HUMAN and EARTH BRAIN
Growing advances in digital technic and computational neurobiology allowing to study to the brain for parts (reverse-engineer), are facilitating to understand how brain functions permitting the construction of 2 brain models to great scale that although run but slow that real brains, simulate big number of neurons and connections. At the moment the models lack sensitive inputs and motor ouputs and of important subcortical structures (cerebellum, amygdale, spine), looking like sleep rhythms. From one side: I) Henry Markram’s model (Blue Brain Project), offering a detailed neuronal reconstruction, complex systems of connections, complex range of gates and communications mechanisms of infinite detail. Of another side the model of II) Dharmendra Modha (IBM), creator of a simulated program of the brain of a cat. In his simplified model of sinapsis and interneuronal connections, the neurons alone have soma with nuclei and simplified spikes. The goal of Modha is to program in parallel, architectures to great scale to carry out simulations of great number of neurons and synapsis. Modha hopes to have supercomputers the 2019, looking like bebe brains or of psychotics. The pattern of Modha has created a conflict among the use of simplified models to run but quick with that of neuronal complex but detailed to understand its operation.
Complementarily (and, if the exponential growth of calculation power doesn't cease), it is hoped to build devices of common use, based on principles of neural computation, probabilístic (not deterministic), able to reason inductively (not deductively). The seminal paper of Simón Haykin (McMaster University): "Cognitive radio: Brain-empowered wireless communications", allowing the generation of wireless nets with devices that use neural computational principles menace to build a new infrastructure of the world, with sensors that will give us information on the use of the electricity, conditions of roads, patterns of weather and of dissemination of illnesses, optimizing goals, regulating the flow of resources, making to the world that surrounds us but useful and efficient.
Complementarily (and, if the exponential growth of calculation power doesn't cease), it is hoped to build devices of common use, based on principles of neural computation, probabilístic (not deterministic), able to reason inductively (not deductively). The seminal paper of Simón Haykin (McMaster University): "Cognitive radio: Brain-empowered wireless communications", allowing the generation of wireless nets with devices that use neural computational principles menace to build a new infrastructure of the world, with sensors that will give us information on the use of the electricity, conditions of roads, patterns of weather and of dissemination of illnesses, optimizing goals, regulating the flow of resources, making to the world that surrounds us but useful and efficient.
AVANCES EN COMPUTACION NEUROCEREBRAL
Crecientes avances en técnicas computacionales digitales y neurobiológicas, que permiten estudiar al cerebro por partes (reverse-engineer), están posibilitando entender como trabaja, conduciéndonos a expectar la construcción de 2 modelos cerebrales a gran escala que aunque corren mas lento que los cerebros reales, simulan gran número de neuronas y conexiones. De momento los modelos carecen de entradas sensitivas y salidas motoras y de importantes estructuras subcorticales (cerebelo, amígdala, medula espinal), semejando ritmos de sueño, funcionales. De un lado: I) El modelo de Henry Markram (Blue Brain Project), ofreciendo una detallada reconstrucción neuronal, sistemas complejos de conexiones, una gama compleja de puertas y mecanismos de comunicación de detalle infinito. De otro lado el modelo de II) Dharmendra Modha (IBM), creador de un programa simulado del cerebro de un gato. En su modelo simplificado de sinapsis y conexiones interneuronales, sus neuronas solo tienen soma con nucleos y espigas simplificadas. La meta de Modha es programar en paralelo, arquitecturas a gran escala para llevar a cabo simulaciones de gran numero de neuronas y sinapsis. Modha espera contar con supercomputadoras el 2019, semejando cerebros de bebes o de psicóticos. El modelo de Modha ha creado una pugna entre el uso de modelos simplificados para correr mas rapido con el de complejos neuronales mas detallados para comprender su funcionamiento.
Complementariamente (y, si el crecimiento exponencial del poder de computacion no cesa), se espera construir dispositivos de uso comun, basados en principios de computación neural, probabilísticos (no, deterministicos), capaces de razonar inductivamente (no, deductivamente). El seminal articulo de Simon Haykin (McMaster University): “Cognitive radio: Brain-empowered wireless communications”, esta permitiendo la generacion de redes inhalámbricas con dispositivos que usan principios neuronales computacionales conformando una nueva infraestructura del mundo, con sensores que nos darán información sobre el uso de la electricidad, condiciones de las vías, patrones del tiempo, patrones de diseminación de enfermedades, optimizando las metas, regulando el flujo de recursos, haciendo al mundo que nos rodea mas útil y eficiente.
Labels: Computational brain
0 Comments:
Post a Comment
<< Home