'Affordable' hypersonics, small business and sustainment lead DoD tech chief's priorities

Sur le même sujet

• 

  9 août 2023 | International, Sécurité

  New Zealand unveils defense strategy documents. Here’s what they say.

  New Zealand has released three defense documents outlining prevailing challenges, principals for its military and ways to improve the force.

• 

  31 décembre 2018 | International, C4ISR

  US Spies Want to Know How to Spot Compromised AI

  BY DAVE GERSHGORN What if you were training an AI, and an adversary slipped a few altered images into its study set? The US government's research arm for intelligence organizations, IARPA, is looking for ideas on how to detect “Trojan” attacks on artificial intelligence, according to government procurement documents. Here's the problem the agency wants to solve: At a simple level, modern image-recognition AI learns from analyzing many images of an object. If you want to train an algorithm to detect pictures of a road signs, you have to supply it with pictures of different signs from all different angles. The algorithm learns the relationships between the pixels of the images, and how the structures and patterns of stop signs differ from those of speed-limit signs. But suppose that, during the AI-training phase, an adversary slipped a few extra images (Trojan horses) into your speed-limit-sign detector, ones showing stop signs with sticky notes on them. Now, if the adversary wants to trick your AI in the real world into thinking a stop sign is a speed-limit sign, it just has to put a sticky note on it. Imagine this in the world of autonomous cars; it could be a nightmare scenario. The kinds of tools that IARPA (Intelligence Advanced Research Projects Activity) wants would be able to detect issues or anomalies after the algorithm has been trained to recognize different objects in images. This isn't the only kind of attack on AI that's possible. Security researchers have also warned about inherent flaws in the way artificial intelligence perceives the world, making it possible to alter physical objects like stop signs to make AI algorithms miscategorize them without ever messing with how it was trained, called “adversarial examples.” While neither Trojan attacks nor the adversarial examples are known to have been used by malicious parties in the real world, researchers have said they're increasingly possible. IARPA is looking at a short timeline as well, expecting the program to conclude after a maximum of two years. https://www.defenseone.com/technology/2018/12/us-spies-want-know-how-spot-compromised-ai/153826

• 

  26 septembre 2018 | International, Aérospatial

  US Air Force turns to data analytics to solve B-1, C-5 maintenance challenges

  By: Valerie Insinna WASHINGTON — The U.S. Air Force is making changes to the way it sustains the B-1B Lancer bomber and C-5 Super Galaxy cargo plane, moving to a maintenance approach that will allow it to use data analytics to predict problems, the acting head of Air Force Materiel Command said. Both the B-1 and C-5 fleets transitioned to a conditions-based maintenance model last month, Lt. Gen. Robert McMurry, commander of the Air Force Life Cycle Management Center, told Defense News in a Sept. 18 interview. “Given the aging fleet situation that we have, we probably need to be using data better to take care of it — which is a drive toward what most everyone right now is saying is the right way to manage fleet sustainment, which is through condition-based maintenance and data analytics,” he said. “So we're trying to bring that on.” The approach — which involves using algorithms to predict the need for repairs rather than waiting for a part to break — is a standard practice in the commercial airline industry to help reduce maintenance-related delays or cancellations, but has been less common in the Air Force. AFMC determined it needed to make a greater push toward conditions-based maintenance as a result of servicewide reviews triggered by rising concerns about the number of aviation-related mishaps. The first review, directed by Air Force Chief of Staff Gen. Dave Goldfein, involved a one-day standdown that would give flying and maintenance units a chance to communicate potential safety concerns up the chain of command. Gen. Ellen Pawlikowski, then the head of AFMC, also directed the organizations under her command, like the Air Force Sustaiment Center, to evaluate its own data. The reviews have since concluded, with the Air Force finding “two systems ... where high risk was accepted,” said McMurry, noting that “operational security does not allow us to identify them.” “Our process is dealing with those responsibly,” he added. The B-1 and C-5 were chosen as pilot programs for the conditions-based maintenance approach because they are sustained by airmen and have older, relatively small inventories, making for a more manageable data set. But the planes have something else in common — a recent history of well-publicized mishaps. The C-5 has sustained a number of nose landing gear malfunctions that led to a standdown and maintenance assessment in 2017. But despite a fix being put in place, there have still been problems with the gear, such as a March 2018 event where one C-5 landed on its nose at Joint Base San Antonio-Lackland, Texas. Meanwhile, the B-1 fleet was temporarily grounded in June after a safety investigation board found problems with ejection seat components while investigating a May 1 emergency landingwhere the ejection seats did not deploy. Full article: https://www.defensenews.com/digital-show-dailies/air-force-association/2018/09/25/air-force-looks-to-data-analytics-to-help-solve-b-1-c-5-maintenance-challenges/