Закрыт

Artificial Intelligence becomes life-long learner with new framework

We need an artist's concept to visualize this news release. Attached is a stock photo. We would like something similar, but original that the laboratory would own.

A project of the U.S. Army has developed a new framework for deep neural networks that allows artificial intelligence systems to better learn new tasks while forgetting less of what they have learned regarding previous tasks.

The North Carolina State University researchers, funded by the Army, have also demonstrated that using the framework to learn a new task can make the AI better at performing previous tasks, a phenomenon called backward transfer.

"The Army needs to be prepared to fight anywhere in the world so its intelligent systems also need to be prepared," said Dr. Mary Anne Fields, program manager for Intelligent Systems at Army Research Office, an element of U.S. Army Combat Capabilities Development Command's Army Research Lab. "We expect the Army's intelligent systems to continually acquire new skills as they conduct missions on battlefields around the world without forgetting skills that have already been trained. For instance, while conducting an urban operation, a wheeled robot may learn new navigation parameters for dense urban cities, but it still needs to operate efficiently in a previously encountered environment like a forest."

The research team proposed a new framework, called Learn to Grow, for continual learning, which decouples network structure learning and model parameter learning. In experimental testing it outperformed pervious approaches to continual learning.

"Deep neural network AI systems are designed for learning narrow tasks," said Xilai Li, a co-lead author of the paper and a Ph.D. candidate at NC State. "As a result, one of several things can happen when learning new tasks, systems can forget old tasks when learning new ones, which is called catastrophic forgetting. Systems can forget some of the things they knew about old tasks, while not learning to do new ones as well. Or systems can fix old tasks in place while adding new tasks -- which limits improvement and quickly leads to an AI system that is too large to operate efficiently. Continual learning, also called lifelong-learning or learning-to-learn, is trying to address the issue."

To understand the Learn to Grow framework, think of deep neural networks as a pipe filled with multiple layers. Raw data goes into the top of the pipe, and task outputs come out the bottom. Every "layer" in the pipe is a computation that manipulates the data in order to help the network accomplish its task, such as identifying objects in a digital image. There are multiple ways of arranging the layers in the pipe, which correspond to different "architectures" of the network.

When asking a deep neural network to learn a new task, the Learn to Grow framework begins by conducting something called an explicit neural architecture optimization via search. What this means is that as the network comes to each layer in its system, it can decide to do one of four things: skip the layer; use the layer in the same way that previous tasks used it; attach a lightweight adapter to the layer, which modifies it slightly; or create an entirely new layer.

This architecture optimization effectively lays out the best topology, or series of layers, needed to accomplish the new task. Once this is complete, the network uses the new topology to train itself on how to accomplish the task -- just like any other deep learning AI system.

We need an artist's concept to visualize this news release. Attached is a stock photo. We would like something similar, but original that the laboratory would own.

A project of the U.S. Army has developed a new framework for deep neural networks that allows artificial intelligence systems to better learn new tasks while forgetting less of what they have learned regarding previous tasks.

The North Carolina State University researchers, funded by the Army, have also demonstrated that using the framework to learn a new task can make the AI better at performing previous tasks, a phenomenon called backward transfer.

"The Army needs to be prepared to fight anywhere in the world so its intelligent systems also need to be prepared," said Dr. Mary Anne Fields, program manager for Intelligent Systems at Army Research Office, an element of U.S. Army Combat Capabilities Development Command's Army Research Lab. "We expect the Army's intelligent systems to continually acquire new skills as they conduct missions on battlefields around the world without forgetting skills that have already been trained. For instance, while conducting an urban operation, a wheeled robot may learn new navigation parameters for dense urban cities, but it still needs to operate efficiently in a previously encountered environment like a forest."

The research team proposed a new framework, called Learn to Grow, for continual learning, which decouples network structure learning and model parameter learning. In experimental testing it outperformed pervious approaches to continual learning.

"Deep neural network AI systems are designed for learning narrow tasks," said Xilai Li, a co-lead author of the paper and a Ph.D. candidate at NC State. "As a result, one of several things can happen when learning new tasks, systems can forget old tasks when learning new ones, which is called catastrophic forgetting. Systems can forget some of the things they knew about old tasks, while not learning to do new ones as well. Or systems can fix old tasks in place while adding new tasks -- which limits improvement and quickly leads to an AI system that is too large to operate efficiently. Continual learning, also called lifelong-learning or learning-to-learn, is trying to address the issue."

To understand the Learn to Grow framework, think of deep neural networks as a pipe filled with multiple layers. Raw data goes into the top of the pipe, and task outputs come out the bottom. Every "layer" in the pipe is a computation that manipulates the data in order to help the network accomplish its task, such as identifying objects in a digital image. There are multiple ways of arranging the layers in the pipe, which correspond to different "architectures" of the network.

When asking a deep neural network to learn a new task, the Learn to Grow framework begins by conducting something called an explicit neural architecture optimization via search. What this means is that as the network comes to each layer in its system, it can decide to do one of four things: skip the layer; use the layer in the same way that previous tasks used it; attach a lightweight adapter to the layer, which modifies it slightly; or create an entirely new layer.

This architecture optimization effectively lays out the best topology, or series of layers, needed to accomplish the new task. Once this is complete, the network uses the new topology to train itself on how to accomplish the task -- just like any other deep learning AI system.

Квалификация: Adobe Illustrator, Drawing, Изобразительное искусство

Показать больше artificial intelligence algorithm, project artificial intelligence india pdf, source code artificial intelligence game snake, artificial intelligence projects user allocation, autolisp artificial intelligence, artificial intelligence connect game computer, artificial intelligence code reverse engineering, artificial intelligence ikea anna, artificial intelligence captcha resume, summarize artificial intelligence, personal website artificial intelligence aspnet, verilog artificial intelligence fpga, artificial intelligence bot, computer graphics artificial intelligence, projects java artificial intelligence, artificial intelligence freelance, artificial intelligence matlab, artificial intelligence bot maker, artificial intelligence pdf file

О работодателе:
( 0 отзыв(-а, -ов) ) Aberdeen Proving Ground, United States

ID проекта: #19720375

12 фрилансеров(-а) в среднем готовы выполнить эту работу за $95

invevolution

Hey, I can help you with completing your project as its completely doable for me. Please message me so we can discuss about it. Thanks

$30 USD за 1 день
(1 отзыв)
1.2
saminakiran2

Hi,i would like to be considered for your writing project.I am a professional research writer with over 6 years of experience in providing high quality and plagiarism free work.I am well versed in evaluation of studie Больше

$100 USD за 1 день
(0 отзывов(-а))
0.0
sodiqa32

Hello, I am pleasure with your job as detailed. Thank you for the job posting. It’s a pleasure to meet you. I’d really like to work with you on this one if possible! I do have a couple of questions, but first I’d like Больше

$30 USD за 1 день
(0 отзывов(-а))
0.0
birdeyeinv

Hi there? I can help you with completing your project. I assure you to deliver high quality work without any delay. I never compromise on the quality of my work. Please message me so that we can discuss further. Reg Больше

$30 USD за 1 день
(0 отзывов(-а))
0.0
LynneAnderson

I provide support in writing essays, multiple questions, term papers, research papers, reports, dissertations and all forms of writing. I am able to do this in the required formats including MLA, APA and Harvard. For 6 Больше

$200 USD за 3 дней(-я)
(0 отзывов(-а))
0.0
Haaly

Dear I can perfectly do the project. I have been working on such projects. I deliver quality, commitment and make sure there is no Plagiarism. I am familiar with all sorts of reference styles. Please s Больше

$90 USD за 3 дней(-я)
(0 отзывов(-а))
0.0
Adamsmulwa34

Hi. I am a well-experienced writer in this subject, with an excellent command in English, and well-advanced with all writing formats (APA, MLA, HARVARD, CHICAGO, TURBIAN, OSCOLA, AMA). Please assign me this order and I Больше

$155 USD за 3 дней(-я)
(0 отзывов(-а))
0.0
iridescent2x15

Hi I am an engineer and have delivered many research and technical projects. You can share more details with me so that we can negotiate the price accordingly. Thank you

$60 USD за 7 дней(-я)
(0 отзывов(-а))
0.0
IdealAftab

Hello I am an experienced Academic Research paper writer. Education: I have done Masters Software engineering and also completed Masters in English Literature. Experience: I am doing Больше

$30 USD за 3 дней(-я)
(0 отзывов(-а))
0.0
kabbylinks

My Bachelor’s degree in Information Technology and four years of experience constructing, managing, and improving computer networks makes me a solid candidate for your organization. My professional experience includ Больше

$155 USD за 3 дней(-я)
(0 отзывов(-а))
0.0
fapcent

Thanks for the project that you have posted. In your project description, I have gathered that you need someone with a deep understanding of on page SEO. I have done hundreds of keyword rich articles that pass plagiari Больше

$155 USD за 3 дней(-я)
(0 отзывов(-а))
0.0
ZhenJin

Dear customer. I have checked your description carefully. To be honest, I have just started working here, but i am sure I have rich experience for your task. Because i am a software developer with several years expe Больше

$100 USD за 3 дней(-я)
(0 отзывов(-а))
0.0