As a presenter I've been given a special discount code to share with my readers. Just quote MH0581 and get:

• Early bird registrations: normally $660, but $550 with the discount code.

• Regular registrations: normally $770, but $660 with the discount code

Last year's conference was great and this years one should be even better. Hope to see you there!

M

Holly Lisle has once again saved the day, but in a most unexpected way.

As the two or three of you who read my blog know, I've been struggling with my 2nd draft outline. I'm making progress, but I haven't completely untangled the mess my pantzer ways got me into.

I had recently purchased Holly Lisle's How to Beat Writer's Block audio course, not because I wanted the audio course, but because she's selling it as a bundle with a bunch of her "clinic" e-books, which I did want. I'm a sucker for writing craft books and software toys. They're fun to read, inspirational, and if I find one useful tip or technique, it's worth it. So I figured the audio course would probably never get listened to. I almost never have trouble coming up with ideas for writing. So why should I listen to an audio course on how to beat writer's block?

But in a way, I was having writer's block with this 2nd draft outline. Not in creating ideas, but in fixing ideas I had already written. So I thought I'd give Holly's course a try.

I don't want to divulge the contents of her course, so let me just say it led me to this very revealing question:

When did writing this novel stop being fun?

Through Holly's technique, the answer blurted out of my fingers onto the keyboard without giving my conscious mind time to construct an answer. I don't want to discuss specific details of my novel, so I'll phrase my answer this way: My novel stopped being fun when Event X happened. This was one of those answers that I already knew inside at some level, but I wasn't taking it seriously. I didn't appreciate the implications of what had happened.

Event X happened to my protagonist. I had planned Event X since I started writing the first draft. My protagonist took Event X in stride, took it very well, and went on with her life as best she could.

My line of questions went something like this:

• When did writing this novel stop being fun? When Event X happened.
• How do I fix this? The easy way is to take out Event X.
• If I leave Event X in, why did it take the fun away?

And then it hit me. The answer to that final question: Because my protagonist didn't react the way I had written.

My Depressed Protagonist

My protagonist was angry. Angry at the person who did Event X to her, angry at life for Event X happening to her, and angry with me for writing her reaction without consulting her first. I didn't let her express any of this anger, and she became depressed. The life had gotten sucked out of my book.

I didn't realize it at the time, but I had abandoned my muse at that point. Writing the novel stopped being fun, and it suddenly became hard work.

My protagonist's reaction to a major event in her life was not truthful.

"Fiction is a lie, and good fiction is the truth inside the lie." -Stephen King

Stephen King has often emphasized telling the truth when writing. I failed to do so because I already had this plot idea in my head and I sailed through writing the passage without my muse by my side. So my muse walked away in a huff.

Thanks to Holly Lisle, I discovered my muse was missing and trapped, and I freed her, promising never to let this happen again.

(Footnote: In searching for photos for this post, I learned a valuable lesson. Google for "woman in cage" at your own risk.)

El enfoque conectista de la IA utiliza un modelo de neurona artificial (el más conocido es el modelo de McCulloch y Pitts. Una neurona artificial imita el comportamiento de una neurona biológica. En un principio se pensaba (y todavía se piensa) que si se modelo adecuadamente una neurona bastaría con añadir un montón de neuronas para tener un "cerebro artificial".

Parece ser que la "arquitectura del cerebro" también cumple un papel importante, además de otros muchos procesos que se están estudiando en la actualidad, por lo que poner un montón de nueronas juntas no basta para hacer un cerebro pero es un principio. En realidad el cerebro y su funcionamiento siguen siendo a fecha de hoy una gran incognita y por lo tanto un gran campo de investigación.

A lo que iba, que me enrollo. Supongamos que desarrollamos un sistema digital con un "cerebro artificial". En un momento dado podríamos intercambiar el "cerebro artificial" por un cerebro biológico. Quizás lo más interesante sería el paso contrario, es decir, tener un cerebro biológico e intercambiarlo por uno digital overclockeado. Cuestiones morales a parte, la idea mola, jeje.

Y ahora retomamos la noticia. Le han puesto un chorro de neuronas de rata a un robot y es capaz de esquivar paredes. La verdad es que no parece mucho pero por algo se empieza. Como bien comenta el científico Kevin Warwick "Ahora hay que enseñarle a hacer cosas". Al fin y al cabo podríamos decir que el robot acaba de 'nacer' y tiene un cerebro sin 'programar'. Tendrá que ir aprendiendo cosas mediante estímulos y lo que él mismo considere positivo o negativo (me gustaría saber como han implementado esto).

A mi es algo que me parece muy curioso y siempre me ha apasionado. ¿A alguien más?

Actualización (19-08-2008): He encontrado un video sobre el robot.

The application form asked for my Title. Options were:

• Miss
• Dr
• Mr
• Mrs
• Ms
• Prof
• Other

Naturally, I selected "Other".  But there was then no-where to enter my actual title, my "other" title.

So last week I received a letter addressed to "Other Gareth Saunders".

While I'm here

Who on earth ordered that list?!  How about the male-centric ordering of:

• Mr
• Mrs
• Miss
• Ms
• Dr
• Prof
• Other

Or the ladies-go-first option of:

• Ms
• Miss
• Mrs
• Mr
• Dr
• Prof
• Other

Or even the mostly-alphabetical but turns out to be partly hierarchical (by placing Other at the end):

• Dr
• Miss
• Mr
• Mrs
• Ms
• Prof
• Other

Aparentemente havia uma peça dentro da própria peça. Os atores eram atores duplos, pois atuavam uma atuação. Ao desenrolar da situação, entretanto, muitas vezes eles saíam dessa encenação da encenação e havia uma história de fundo. Se alguém me perguntar que parte é qual, eu não vou saber dizer. Muito confuso para mim. Apesar de tudo, a história era ótima. Basicamente no contexto da Revolução Francesa, o agente principal que desencadeava tudo era o tal do Marat. Não sei, contudo, em qual peça o Marat estava: se na encenação dos "atores reais", ou na encenação dos "atores virtuais" - arrisco essa definição para conseguir diferenciar as duas camadas.

Mas fato é que havia uma forte discussão acerca da ideologia empregada na época da Revolução Francesa. Algo do tipo "será que os fins justificam os meios?", além de discussões acerca de um senso de justiça e direito da população camponesa. É uma problematização bem comum em revoluções em geral, passar ou não por cima de tudo que se acredita em prol de um bem que poderá vir.

De uma maneira geral, esse negócio de peças é muito louco. Do nada as pessoas começam a cantar, ou a se atirar de um lado para o outro. Teve até uma hora que um dos atores atirou um copo de vidro no chão durante uma correria no palco e detalhe: estavam todos de pés descalços. Gurizada tão fissurada no enredo que começa a extrapolar. Outro ponto alto da peça é, por unanimidade, quando uma das atrizes mostra os peitos (numa peça que envolve o Marquês de Sade isso não podia faltar). Parece que ela mostrou mais de uma vez, infelizmente só vi a primeira. O Pedro disse que chegou a esquecer a história e ficou só esperando esse momento derradeiro. E valeu a pena.

Assortment of apples in the orchard

In the gift shop they have a variety of books I found interesting and--most irresistible--hundreds of seed packets!  I have managed so far to resist buying seeds for myself but do have a full color glossy catalog that I will read cover to cover and hopefully I can have a garden next summer, in which I will go completely overboard and plant too many things.  Maybe I will plant Queen Anne's Pocket Melon, which Victorian ladies used to carry in their pockets for their sweet perfume.  Their flesh is bland, though.  And maybe I will have Chet's Italian Red Garlic, which was found growing in an abandoned garden near Tonasket, WA.  And Blue Jade corn, which I saw growing at the farm.  The plants are only 2-3' tall and 3 to 6 ears grow near the ground.  It is one of the only corn varieties that can be grown in a container.

Herb and flower garden

The heritage farm is more than just test plots and gift shop, too.  They have a couple hiking trails and they are keeping a herd of Ancient White Park Cattle, a rare breed from the British Isles. 

What a great visit!  Afterward we walked around downtown Decorah (a town where people still sit on their front porches) for a little while and I went to the food co-op hoping to find avocadoes and cilantro.  I have a taste for tacos but no luck finding those ingredients.  I did find chipotle hummus and really delicious green beans.  Now I am settling in to read my seed catalog, though I wonder if I should save it for winter.

O cérebro biológico do robô, batizado Gordon, foi gerado a partir de neurônios extraídos de um rato. Os tecidos foram colocados numa solução, separados e depois colocados em uma espécie de leito com 60 eletrodos.

"Em 24 horas, as conexões se reforçaram, formando uma rede como num cérebro normal", explicou o responsável da equipe, Kevin Warwick. "Em uma semana ocorreram impulsos elétricos espontâneos e o que parecia ser uma atividade de cérebro comum."

"Utilizamos esta reação para vincular o cérebro ao robô com os eletrodos. Agora o cérebro controla o robô e ele aprende por repetição", disse o cientista.

Muito louco não???

• Fonte: Folha Online
• http://www1.folha.uol.com.br/folha/ciencia/ult306u433530.shtml

I've been thinking that traveling has been more comfortable lately than I remember it being in the beginning, partially I assume because I have gotten used to it, but also because we have made some improvements to our system. 

Crucial improvement #1: the capability of running the fridge on propane.  I've mentioned it before, and not only does it allow us to keep groceries more reliably, we have a better selection of campsites.  Frequently, the sites with no hookups are more attractive to us and now we can indulge in them. 

#2: ant-proof dog food container.  We picked this up in Livermore and I had wanted one for a while.  It has a screw top and I think my roommates will like it, too, since the top doesn't come off its hinges and fall off anymore.  I left the dog food outside in Mississippi and we had an ant infestation.  Fortunately, there wasn't much food left.  And they didn't appear to be biting ants.

#3: 6-gallon water container.  First we had the kind of couple-gallon disposable plastic container you buy water in from the grocery store that we reused all the way into California.  Then I found a bug in it that resembled a worm and I couldn't drink out of it anymore.  Next we had a 1-gallon Coleman container I bought at a yard sale in Livermore for $1, but it leaked even after my attempts to seal it with a sealant most likely not recommended for food containers.  Last week Derek called my attention to a couple items at his neighbors' yard sale in Leadville, one of which was this great aquatainer which is watertight and has a convenient spout.  (I also got a food dehydrator for $5!)  I should have paid more attention to yard sales in a place like Leadville where the people tend to be outdoorsy and do-it-yourselfers.

#4: The Stroller.  Obviously, this allows us to go some places when we might otherwise be hampered by Angus' lethargy.  He's been doing pretty well lately but it's nice to have, especially when you have access to a smooth bike path.

#5: New camper tires.  We had a lot of flat tires in the beginning, but I think since New Mexico the camper has been on new tires, and by 'new' I believe I mean tires that did not come with the camper in 1998.  We actually had a couple more tire changes recently: one was bald and the other had a variable but constant leak.  I hope to make it back to VA on these ones. 

I'm off to see if I can get a run in before Angus starts barking about dinner.  This is a picture of a calf we saw at the Picket Fence Creamery in Woodward, IA.  I wanted a better shot, but Lucy was crying and the yellow lab was barking and my groceries, including a tub of peanut butter, were in the car with Angus so I was rushed.

is what Iowa looks like, you would only be partially correct.  Yes, unfathomable amounts of corn grow here, but that doesn't make it as dull to drive through as you might think.  After we drove through Omaha and crossed into Iowa, suddenly the terrain became hilly and sometimes terraced.  We passed scores of farms and I was enthralled. 

A view in southwestern Iowa

Lucy was not, however.  I'm not sure what happened to her, but since leaving Leadville every time we get in the car she starts panting very fast, as if she's nervous.  It hasn't been hot, so I know it's not that.  I wonder if it is related to the diarrhea incident...oh, did I fail to mention that?  En route to Flaming Gorge NRA with Heather from Yellowstone, Angus snagged the front seat and Lucy stayed close to the front of the car with us.  It was warm, so I assumed she was panting because she was hot.  Usually if she needs to get out of the car she paws my shoulder and she never did.  Just before arriving at the campground I smelled it, and congratulated Angus since I assumed it was his usual gas.  Then I realized that Lucy had diarrhea in the back.  Heather patiently tolerated my little temper tantrum (she's good at that) as we were choosing a site and soon we were drinking beers and joking about whether the pop-up would blow over in the wind if we zipped up the windows to keep out the rain.  Anyway, Lucy doesn't seem a happy traveler in the car lately but she hasn't yet refused to get in. 

We are staying in Story City, which is 13 miles north of Ames.  This is the first time since leaving Leadville that we are staying two nights in one spot, which feels great.  Plus, we have an excellent site.  Our three main criteria for choosing a site are these 1) shade, 2) grass, 3) proximity to neighbors.  We are camping in the tent area of a beautiful private campground across from, yes, a cornfield.  Our site is completely grassy, shady except for the early morning when it doesn't matter, and our nearest neighbor is 40 meters away.  Behind us stretches a large and half shady park with a couple trails for walking, and if we go sit near the office we have wireless internet access.  Also, we share the park with this cool tractor and have a sweet-smelling port-a-potty close by which only I appear to be using, all for $16.05 per night.  Oh yeah, and the showers are free and hot.  I feel so pampered.

So here I am, working to finish my second draft outline, when I get a great idea for a completely different novel. I'm doing my best to focus focus focus on finishing my first novel, but I can't help but brainstorm ideas for the second one. It's a definite juggling act not to lose momentum on novel one and get carried away on novel two.

I wonder what other authors do when they are working on more than one book at a time?

Now, I’m new to the world of HE but I think it’s fair to say that brand and design don’t necessarily sit comfortably or naturally with many university web teams. You only have to look at most university websites - particularly those of the “old” universities (sorry, my language in this space is still naïve, but I basically mean “the universities that didn’t used to be polytechnics”…) to see - well, a lack of design. These sites aren't incompetent, visually: they often look ok, but at the same time there is often a missing coherence to the visual design; a lack of deeper thought about what the site represents and why.

There are many possible reasons for this - first off, it can be expensive - inevitably it involves agencies and not just "someone in-house who can use Photoshop"; second, it requires a series of skills that are specialist and often not available to these institutions; and thirdly, it just isn’t (or hasn’t been) seen as a priority. And in a world where sites have grown organically, without any strategic overview (much like many large museum sites), it’s hardly a surprise that the end result often looks like an explosion in a paint and typography factory.

Here's the thing - personally, I think it was a mistake to have Precedent doing the talk at IWMW. This isn't - contrary to what would be an easy assumption - because "I hate that London designer type". Far from it. I've worked with agencies like this day in, day out for (scarily) a decade, and see the extraordinary work these approaches can have in supporting and extending "the brand". I’ve been the budget holder for projects where we absolutely reaped the benefits of getting the brand right. Where I once squirmed at handing over tens of thousands of pounds for “a logo I could have knocked up in Photoshop in an hour”, I now see that the work surrounding brand development can be absolutely fundamental to aligning the various voices held by institutions. I have - I guess - grown up, and now find myself fighting the battle on the side of the design and brand agencies; or, to put it another way, against mediocre or non-existent visual design.

The fact is, though, you couldn't have scripted the opposites at IWMW more perfectly if you'd tried. One the one hand, a media type with his SAFE matrix; on the other, a room full of techies with SQL, Java and repositories higher up in their mental wish-lists than all that high-level brand stuff. Which is not as insulting as it sounds: these guys are busy people with the daily requirements of huge, political and sprawling organisations in their in-boxes. Why should they care too much about what the thing looks like?

The reason I think Precedent were the wrong choice for IWMW is that what they do is so far in one direction - they are at a distance that is worlds apart from where most HEI’s and other institutions are today; it assumes an acceptance of the need for brand; it assumes a considerable budget; it assumes that anyone actually cares about what brand and visual design can do for web users and the coherence of an online offering. Out there in the commercial web world, these things can - sometimes - be taken as read; in here, with museums and HEI’s, many sites haven’t even got coherent colours, style guidelines or a consistent tone of voice sorted yet, let alone a "brand" or online strategy.

The net result, of course - through no fault of the speakers, but more the context - was that the already-held misconceptions about brand agencies (expensive, marauding, slick) were reinforced rather than broken down. No-one went away thinking “great, a SAFE matrix: I’ll take this back and implement it”. Instead the audience sniggered a bit and then went down the pub to laugh at iSoton for spending money on a text based logo that - yes, you guessed it - “…I could have knocked up in Photoshop in an hour”.

What would have worked better, I think, is an online marketing agency talking about why visual design and coherence is important; someone with some understanding of the barriers that museums and HEI’s face when trying to get to this coherence; what it means to have to battle with differing internal pressures about what the website is for and how brand work can help win these battles.

Even amongst web people, there is a huge divide - astonishingly huge, IMO - between those who think visual design is important, and those who don’t. Go to any web conference like FOWA and you’ll see beautifully crafted visual design on everything you come across; go to any repository, VLE or research site and you’ll get Times New Roman-esque nothingness, or worse, a level of extremely nasty amateurism. Just take one look at the Moodle logo on the left, for example. I mean, that hurts, right? That's a classic "hey, the bevel filter. Cool! Where can I use it...?" moment. That isn't design...it's...horrible..

The question - either to the HEI's, museums, public bodies or government sites that are lacking in visual appeal is - well, hey, does it actually matter? I mean - Moodle works, right?

My personal take is - yes - it matters a great, great deal. What Precedent are saying I think is that visual design and brand are absolutely vital in communicating trust, and increasingly this will be the case as time goes by. The trust landscape is the place that we will increasingly be inhabiting as the web evolves. The very fact that the social web is a leveller, a democratiser, makes it even more vital that institutions have coherence in brand, voice and visual appearance.

Of course, the debate (as was highlighted by a recent twitter moment about the new delicious re-design) is only made more confusing when you bring in non-visual design: how the IA helps or hinders a user; whether (as I strongly believe), “accessibility” is about brand, look and feel, identity and not just about whether the thing “works”. That, as they say, is a whole different story…

One thing is clear: the bar is continually being raised, and with it the expectation of users. All institutions, whether HEI, museum, public body or otherwise, are continually being compared to commercial websites. We should therefore learn to listen to people like brand and marketing agencies. They have something useful to say - whether you choose to engage with a “SAFE matrix” is up to you...

Esto mismo podría ocurrir en todos los juegos de información perfecta, lo que ocurre es que el espacio de estado es tan grande que no es manejable, ni para el humano ni para la maquina (aunque el humano juega con estrategias diferentes a las utilizadas por los ordenadores).

Lo que hace que el juego del Go sea tan complicado para la IA es encontrar una función de evaluación para una jugada. La explosión combinatoria que aparece como resultado de cada movimiento es completamente inmanejable para un ordenador. En cambio, los humanos al utilizar estrategias distintas, basadas en reconocimiento de patrones parecidos (no es necesario evaluar la situación sino simplemente saber que en el pasado una situación parecida tuvo buen o mal fin) podemos afrontar la explosión combinatoria con un enfoque distinto, más simplificado.

Bueno, pues por primera vez en la historia un programa de IA gana a un jugador profesional de Go. Quizás no haya sido tan sonado como cuando Deep Blue ganó a Kasparov pero se trata de un hito igual de importante.

Por último os dejo un enlace para que el quiera pueda jugar al Go online.

Este juego ha sido considerado durante años como un reto extremadamente difícil para la Inteligencia Artificial, situándose a un nivel de complejidad muy por encima del Ajedrez. El 7 de Agosto de 2008, el programa Mogo, que funciona sobre una supercomputadora, fue capaz de batir al jugador de Go profesional Myungwan Kim, en un tablero de 19x19, noticia que se ha convertido en un logro histórico para la Inteligencia Artificial.

Si bien ya hace años que las computadoras pueden jugar al Ajedrez al nivel de los profesionales, el Go se ha mostrado como un reto más duro y complejo debido a varias diferencias claves con respecto al Ajedrez:

• El tablero es demasiado largo (19x19, con 361 intersecciones), lo cuál genera una explosión combinatoria imposible de manejar, ni con la mejor de las heurísticas. De hecho, en tableros simples de 5x5, ya se consiguió resolver este juego en el año 2002.
• Las piezas tienen más grados de libertad que las del Ajedrez, ya que los movimientos del Go no se limitan por las regla del juego.
• Si bien en el Ajedrez, a medida que avanza la partida se van eliminando piezas y, con ello, cortándose la explosión combinatoria de movimientos, en el Go se van añadiendo posibilidades según avanza la partida.
• Las técnicas utilizadas en el Ajedrez computacional no son transladables al Go. Como ejemplo, si bien en el Ajedrez se puede utilizar un simple conteo de las piezas como un punto de apoyo en las exploraciones de su árbol de búsqueda, en el Go un simple conteo no sirve de nada, ya que las posiciones dependen más de los grupos que hay formados, de las conexiones entre piezas, etc.

Toda esta complejidad, ha permitido evolucionar un campo dentro de la Inteligencia Artificial, el denominado "Go computacional", que se encarga de desarrollar técnicas y programas capaces de jugar al Go como si fueran expertos humanos.

Si bien hasta la fecha ningún programa había sido capaz de derrotar a ningún jugador de Go profesional en un tablero estándar (19x19), el pasado 7 de Agosto, el programa MoGo (que funciona sobre una máquina con 800 procesadores a 4.7Ghz, con una potencia total de 15 Teraflops) derrotó a Myungwan Kim, quién dijo del programa: "ha jugado realmente bien, con una potencia de entre 2 y 3 dan, aunque ha realizado movimientos propios de un jugador 5 dan".

Enlace: Fuente

Para empezar crearé 3 clases: una para los nodos (GraphNode), otra para los ejes que conectan los distintos nodos generando la navegación entre ellos (GraphEdge) y, para terminar, la clase que representa el grafo de navegación (NavGraph).

Inicialmente (seguramente se deba añadir o quitar algunos atributos o métodos, con el tiempo iré mejorándolo) voy a crear estas clases:

A partir de ahora comentaré más o menos como tengo pensado utilizar las distintas clases y como las iré relacionando con los otros elementos que compondrán el engine de IA a crear para el proyecto Cragpet.

I think I've finally freed myself from my second draft outline quicksand. I had a number of questionable motivations that triggered key plot events, but I finally (I think!) untangled the mess and am ready to move on.

I give credit to Jocelynn Drake and her recently released debut novel, Nightwalker. I don't want to talk too much about this book, as I'll be discussing it soon in an "Author Spotlight" post. Except this: I loved it. The atmosphere she creates is why I love vampires in fiction. It definitely awoke my vampire muse and helped me get moving again.

"A chain is as strong as its weakest link."

Come to think about it, I was quite fortunate to be able to attend this safety supervisor course. This 6-days course cost $325.00 and I got it for free since Siemens is paying for it... hahaha... I bet NTUC must have earned loads from it...
Well, another quote from the trainers:

"Learn from the PAST,
Apply to the FUTURE"

Las redes neuronales pretenden imitar el funcionamiento del conjunto de neuronas que funcionan y forman el cerebro humano. Alan Turing fue el primero en estudiar el cerebro como una forma de ver el mundo de la computación. Quienes han realizado estudios, investigaciones y ha podido modelar una red neuronal y, se acredita la creación de redes neuronales para otros fines que no sea solo el biológico, son Warren McCulloch y Walter Pitts. Estos dos investigadores propusieron un modelo matemático de neurona. La meta principal era obtener información referente al funcionamiento y comportamiento del cerebro. El modelo consistía en proporcionar a cada neurona de un conjunto de datos de entrada y salida, para activar la neurona era necesario calcular la suma de los productos de cada una de las entradas y, la salida en cambio, es una función, resultado de la activación. El recurso principal de este modelo son los pesos que se proporciona a cada entrada, es decir, los pesos de las conexiones entre neuronas. Si los pesos llegan a variar, el resultado de las salidas se modificará. Por lo tanto, el valor de los pesos de cada entrada es directamente proporcional al resultado de las salidas del modelo. A finales del siglo XIX se logró una mayor claridad sobre el trabajo del cerebro debido a las investigaciones de Ramón y Cajal en España y Sherrington en Inglaterra. El primero trabajó en la anatomía de las neuronas y el segundo en los puntos de conexión de las mismas o sinapsis.

Aproximadamente en el cerebro hay 50.000 neuronas por cada milímetro del cerebro, y billones de neuronas en todo el sistema neuronal. El tamaño y forma de las neuronas varía, lo que no varía son las partes de las cuales se compone, soma de la neurona, dendritas y axón. Se puede decir que la creación de redes neuronales y sus diferentes interpretaciones en cuanto a modelos, se puede aplicar en el campo informático y que ha servido en la mejora de procesos de sistemas computacionales complejos, como se comportan y la experiencia que adquieren estos en base a la información recibida y procesada. En el campo de la medina, ha servido para obtener información del comportamiento del cerebro y mejorar los diagnósticos de las diferentes anomalías que se generan en él.

El desarrollo de la tecnología y la necesidad de construir sistemas computacionales sumamente complejos, ha implicado el desarrollo de potentes herramientas y productos tanto hardware como software, que permitan simular, en este caso, redes neuronales para determinar u obtener una solución a determinado problema. En lo que se refiere a inteligencia artificial se refiere a redes de neuronas o redes neuronales, estas son una representación de aprendizaje y procesamiento automático. Se trata de un conjunto o sistema de interconexión de neuronas en una red que en la que interactúan para producir información de salida, previo un estímulo o datos de entrada.

Las redes neuronales están compuestas por un gran número de neuronas, las cuales a través de sus interconexiones generan un nivel muy alto de procesamiento, estas al momento de trabajar, proporcionan soluciones a problemas muy complejos. Las redes neuronales, aprenden con la experiencia. El aprendizaje automático es inevitable en las redes, esto es una ventaja, ya que les permite adquirir la experiencia necesaria para dar soluciones eficaces y eficientes, sin la necesidad de proporcionarle datos o información a la red.

Prácticamente se puede decir que, es una nueva forma de supercomputación con capacidad de manejar los estímulos e incertidumbres que aparecen cuando se desea resolver cualquier tipo de problemas asociados al mundo real, de esta manera las redes neuronales ofrecen soluciones completas, sin inconsistencias y fáciles de implementar.

Las redes neuronales se componen de muchos elementos, estos operan de forma paralela. El diseño no es único, este se determina por las interconexiones que tienes los elementos. Las principales ventajas de las redes neuronales son:

• Aprendizaje adaptativo.
• Auto-organización.
• Tolerancia a fallos.
• Operación en tiempo real.
• Facilidad para interactuar con las nuevas tecnologías.

Aplicabilidad

Las redes neuronales pueden utilizarse en diferentes áreas y aplicaciones. La construcción de las redes neuronales puede tardar de acuerdo al problema para el cual se quiere construir la red y a las funcionalidades y tareas que se quiere cumplir. Cuando se implementa mediante hardware, es decir, a través de un chip, existe un nivel muy alto de tolerancia a fallos del sistema, lo que permite la inserción de otras redes neuronales de bajo costo y funcionalidades no completas en los sistemas desarrollados.

Existen distintos tipos de redes neuronales, cada uno de estos tiene aplicabilidades diferentes y distintas áreas como:

Electrónica y Sistemas Computacionales:

• Reconocimiento de patrones de voz, imágenes, señales.
• Algoritmos genéticos para crear controladores para robots.

Biología:

• Aprender más acerca del cerebro y otros sistemas.
• Obtención de modelos de la retina.

Empresa:

• Evaluación de probabilidad de formaciones geológicas y petrolíferas.
• Identificación de candidatos para posiciones específicas.
• Explotación de bases de datos.
• Optimización de plazas y horarios en líneas de vuelo.
• Optimización del flujo del tránsito controlando convenientemente la temporización de los semáforos.
• Reconocimiento de caracteres escritos.
• Modelado de sistemas para automatización y control.

Medio ambiente:

• Analizar tendencias y patrones.
• Previsión del tiempo.

Finanzas:

• Previsión de la evolución de los precios.
• Valoración del riesgo de los créditos.
• Identificación de falsificaciones.
• Interpretación de firmas.

Bibliografía:

• Historia de las Redes Neuronales. Disponible en:http://ingenieria.udea.edu.co/investigacion/mecatronica/mectronics/redes.htm
• TREC INTERNET. Redes Neuronales Artificiales. Disponible en: http://electronica.com.mx/neural
• Introducción a las Redes Neuronales Artificiales. Disponible en: http://www.redes-neuronales.tk
• MATICH, Damián. Redes Neuronales: Conceptos básicos y aplicaciones. Universidad Tecnológica Nacional – Departamento de Ingeniería Química. Rosario-Argentina. Documento PDF.
• Red Neuronal Artificial. Disponible en: http://es.wikipedia.org/w/index.php?title=Red_neuronal_artificial
• Neural Networks Framework. Disponible en: http://www.redes-neuronales.netfirms.com/tutorial-redes-neuronales/que-son-las-redes-neuronales.htm
• ARAUJO SIERRA, Basilio. Aprendizaje Automático: Conceptos básicos y avanzados. Editorial Pearson. Madrid-España. 2006.
• Red Científica – Ciencia, Tecnología y Pensamiento. Redes Neuronales. Disponible en: http://www.redcientifica.com/doc/doc199903310003.html
• Tipos de Redes Neuronales. Clasificación de las Redes Neuronales. Disponible en: http://softwarelibre.unsa.edu.ar/docs/descarga/2003/curso/htmls/redes_neuronales/c81.html
• Redes Neuronales. Funcionamiento del Perceptrón. Disponible en: http://www.wikipedia.es/enciclopedia/Redes_neuronales_artificiales
• Redes Neuronales aplicadas al avalúo inmobiliario. Disponible en: http://www.ilustrados.com/publicaciones/EpyVZEuElVVFXlQkSb.php

]]>