computer, the memory and processor are separated by a data channel or bus. Data bus (channels) have fixed capacity. Processor reserves a small number of slots called registers for storing data during computation, the processor writes the results back to memory. Though modern processors have cache memory. Simple brain have tens of millions of neurons connected by billions of synapses, any attempt to simulate their interconnection and computational power requires a cache memory of processor as big as the computers main memory. High power consumption in contrast with brain which can operate at around 100 millivolts in most crucial state.

NEUROMORPHIC ARCHITECTURE
Why a biological brain is able to quickly execute this massive simultaneous information? And here is happens in a brain: Here's what the important part: By the time that information Neuron 2) (Consider two neurons; Neuron 1 and reaches the body of Neuron 2, there is only a and the resultant processing sent Neuron 1 gives an impulse, single valueallinformation ishas down thetaken to the synapsethe its target, Neuron 2. There is already axon place during of information transfer. The synapse of Neuron 2, having stored its own state locally, never any need for the brain to take information out of one evaluates the importance of the information coming fromto a neuron, spend time processing it, and then return it Neuron 1 by integratingof with its own previous state and the strength of its different set it neurons. connection to Neuron 1. Instead, in the mammalian brain, storage and processing Then, these two pieces of informationthe information from happen at the same time and in the same place. Computer Neuron 1 and the state of Neuron 2's synapseflow toward the scientists call it2a neuromorphic architecture. body of Neuron over the dendrites.

HOW TO BUILD NEUROMORPHIC ARCHITECTURE?

A true artificial intelligence could hypothetically run on conventional hardware, but it would be fantastically inefficient. So how do you build something that has an architecture like the brain's? Change the architecture to merge memory and computation . Memristor is the best technology. The concept wasnt new. The concept wasn't new. In 1971 , professor Leon Chua of the University of California, Berkeley, reasoned that a memristor would behave like a resistor with a conductance that changed as a function of its internal state and the voltage applied. a memristor could remember how much current had gone through it, it could work as an essentially nonvolatile memory . In 2008, HP labs had created a functioning memristor.

FOURTH FUNDAMENTAL ELECTRONICS COMPONENT MEMRISTOR

A memristor is a two-terminal device whose resistance changes depending on the amount, direction, and duration of voltage that's applied to it. But here's the really interesting thing about a memristor: Whatever its past state, or resistance, it freezes that state until another voltage is applied to change it. Maintaining that state requires no power. That's different from a dynamic RAM cell, which requires regular charge to maintain its state . A memristor is a two-terminal device whose resistance changes depending on the amount, direction, and duration of voltage that's applied to it. But here's the really interesting thing about a memristor: Whatever its past state, or resistance, it freezes that state until another voltage is applied to change it. Maintaining that state requires no power. That's different from a dynamic RAM cell, which requires regular charge to maintain its state.

STRUCTURE OF MEMRISTOR

MEMRISTOR VS. SYNAPSES

Memristor's "state" can be considered analogous to the state of the synapse as they preserve the previous amount of current passing from them. The state of the synapse depends how closely any two neurons are linked, which is a key part of the mammalian ability to learn new information.

MONETA: A MIND MADE OF MEMRISTORS

Human Cortex About 106 neurons per square centimeter About 1010 synapses per square centimeter About 2 milliwatts per square centimeter Total power consumption: 20 watts DARPA SyNAPSE Hardware Goals 106 neurons (neuron cores) per square centimeter 1010 synapses per square centimeter (memristors) About 100 milliwatts per square centimeter Total power consumption: 1 kilowatt Cortical-Scale Hardware System 10 000 neuromorphic chips 1010 neurons 1016 synapses Total power consumptioin: 1 kilowatt

CONCLUSION

The use of the memristor addresses the basic hardware challenges of neuromorphic computing: the need to simultaneously move and manipulate data, thereby drastically cutting power consumption and space. Neuromorphic computation means computation that can be divided up between hardware that processes like the body of a neuron and hardware that processes the way dendrites and axons do.