Friday 24 April 2015

neuromorphic chips -replacement for brain cells AND MANY MORE

                                   NEUROMORPHIC  CHIPS -  FUTURE OF  MANKIND

 Researchers have increasinglly made better intracranial implants of devices that are able to communicate with the brain. They can be used for a variety of brain disorders in order to restore either motor, sensory or cognitive functions.


this  chips  could replace  brain cells and could even  increase its  functioning  it could be the first step  towards the world  of  cyborg.






application of neuromorphic  chips :-











In recent years there is a growing interest in neuroscience and the prospect to understand the brain functionality and use the principles of neural computation to address the limitations of toady’s technology. The research community by recognizing this potentiality has lunched remarkable projects to support the field of computational neuroscience, which studies the information processing properties of the nervous system. A notable ongoing project is the Blue Brain Project at the Ecole Polytechnique Federale de Lausanne in Switzerland. This project uses detailed studies of the nervous system to simulate 10.000 neurons of the rat brain. Another project at IBM Almaden Research Center, in San Jose, California, focuses at the understanding of the outer processing layer of the brain, which is the cortex. They use neuron models to simulate 900 million neurons connected by 9 trillion synapses. Both of the above approaches have given great results, but rely on large-scale simulations of the studied models in High Performance Computing clusters (supercomputers) like an IBM BlueGene/ P. These simulations are quite slow, since the computers need many minutes to model seconds of worthy brain activity, given the large amount of studied parameters and the drawbacks of the von-Neumann computing architecture.

                       chip :-
http://blog.pcnews.ro/wp-content/photo/2010/12/iconchip.jpg

Very large scale integrated (VLSI) systems are the products of commercial electronics that have the greater performance, by placing hundreds of thousands of electronic components on a single circuitry. The size of VLSI systems is steadily increased in order to handle progressively more advanced computational tasks. This allows them to represent an attractive alternative to conventional numerical simulations, by offering high-speed execution of neural models in silicon. 

http://ncs.ethz.ch/projects/neurop/FCorradiMGiulioni2012.png/image_preview 
The delivered systems, by the neuromorphic engineering, can perform efficient emulation of large spiking neural networks, which is a process very demanding in energy and processing time in supercomputers. With the existing conventional technology the simulation of a model that represents the complete brain activity, would require a supercomputer 1000 times more powerful as the best one that are available today and will consume energy equal with one that a large city needs.
One approach that seeks to exploit the non-linear characteristics of the transistors and emulate the behaviour of the neural cells is the Neurogrid project, which runs at Stanford University, Prof. Kwabena Boahen. The aim is to build a large-scale system, capable to perform simulations that include enough cortical areas, in order to reveal the interactions among them. This system will be an alternative to supercomputers, which can demonstrate with affordable way how the cortex works. The project uses analog circuits, below their transistors’ threshold voltage. This approach slows down the operations down to a biological realistic rate, giving the opportunity of simulating a million cortical neurons in real-time. The analog computation provides the emulation of ion-channel activity and the digital communication the synaptic connections. The neurons and the synapses are full programmable and can model a variety of neuronal cells and interactions. The aim is the Neurogrid finally to simulate a million neurons connected by billions of synapses in real-time, by consuming many orders of magnitude less energy than the typical supercomputers.
Large-scale simulations of neural networks may be performed also by standard digital processors that are connected by dedicated pulse communication architecture. This is realized by the SpiNNaker project at the University of Manchester, Prof. Steve Furber. This project aims to model brain activity in real time and with high flexibility and programmability as a general purpose computer. The final system will consist of 57 600 custom-designed chips, each of which contains 18 low-power ARM9 processor cores. At the center of each chip a custom designed router is placed, which receives and forwards all the incoming packets from the cores to the neighbouring chips. At the top of each chip a sufficient size of synchronous dynamic RAM holds the connectivity information for up to 16 million synaptic connections. The main advantages of this proposed system are the high programmability of the overall digital system and the optimized distribution and routing of pulse events, with respect to conventional supercomputers.
As an alternative, highly integrated mixed-signal circuits could be used for emulating the individual neurons. This approach is employed by the BrainScaleS project, which is currently ongoing at University of Heidelberg, Prof. Karlheinz Meier. Custom-made analog circuits are developed to mimic the operation of the brain. This project replicates these custom-designed neuromorphic circuits over whole silicon wafers, interconnecting them with a high-density routing grid. Each wafer implements up to 200 000 analog neurons and 40 million learning synapses that are configurable. A sophisticated frame-based communication infrastructure for inter-wafer connectivity and control of the system has been developed from the University of Technology in Dresden (TUD), Prof. Rene Schüffny. The whole system is significantly more energy efficient than supercomputers and accelerates the computation by a factor of 10 000 compared to biological real time. The latter is an important feature as it shrinks the simulation time of an implemented neural network from hours to minutes or seconds. This is translated to significant power savings, given the large energy consumption of supercomputers and of course to a more affordable way to simulate and understand the brain. The final delivered system will be a flexible research platform, capable to study the dynamics of large-scale biologically inspired neural networks that the contemporary computational neuroscience investigates. The accelerated computation and the significant less power consumption (and therefore more affordable), with respect to conventional supercomputers are the main advantages of this large-scale neuromorphic system.
As we understand from all the above, there are three kind of approaches for the developing of large-scale, brain-inspired systems; analog, mixed-signal and digital. Each one has its own advantages and disadvantages as concerns important factors. With regard to thereconfigurability the digital domain is far more advance than the analog, with the mixed-signal solutions to be somewhere in the middle. With respect to power consumption of the overall system the analog solutions have clearly the advantage over the digital, with the mixed-signal to be also in the middle. As concerns the density of components for all the prementioned approaches it is more or less at the same level. Consequently, we can realize that the mixed-signal designs are a compromise between the analog and the digital domains, that’s why they are often preferred by the IC industry for commercial application.


QUALCOMM CHIPS - BETTERMENT   OF ROBOTS 

Qualcomm’s chips won’t become available until next year at the earliest; the company will spend 2014 signing up researchers to try out the technology. But if it delivers, the project—known as the Zeroth program—would be the first large-scale commercial platform for neuromorphic computing. That’s on top of promising efforts at universities and at corporate labs such as IBM Research and HRL Laboratories, which have each developed neuromorphic chips under a $100 million project for the Defense Advanced Research Projects Agency. Likewise, the Human Brain Project in Europe is spending roughly 100 million euros on neuromorphic projects, including efforts at Heidelberg University and the University of Manchester. Another group in Germany recently reported using a neuromorphic chip and software modeled on insects’ odor-processing systems to recognize plant species by their flowers.
Today’s computers all use the so-called von Neumann architecture, which shuttles data back and forth between a central processor and memory chips in linear sequences of calculations. That method is great for crunching numbers and executing precisely written programs, but not for processing images or sound and making sense of it all. 



how neuro chips work________________________
http://ej.iop.org/images/0957-4484/24/38/384009/Full/nano458941f1_online.jpg


             NEURONS BIOLOGY______________


http://www.nature.com/polopoly_fs/7.13495.1383568372!/image/Brain-chip.jpg_gen/derivatives/landscape_630/Brain-chip.jpg











                           VISION CHIPS -HOW IT WORKS ?

A vision chip is an integrated circuit having both image sensing circuitry and image processing circuitry on the same die. The image sensing circuitry may be implemented using charge-coupled devices, active pixel sensor circuits, or any other light sensing mechanism. The image processing circuitry may be implemented using analog, digital, or mixed signal (analog and digital) circuitry. One area of research is the use of neuromorphic engineering techniques to implement processing circuits inspired by biological neural systems. The output of a vision chip is generally a partially processed image or a high level information signal revealing something about the observed scene. Although there is no standard definition of a vision chip, the processing performed may comprise anything from processing individual pixel values to performing complex image processing functions and outputting a single value or yes/no signal based on the scene.
http://www.nature.com/scientificamerican/journal/v292/n5/images/scientificamerican0505-56-I4.jpg


 IF U HAVE SOMETHING TO SAY JUST TYPE IN COMMENT  BOX.


                             THANK UUUUUUUUUUU_____________