[phpBB Debug] PHP Notice: in file /viewtopic.php on line 920: date(): It is not safe to rely on the system's timezone settings. You are *required* to use the date.timezone setting or the date_default_timezone_set() function. In case you used any of those methods and you are still getting this warning, you most likely misspelled the timezone identifier. We selected the timezone 'UTC' for now, but please set date.timezone to select your timezone.
[phpBB Debug] PHP Notice: in file /viewtopic.php on line 920: getdate(): It is not safe to rely on the system's timezone settings. You are *required* to use the date.timezone setting or the date_default_timezone_set() function. In case you used any of those methods and you are still getting this warning, you most likely misspelled the timezone identifier. We selected the timezone 'UTC' for now, but please set date.timezone to select your timezone.
[phpBB Debug] PHP Notice: in file /includes/functions.php on line 3534: Cannot modify header information - headers already sent by (output started at /includes/functions.php:2956)
[phpBB Debug] PHP Notice: in file /includes/functions.php on line 3536: Cannot modify header information - headers already sent by (output started at /includes/functions.php:2956)
[phpBB Debug] PHP Notice: in file /includes/functions.php on line 3537: Cannot modify header information - headers already sent by (output started at /includes/functions.php:2956)
[phpBB Debug] PHP Notice: in file /includes/functions.php on line 3538: Cannot modify header information - headers already sent by (output started at /includes/functions.php:2956)
www.portierramaryaire.com • Ver Tema - Inteligencia artificial, redes neuronales, machine learning

Inteligencia artificial, redes neuronales, machine learning

Foro dedicado a otros temas sobre defensa sin cabida en el resto de foros

Moderadores: Lepanto, poliorcetes, Edu, Orel

Inteligencia artificial, redes neuronales, machine learning

Notapor poliorcetes el Dom Sep 08, 2019 10:22 am

Abro este hilo para cubrir por separado una parte de la revolución de asuntos militares que se nos echa encima. No se nos echa como el terminator de la peli, ni siquiera como la IA fuerte y autoconsciente de la que advierten los attention whores como Elon Musk. Se nos echa encima con utilidades menos humanizadas, quizás menos glamourosas, pero con un potencial disruptivo ilimitado... y no puramente militar, sino al alcance de actores no estatales.

Comienzo con un hilo abierto en otro foro de defensa

https://www.defencetalk.com/military/fo ... rks.14894/

Cheap visual processors & neural networks

I've been playing with the $100 Intel Neural Compute stick 2, a USB sized trainable neural network and visual processing unit - it enables all sorts of real time recognition of visual data such as facial recognition, number plate recognition etc It runs OpenCV and OpenVino and comes with a lot of samples, including pretrained models. It's not for the feint hearted and you need some experience with C++ to use it fully. You can deploy the trained models on a $50 Raspberry Pi, though training needs to done on more powerful Linux box.

We have set up a an example which recognises and reads number plates, allegedly up to 13 fps for video data, though we have only used still images. It's mind blowing, we are getting 100% correct reading of number plates. You can read about this stuff all day but until you see it in action I don't think you can fully grasp the implications. In robotics the 'perception problem' has been a real stopper for decades and this sort of cheap, OTS technology is going to be an incredible boost.

There would be many uses for this in the military. A simple use, you could put this very technology in a head of a missile after training it to recognise targets, such as warships, and get accurate target recognition. The Raspberry Pi has worked in space, so I don't see too many issue in a missile.

Understanding how an opponents 'see' you and how they process that becomes essential for defensive measures. With LO aircraft they already fly them with the lenses to increase their radar returns, this sort of strategy may have to be applied to ships to change their appearance in MW, visual and infrared appearance between peacetime and conflict. Using a decoy, like the nulka is a simple example of it but I wonder if it is going to need to be more profound.

It is also the technology that will most enable the 'killer robot' of science fiction, which was most hamstrung by the perceptual problem, now looking like it's going to be solved in the near future. Certainly any autonomous robotic system could benefit from this < $200 worth of hardware, if trained properly.


Fijaos en los términos:

* menos de 200$ australianos
* Raspberry pi
* Desarrollador sin apoyo de una megacorporación. De hecho, aparentemente con un equipo mínimo
* Software libre / de fuentes abiertas

El usuario habla de matrículas, pero las implicaciones son enormes. Sin megacorporaciones y cientos de millones de dólares, parece perfectamente factible emplear estos equipos ridículamente potentes para entrenarlos en que reconozcan otros inputs visuales.

Por ejemplo, un vehículo

Las implicaciones son sencillamente bestiales. Por un precio anecdótico, un desarrollador comparativamente modesto puede generar una cabeza buscadora para un dron. Por las mismas razones, no sería infinitamente caro generar una solución que entrenara para reconocer drones, como ya adelantó un socio mío el año pasado.

Precios. software libre. Conocimiento disponible

Components off-the-shelves

Todos estos componentes nos llevan a la distribución: muchos pueden generar soluciones capaces por muy poco dinero e invirtiendo relativamente poco dinero. Yo puedo hablar en primera persona de algo parecido. Empezando absolutamente de cero, leyendo y viendo videos, por menos de 100 euros he construido un dron de carreras y lo he programado con software libre. De cero, insisto.

Lo peor, en mi opinión: que mientras este tsunami de innovación se nos echa encima, Europa se empeña en seguir el camino de desarrollo de los 80. En otros campos Turquía, una nación con capacidades varios órdenes de magnitud inferiores a las del conjunto de Europa nos da un baño cada semana, con láseres, munición guiada, drones de ataque y un largo etc. Pero ni siquiera es un ejemplo completamente válido: actores no estatales van a poder producir un amplio abanico de soluciones de C3, defensa y ataque con unas capacidades que ahora mismo no están a la disposición de la todopoderosa OTAN.

En fin, creo que es importante mantener un hilo sobre IA, redes neuronales, machine learning y un largo etc. Obviando a hollywood o a los alarmistas, hay un campo enorme y revolucionario
Nunca digas que éste es mi último sendero
**
podcast de portierramaryaire
https://www.ivoox.com/podcast-portierra ... 223_1.html

y recuerda nuestro patreon para actualizar el foro y crecer
https://www.patreon.com/portierramaryaire
Avatar de Usuario
poliorcetes
Site Admin
 
Mensajes: 13139
Registrado: Vie Abr 17, 2009 11:54 am

Re: Inteligencia artificial, redes neuronales, machine learning

Notapor Orel el Dom Sep 08, 2019 12:50 pm

Ya sabéis que hasta casi octubre estaré bastante desconectado, pero espero que Pyro se anime a comentar por aquí algunas cosas :wink:
Avatar de Usuario
Orel
Moderador
 
Mensajes: 46184
Registrado: Sab Sep 24, 2005 11:33 am
Ubicación: España, en el bocho

Re: Inteligencia artificial, redes neuronales, machine learning

Notapor champi el Vie Ago 14, 2020 2:49 pm

Contrato de EEUU para el centro donde se impulsará el desarrollo de IA (12/8/2020): https://www.defense.gov/Newsroom/Contra ... e/2310984/
...
JOINT ARTIFICIAL INTELLIGENCE CENTER

Deloitte Consulting LLC, Arlington, Virginia, was awarded a Systems Engineering, Technology, and Innovation prime integrator task order (HC102820F2000) for an estimated $106,352,518 to design and build the Joint Common Foundation artificial intelligence development environment for the Department of Defense Joint Artificial Intelligence Center. The period of performance is a one-year base period from Aug. 17, 2020, through Aug. 16, 2021, valued at approximately $31,000,000, with three one-year option periods through August 16, 2024. Work will be performed in the greater Washington, D.C., area. The contracting activity is the Defense Information Systems Activity/Defense Information Technology Contracting Organization, Scott Air Force Base, Illinois.
...

Artículo al respecto: https://www.ai.mil/blog_03_25_20-develo ... odges.html
Developing the Joint Common Foundation
By: The JAIC 25 MAR 2020
...
Why is the Joint Common Foundation critical to the U.S. Military to field AI?

We are designing the JCF to:

· Reduce technical barriers to DoD-wide AI adoption.
· Accelerate security assessments of AI products to support rapid authorization decisions and AI capability deployment.
· Create standardized development, security, testing tools, and practices to support secure, scalable AI development.
· Facilitate the concept of secure re-use of AI resources, software, tools, data, and lessons learned that capitalize on the progress made by each JCF AI project.
· Encourage efficiencies by finding patterns in JCF customer needs and creating solutions that are repeatable to build core products that advance AI development
· Mitigate risk by providing a common, standardized, and cyber-hardened infrastructure and platform for AI development, assessments, and rapid deployment promotion.
...
champi
Moderador
 
Mensajes: 13626
Registrado: Vie Nov 21, 2008 10:53 pm

Re: Inteligencia artificial, redes neuronales, machine learning

Notapor poliorcetes el Vie Ago 21, 2020 4:02 pm

En este video del carmel sale el fire weaver en el minuto 2:26


Nunca digas que éste es mi último sendero
**
podcast de portierramaryaire
https://www.ivoox.com/podcast-portierra ... 223_1.html

y recuerda nuestro patreon para actualizar el foro y crecer
https://www.patreon.com/portierramaryaire
Avatar de Usuario
poliorcetes
Site Admin
 
Mensajes: 13139
Registrado: Vie Abr 17, 2009 11:54 am

Re: Inteligencia artificial, redes neuronales, machine learning

Notapor champi el Vie Oct 16, 2020 7:56 pm

Contratos muy importantes de EEUU para desarrollar IA (15/10/2020): https://www.defense.gov/Newsroom/Contra ... e/2383996/
...
JOINT ARTIFICIAL INTELLIGENCE CENTER

Redhorse Corp., San Diego, California (FA701420A0020); Cyber Point International LLC, Baltimore, Maryland (FA701420A0021); Elder Research Inc., Charlottesville, Virginia (FA701420A0019); Barbaricum LLC, Washington, D.C. (FA701420A0018); and Enterprise Resource Performance Inc., Fairfax, Virginia (FA701420A0022), were awarded five-year competitive blanket purchase agreements (BPAs), each with a $100,000,000 ceiling, to provide services to the Department of Defense Joint Artificial Intelligence Center Missions Directorate. The services include software development, machine learning, cognitive and systems engineering, operations research, and user experience design. Work on the contract will occur in Arlington, Virginia. The ordering period is from Sept. 25, 2020, through Sept. 24, 2025. The contracting activity is the Air Force District Washington, Joint Base Andrews, Maryland.
...
champi
Moderador
 
Mensajes: 13626
Registrado: Vie Nov 21, 2008 10:53 pm

Re: Inteligencia artificial, redes neuronales, machine learning

Notapor jesromgar el Sab Oct 24, 2020 9:13 am

Buenas Juan,

Soy Jesus F Roman (@jesusfroman) en twitter.
Te parece usar este Tema como base para volcar información? O prefieres que creemos uno nuevo?

Saludos
Avatar de Usuario
jesromgar
 
Mensajes: 82
Registrado: Jue Jul 12, 2018 5:52 pm

Re: Inteligencia artificial, redes neuronales, machine learning

Notapor jesromgar el Sab Oct 24, 2020 9:43 am

"Army Looks to Disperse Command Posts to Boost Survivability"

https://www.nationaldefensemagazine.org ... Hw.twitter

Puede parecer que tiene poco que ver con el tema, pero tiene bastante que ver con IA, IoT (IoBT) y con el estudio de vulnerabilidad en redes, que es uno de los temas de mi tesina. A ver si un día de estos explico aquí qué hice y para qué pienslo que es útil.

Saludos,
Jesus F Roman
Avatar de Usuario
jesromgar
 
Mensajes: 82
Registrado: Jue Jul 12, 2018 5:52 pm

Re: Inteligencia artificial, redes neuronales, machine learning

Notapor jesromgar el Sab Oct 24, 2020 10:51 am

Buenas,

Volviendo al tema original del Tema, dejo aquí un muy buen artículo de uno de los grandes: Antulio J Echevarrria II (MWI) sobre AI-driven analytics recien publicado en Military Strategy Magazine que me pasó Kique_42 (tb en el foro).

(Si alguien tiene problemas con el inglés: Google Chrome y traducción automática)

https://www.militarystrategymagazine.co ... analytics/
Avatar de Usuario
jesromgar
 
Mensajes: 82
Registrado: Jue Jul 12, 2018 5:52 pm

Re: Inteligencia artificial, redes neuronales, machine learning

Notapor Orel el Dom Oct 25, 2020 10:31 am

Gracias jesromgar
Avatar de Usuario
Orel
Moderador
 
Mensajes: 46184
Registrado: Sab Sep 24, 2005 11:33 am
Ubicación: España, en el bocho

Re: Inteligencia artificial, redes neuronales, machine learning

Notapor poliorcetes el Dom Oct 25, 2020 11:25 am

jesromgar escribió:Buenas Juan,

Soy Jesus F Roman (@jesusfroman) en twitter.
Te parece usar este Tema como base para volcar información? O prefieres que creemos uno nuevo?

Saludos


Perdón por la tardanza en responder. De momento yo creo que es buena idea, y si da para que la Poliorcética IoT tenga su propio hilo, pues lo movemos
Nunca digas que éste es mi último sendero
**
podcast de portierramaryaire
https://www.ivoox.com/podcast-portierra ... 223_1.html

y recuerda nuestro patreon para actualizar el foro y crecer
https://www.patreon.com/portierramaryaire
Avatar de Usuario
poliorcetes
Site Admin
 
Mensajes: 13139
Registrado: Vie Abr 17, 2009 11:54 am

Re: Inteligencia artificial, redes neuronales, machine learning

Notapor jesromgar el Lun Oct 26, 2020 7:09 pm

poliorcetes escribió:
jesromgar escribió:Buenas Juan,

Soy Jesus F Roman (@jesusfroman) en twitter.
Te parece usar este Tema como base para volcar información? O prefieres que creemos uno nuevo?

Saludos


Perdón por la tardanza en responder. De momento yo creo que es buena idea, y si da para que la Poliorcética IoT tenga su propio hilo, pues lo movemos



Sin problemas, todos tenemos nuestros asuntos personales ;)

He aquí un artículo publicado hoy por el US Mad scientist (blog que recomiendo encarecidamente) sobre "Densos Entornos Urbanos" (DUE, Dense Urban Environments) desde ahora hasta 2050.
Los que me conozcan un poco sabrán que me gusta bastante el tema de combate urbano y creo que el campo de operaciones de la IoT (IoBT) será eminentemente urbano, pues será donde se concentren todas las rendes sensoriales propias, de los enemigos y neutrales.

https://madsciblog.tradoc.army.mil/279- ... ough-2050/

La última parte me parece bastante informativa:

"The information space within DUE is extremely congested and will increase as the internet of things (IoT) becomes more pervasive. Consequently, forces operating in DUE will have to move, shoot, and communicate amongst an internet of battlefield things (IoBT) that includes friendly, adversarial, and civilian entities – platforms, sensors, emitters, and all other connected devices. This creates a digital and information “fog of war.” However, there is also an opportunity to exploit adversarial and civilian components of the IoBT to gain a better situational understanding. Cities are, and will be, flooded with millions of agnostic civilian intelligence, surveillance, and reconnaissance (ISR) assets that an operating force can tap into and exploit. Cell phones, smart security systems, networked cameras, social media, and a plethora of other sensors are all capturing relevant information that can be used in a military operation. So too, databases with valuable information will be available to those who can access them.

In the future, military operations in DUE will face even more difficult challenges. Population growth is expected to continue as a larger percentage of the planet migrates to cities. With that will come more and taller buildings, an increase in vehicular traffic, and an increase in connected devices – the last potentially numbering in the trillions. Smart cities will be the norm. Autonomous vehicles will flood the streets. Army forces could use those as an opportunity to gain access and control traffic flow during an operation or an adversary could surreptitiously create a swarm of vehicles that overpowers and overwhelms friendly forces. Both instances will bring about challenges and authorities in ethics and law. Unmanned aerial systems will be prolific. Drones will contest and congest the already volatile airspace, further clouding sensors and surveillance systems. The potential for nations to fracture, censor, or remove themselves from the global internet will further complicate ISR and open source collection.

The fog of war and challenges presented by conflict will tend to be exacerbated by DUE. The density of features means that each bomb, bullet, and missile carries with it the potential for collateral damage, thereby exacerbating hostility or, at the very least, diminishing support from the civilian populace. Fortunately, there are opportunities among those buildings and populations. They are information-rich surroundings not only in terms of sensors and databases – every urban resident is a potential source of information. Cities are also wealthy in other potential resources: building supplies; means of communication; civilian vehicles; and those aforementioned members of police, fire, and other authorities who will have intimate knowledge of their environments. Innovation, decentralized decision-making guided by clear missions and intents, and superior situational awareness are among those capabilities that will allow us to meet the challenges and seize available opportunities."

(Si alguien tiene algún problema con el inglés o necesita aclarar algúna parte, que avise y lo intentamos interpretar juntos sin problemas).

Saludos,
Jesus F Roman
Avatar de Usuario
jesromgar
 
Mensajes: 82
Registrado: Jue Jul 12, 2018 5:52 pm

Re: Inteligencia artificial, redes neuronales, machine learning

Notapor jesromgar el Lun Oct 26, 2020 7:14 pm

Buenas,

Además, también recomendaría encarecidamente escuchar este podcast (The Convergence, también del Mad Scientist). Es una entrevista antigua del Mad Scientist a Dr. Alexander Kott, Chief Scientist del Army Research Laboratory (ARL) sobre Internet of Battlefield Things y la modernización del ejército. Son 36min. y no tiene desperdicio.

https://theconvergence.castos.com/episo ... ander-kott
(Si alguien tiene algún problema entendiendo alguna parte que avise que sin problemas le hacemos un transcript&traducción).

(A modo resumen)
Highlights from the conversation include:

-The battlefield is becoming saturated with devices that can do computation, some kind of thinking, and can communicate. These are not just things the Army owns.
-Complexity can actually be a good thing. Being able to “hide” on the battlefield is a good thing and we can hide in the complexity of the Internet of Battlefield Things.
-The battlefield of the future will be populated by multiple intelligent species. Humans will be very important but just one among them. How do humans co-exist with those intelligent species? We humans are not known for working and living well with other species, not even ourselves.
-The Army (and larger Department of Defense) has a collaborative relationship with industry that is actually beneficial. It is not just a competition for talent but rather a relationship that is a strength. A rising tide lifts all our boats.
-Every war has seen greater and greater ranges in magnitudes from the Civil War up to the Global War on Terrorism. In the future we may see an Army missile that could be intercontinental. We could see artillery “spanning a fraction of the globe.” This leads to global ground warfare and changes the battlefield calculus. Such a shift in warfare could change the Army’s relationship with other services and actualize the reality of multi-domain operations even more.
-Regimes that are unethical, immoral, and authoritarian lose the technological edge in the long run because as they run out of ways to use technologies they have developed for unethical and immoral purposes, they do not have thorough investment in other technology areas where the United States is excelling.
-Long-range, intelligent, precision fires may be a major threat to our Homeland in the future. The Homeland may not be as defensible as it has been for centuries.


Saludos,
Jesus F Roman
Avatar de Usuario
jesromgar
 
Mensajes: 82
Registrado: Jue Jul 12, 2018 5:52 pm

Re: Inteligencia artificial, redes neuronales, machine learning

Notapor poliorcetes el Mar Oct 27, 2020 3:36 pm

Un tema que me llama la atención del futuro de los combates urbanos es de dónde sale el fluido eléctrico para alimentar la red de sensores, Internet of Battle Things o Poliorcética IoT. O se genera o se almacena, porque la red eléctrica será lo primero que se pierda por amplias zonas

Si se genera, o es durante el día y llamando la atención, o es de noche y llamando más la atención

Si se almacena, añadimos costes y sobre todo volumen y peso
Nunca digas que éste es mi último sendero
**
podcast de portierramaryaire
https://www.ivoox.com/podcast-portierra ... 223_1.html

y recuerda nuestro patreon para actualizar el foro y crecer
https://www.patreon.com/portierramaryaire
Avatar de Usuario
poliorcetes
Site Admin
 
Mensajes: 13139
Registrado: Vie Abr 17, 2009 11:54 am

Re: Inteligencia artificial, redes neuronales, machine learning

Notapor Vorlon el Vie Oct 30, 2020 1:03 pm

https://www.c4isrnet.com/artificial-int ... elligence/

Como la IA reducira los tiempos de localización, adquisión y fuego sobre objetivos.

Actualmente, las tripulaciones de los tanques utilizan un proceso muy manual para detectar, identificar y atacar objetivos", explicó Abrams Master Gunner Sgt. Primera Clase Dustin Harris. “Los comandantes de tanques y artilleros se mueven manualmente, tratando de detectar objetivos usando sus sensores. Una vez que encuentran un objetivo, tienen que seleccionar manualmente la munición que van a usar para dar servicio a ese objetivo, aplicar láser al objetivo para obtener un alcance preciso y algunos otros factores ".

El proceso debe repetirse para cada objetivo.

"Eso puede llevar tiempo", añadió. "Todo se hace manualmente todavía".

En el campo de batalla del siglo XXI, es un anacronismo.
Imagen

saludos
Algún día cuando toda civilización y ciencia hayan sido igualmente arrasadas,rezaréis por un hombre con una espada . (Robert E Howard)
Avatar de Usuario
Vorlon
Moderador
 
Mensajes: 7588
Registrado: Dom Jun 16, 2019 7:03 pm
Ubicación: Españistán

Re: Inteligencia artificial, redes neuronales, machine learning

Notapor jesromgar el Vie Oct 30, 2020 3:40 pm

Muchas gracias por el aporte Vorlon.

No relacionado con el IoT, sino con la IA: el CSET ("Center for Security and Emerging Technology", al cual os recomiendo subscribiros si os interesa el tema de la IA, pues tienen una newsletter casi diaria), acaba de publicar un análisis de varios autores (entre ellos una de mis favoritas, Margarita Konaev ya que escribe bastante sobre combate en zonas urbanas, nuevas tecnologías e IA) titulado "U.S. Military Investments in Autonomy and AI: Costs, Benefits, and Strategic Effects". El documento está dividido en 3 partes y se puede acceder en PDf al final de la página.

https://cset.georgetown.edu/research/u- ... c-effects/
Avatar de Usuario
jesromgar
 
Mensajes: 82
Registrado: Jue Jul 12, 2018 5:52 pm

Siguiente

Volver a Defensa

¿Quién está conectado?

Usuarios navegando por este Foro: No hay usuarios registrados visitando el Foro y 1 invitado