Inteligencia artificial, redes neuronales, machine learning

[Truquichan]

A ver si encuentro el origen de un caso divertido de una IA para reconocer tanques, que a pleno sol y con el tanque bien contrastado, fallaba en reconocerlo. Resultó que la dieron de comer fotos de tanques en dias nublados.

[Atticus]

En temas de defensa, a muchas cosas tecnologicas se les esta dando el tratamiento de la magia. Cuando lo importante es llenar power point de afirmaciones grandilocuentes y acronimos ingeniosos y "agresivisisimos de la muerte" pasan estas cosas. Se vive en algunos lados bajo una presion terrible por ofrecer "molonidad", no importando que no haya nada mas que luces y humo de colores tras ello. El tema de las IA, etc, es fundamental para un futuro cada dia mas cercano, pero no le hace ningun bien que se pongan sobre sus hombros pesos que aun no pueden soportar. Ni que se les poco menos que obligue a aparecer como fuentes de cosas que ni siquiera se sabe como se va a llegar a ellas. Me quejo de una presion insana por sacar power point sin importar la velocidad real en que avanzan los conocimientos.

[poliorcetes]

En temas de defensa, a muchas cosas tecnologicas se les esta dando el tratamiento de la magia. Cuando lo importante es llenar power point de afirmaciones grandilocuentes y acronimos ingeniosos y "agresivisisimos de la muerte" pasan estas cosas. Se vive en algunos lados bajo una presion terrible por ofrecer "molonidad", no importando que no haya nada mas que luces y humo de colores tras ello. El tema de las IA, etc, es fundamental para un futuro cada dia mas cercano, pero no le hace ningun bien que se pongan sobre sus hombros pesos que aun no pueden soportar. Ni que se les poco menos que obligue a aparecer como fuentes de cosas que ni siquiera se sabe como se va a llegar a ellas. Me quejo de una presion insana por sacar power point sin importar la velocidad real en que avanzan los conocimientos.

Ocurre igual fuera de la defensa. Ya no viene un nuevo AI winter, pero sí que hay hypes descontrolados por momentos.

El principal problema de la IA respecto a defensa EN OCCIDENTE es que la industria IT occidental está demasiado poco conectada con la industria de defensa. Aún las empresas de defensa más grandes son cada vez menos relevantes en capital financiero y humano comparadas con la GAFAM

[poliorcetes]

Contrato de EEUU para desarrollar el software CATANA ("Command, Acquire, Track, Assess, eNgage, and Analyze", 24/11/2021): https://www.defense.gov/News/Contracts/ ... e/2854300/

...
MISSILE DEFENSE AGENCY

BlueHalo LLC, Huntsville, Alabama, is being awarded a cost-plus-fixed-fee contract. The total value of this contract is $30,341,268. Under this new contract, the contractor will develop a Command, Acquire, Track, Assess, eNgage, and Analyze (CATANA) cloud development that includes: 1.) Infrastructure and interface development; 2.) Core component and service development; and 3.) External service integration and assessment. The work will be performed in Huntsville, Alabama, with an estimated completion date of November 2025. This contract is awarded under broad agency announcement (BAA) procedures and fulfills the competition requirement, in accordance with Federal Acquisition Regulation (FAR) 6.102(d)(2)(i). The BAA, HQ0852-21-S-0001, posted on beta.SAM.gov on Dec. 14, 2020, fulfills the publication of this contract on the Government-wide Point of Entry website requirement in accordance with FAR 35.016(f). Fiscal 2021 research, development, test and evaluation funds in the amount of $5,715,516.74 are being obligated on this award. The Missile Defense Agency, Huntsville, Alabama, is the contracting activity (HQ0852-22-C-0001).
...

Esta mierda va a ser de los multiplicadores mas importantes: parte del orquestador o al menos su preprocesado del batiburrillo distribuido que quieren.

Por cierto, para todos los follapostes aquí presentes: sabéis que 30M son jodidos cacahuetes para un software así. Ni remotamente llega

[Truquichan]

Y el presupuesto para el core/routing físico de las comunicaciones direccionales furtivas me pareció poco para todo lo que es. Parece que van a la baja a ver si sale alguien/alguna universidad que se lo haga.

[Truquichan]

... Cosas de Champi y contrato...

... Cosas del menda...

Por cierto, para todos los follapostes aquí presentes: sabéis que 30M son jodidos cacahuetes para un software así. Ni remotamente llega

Es tan abierto que quieren que el boceto se lo resuelvan, orienten o montárselo en plan startup. Además va sobre plataformas conocidas como azure o aws. Quieren abaratar costes desde el principio para estos monstruos. Solo la rebaja de listón en los nuevos proyectos al usar MOSA o IaS conocidas abre enormemente el abanico de contendientes: todo el ambito academico universitario, para el que 30M$ si que es jugoso junto al "prestigio" para cazar mas estudiantes y algún royalti.

Adicionalmente ten cuidado con los telecos milenialls que no saben del verdadero lenguaje inclusivo.

[champi]

"AI and Autonomy in Russia Issue 27, December 6, 2021": https://www.cna.org/CNA_files/PDF/AI-an ... sue-27.pdf

[Orel]

https://www.defensa.com/espana/cinco-te ... sta-decada

[Orel]

Insisten mucho en el uso de IA en el NGAD y que tomará decisiones:

US Air Force to advance stealthy successor for F-22
23 December 2021

The USAF’s Next-Generation Air Dominance (NGAD) fighter remains under wraps, but some clues about capabilities the service wants in the aircraft are surfacing.
The sixth-generation fighter aircraft (one full-scale flight demonstrator secretly flew for the first time in 2020) is expected to replace the service’s F-22 Raptor fleet, starting in the 2030s.

ARTIFICIAL INTELLIGENCE TO DIGEST DATA

Whereas past generations of fighter aircraft relied upon speed and manoeuvrability to defeat foes, it seems the NGAD will lean heavily on stealth characteristics to hide from opponents and on computing power to outsmart them, according to comments from the USAF and likely development partners.

Disclosures point to a hyper-connected stealth aircraft that will use artificial intelligence programs to rapidly digest and make sense of multiple streams of sensor data – information that will help combat pilots beat their adversaries to the punch.

NGAD will be a multi-role combat aircraft, but air dominance will be its primary mission, General Charles Brown, USAF chief of staff, told the US House Armed Services Committee in June. He added that the service wants the aircraft to have an increased weapons load and increased range. Greater range would be useful flying across the vast areas of the Indo-Pacific region, Brown said. Greater weapons load would probably be needed in combat against China’s air force, which the Pentagon expects to have a numerical advantage.

Winning air battles will require more than a bigger arsenal of missiles. In order to eliminate China’s numerical advantage, each fighter will have to be able to repeatedly find enemy aircraft and fire quickly – again and again.
Lockheed, a leading contender to develop the sixth-generation fighter, says new digital technologies will give NGAD “omniscient situational awareness”.

BIG DATA

An all-knowing capability also fits into the USAF’s desire for an Advanced Battle Management System (ABMS), a battlefield network that would allow pilots to make decisions faster using data gathered from around the combat theatre.

For example, ABMS might be used to pass intelligence, surveillance and reconnaissance information gathered by autonomous loyal wingman unmanned air vehicles (UAVs) back to the NGAD platform, which could then use artificial intelligence programs to make sense of it all. The USAF has described NGAD as being a “family of systems” with the manned fighter at its centre.

In October, the USAF awarded Kratos Defense & Security Solutions and General Atomics Aeronautical Systems each a contract to develop an “Off Board Sensing Station” UAV. Such a loyal wingman would probably fly in advance of NGAD, search for targets and threats using its sensors, such as radar and infrared search-and-track sensors, and then possibly act as a weapons launch platform.

Raytheon, a manufacturer of advanced radars, expects sensors on NGAD to be automatically “harmonised” to find targets. Some sensors, such as radar, might also be automatically turned off in certain situations to reduce the jet’s electronic signature to avoid detection, the company says.

ADVANCED ADVERSARIES

Interest in artificial intelligence partly comes from a belief that the future battlefield is going to be overwhelmingly chaotic and complex – challenging circumstances created by large numbers of sophisticated radars, electronic warfare systems, surface-to-air missiles, and fighter aircraft fielded by advanced adversaries such as China and Russia.

“It’s going to take a suite of sensors integrated together,” says Eric Ditmars, vice-president of secure sensor solutions, Raytheon Intelligence & Space. “There are environments where the radar performs phenomenally,” he says. “There are environments where the radar is jammed, where infrared search-and-track systems are phenomenal.”

Integrated systems must be able to respond to changing circumstances.
“The environments are getting so contested that you really have to have the ability to be more adaptive,” says Ditmars. “The intent is to allow that pilot to be able to be more flexible in the mission that they are executing, and not be as reliant upon the pre-planning that has been done.”

“This [system] is deciding, ‘In this environment, I need to use my AESA radar in this mode. I’m not going to use my [electronic warfare] system because that’s going to be detected,’” he says.
Potential scenarios might be solved ahead of time by training artificial intelligence (AI) programs using computer simulations of combat, he says.
“That’s the great thing about artificial intelligence. You give it a set of defined criteria and it figures it out,” Ditmars says.

The concept has a precedent. Researchers with Air Combat Command recently developed the ARTUµ software, a machine learning program that used more than half a million computer simulations to train the radar on the U-2 surveillance aircraft to find enemy missile launchers. In late 2020, the artificial intelligence program was demonstrated aboard a U-2 at Beale AFB in California.

“ARTUµ was responsible for sensor employment and tactical navigation, while the pilot flew the aircraft and co-ordinated with the AI on sensor operation,” explained the service. “Together, they flew a reconnaissance mission during a simulated missile strike. ARTUµ’s primary responsibility was finding enemy launchers while the pilot was on the lookout for threatening aircraft, both sharing the U-2’s radar.”

NEW TEAM

The USAF said the AI software was “easily transferable” to other systems, and that it planned to refine the technology.
“Putting AI safely in command of a US military system for the first time ushers in a new age of human-machine teaming and algorithmic competition,” said Will Roper, who was assistant secretary of the USAF for acquisition, technology and logistics at the time. “Failing to realise AI’s full potential will mean ceding decision advantage to our adversaries.”

In other words, asking a pilot to make sense of complex sensor data in the middle of a pitched battle might lose precious seconds to the enemy.

“We’re trying to take some of this workload off the pilot. They are human and they can only do so much,” Ditmars says. “As the systems get more and more complex, it becomes very challenging for them.”

https://www.flightglobal.com/defence/us ... 92.article

[poliorcetes]

Winning air battles will require more than a bigger arsenal of missiles. In order to eliminate China’s numerical advantage, each fighter will have to be able to repeatedly find enemy aircraft and fire quickly – again and again.
Lockheed, a leading contender to develop the sixth-generation fighter, says new digital technologies will give NGAD “omniscient situational awareness”.

Qué batallas aéreas? Qué ventaja numérica china presente o futura???


Enviado desde mi iPhone utilizando Tapatalk

[Atticus]

says new digital technologies will give NGAD “omniscient situational awareness”.

Ah, pues ya me quedo mas tranquilo. Aqui, poniendo un poco de "omniscient situational awareness" al diseño.

[Orel]

Qué batallas aéreas? Qué ventaja numérica china presente o futura???

Las batallas aéreas que habría si se enfrentan dos fuerzas similares como la yanqui y la china, incluyendo contra drones y entre drones. No pondremos en duda el combate AA porque apenas ha habido en conflictos disimilares, ¿no?, sería como poner en duda el combate ASW/ASuW porque ¿cuántos ataques aéreos antibuque y antisubmarino ha habido en décadas? Si vis pacem...

Y ventaja numérica china, pues dado que EEUU contempla entre otros escenarios Japón o Taiwán y similares al ladito mismo, China tiene ventaja numérica pues parten de sus propias bases, no tienen que hacer despliegues en portaaviones ni en bases lejanas como tendría que hacer EEUU, lo que siempre limita las cantidades, la logística, etc.

Aqui, poniendo un poco de "omniscient situational awareness" al diseño.

Sabía que se comentaría, jeje, y con razón. Eso es como cuando nació el F-22 y lo denominaron de "supremacía aérea" para que quedase claro que era mucho mejor que los de "superioridad aérea" de entonces. O al Rafale que lo llamaron "omnirrol" para dejarlo por encima de los "multirrol". Se les va un poco a los de mercadotecnia, jeje.

[Azael]

says new digital technologies will give NGAD “omniscient situational awareness”.

Ah, pues ya me quedo mas tranquilo. Aqui, poniendo un poco de "omniscient situational awareness" al diseño.

No me caes tan bien, pero después de esto, no me caes tan mal. Te has lucido con Salt Bae.

[champi]

China iguala a EEUU en inversión en IA: https://www.nationaldefensemagazine.org ... ding-on-ai

[champi]

Informe del "Inspector General" sobre el programa Maven (" Algorithmic Warfare Cross-Functional Team"), "Evaluation of Contract Monitoring and Management for Project Maven": https://media.defense.gov/2022/Jan/10/2 ... DACTED.PDF

Algo sobre este macroproyecto: https://www.c4isrnet.com/intel-geoint/i ... g-project/

By Mark Pomerleau
May 2, 2018
...
Known as Project Maven, the DoD initiative aims to accelerate the integration of big data and machine learning, first focusing on processing of full motion video from tactical drones and from medium altitude sensors.
...
To exemplify the problem, military personnel at the annual GEOINT symposium in Tampa, Florida in late April said in fiscal 2017, 127 terabytes of captured enemy data was collected. In addition, the annual video that Central Command collects could cover 325,000 feature films and the annual signals intelligence Central Command collects is equal to roughly 5.5 million songs.
...