AI is making great use of the museum's vast data

Leifeng.com: What kind of sparks would emerge if a museum with an exploding amount of information were combined with advanced artificial intelligence? The article "Israel Holocaust Museum Embraces Artificial Intelligence to Help Visitors Get Inspired by Its Vast Archives" published on NVIDIA Blog may tell you. Leifeng.com's full text is translated as follows:

Last week, people around the world marked Holocaust Remembrance Day, remembering not only Hitler's victims but also the non-Jews who risked their own lives to try to save them.

The world's most famous Holocaust memorial center, Yad Vashem, a site that attracts one million visitors a year and has been visited by six U.S. presidents, is dedicated to preserving for future generations the memory of the Jews who died six million years ago at the hands of the German Nazis and their collaborators. Yad Vashem's archive includes unique, intense video testimonies, short films, photographs, personal written records, Nazi documents, and audio files. This 800 million digital asset, containing 4 petabytes of data, is twice as much as the U.S. Library of Congress.

The World Holocaust Remembrance Center, a place of document sources for scholars around the world, stores a large archive of digital media about victims and survivors that is difficult to access and find. It is a daunting challenge for researchers to keep up with the pace of compiling history, let alone for younger generations to access it. Since these documents were submitted and discovered over decades, they will become a source of information for Holocaust scholars around the world after they are fully digitized. Therefore, the Jerusalem-based organization is seeking the help of artificial intelligence to help identify, organize and link photos and other historical documents from the vast amount of data for easier discovery.

Yad Vashem’s team used deep neural networks to let image recognition algorithms help index and categorize its digital history. Michael Lieber, Yad Vashem’s chief information officer, said this could help find new relationships and stories about Holocaust victims.

“We are the first institution in the world that deals with cultural heritage, and we decided to digitally reproduce everything because it is the way to reach a wider audience around the world,” Lieber said. He is optimistic that artificial intelligence will help better identify resources to tell the stories of Holocaust victims and survivors on its social media accounts. This could help connect with younger audiences, he said. In addition, he hopes that researchers can use deep learning methods to reveal new historical information that cannot be easily discovered.

Many people visit Yad Vashem to research what happened to their grandparents and great-grandparents and piece together their family histories. The problem is that the collection of digitized data, which is likely to double in the coming years, will be even harder to search.

Yad Vashem's technical team aims to change that by leveraging high-performance computing-driven deep learning. The company plans to use the supercomputing power of the NVIDIA DGX-1 AI system to help organize and augment its history through deep learning. The DGX-1 provides the power of hundreds of CPU-based servers in a system that can perform more than a petaflop of AI computing power per second.

Yuval Mazor, senior solutions architect at NVIDIA, said DGX-1 brings Yad Vashem together with the world's most innovative organizations deploying AI to meet challenges. "They are gaining tangible benefits from the application of AI," he said. "For example, Yad Vashem can use video analytics to understand and predict museum traffic and the impact of individual exhibits, and can extract deep insights from rich historical data. These can help Yad Vashem achieve its primary mission of reaching and educating as many people as possible."

Lieber said unsupervised learning holds the promise of enabling trained neural networks to create meta-tags for digital products, allowing deep learning to connect the dots of various information. "If you manage to find a prison card from the Mauthausen concentration camp, the system will know that it is a prisoner card. It will lead you to the relevant data fields and documents, and you will be able to locate and identify the type of document and provide additional information without human intervention," he said.

Another option is to have a large number of people tag hundreds of millions of digital media assets and continue to track and update the database. NVIDIA researchers in Israel are working on this work in collaboration with Yad Vashem.

Leifeng.com Note: This article is compiled from NVIDIA Blog

Microsoft Build 2019: Windows 10 retires, IE is reborn, and Azure becomes the protagonist

Lenovo responds to online rumors; WeChat's Drift Bottle feature is officially offline; Cook: I don't want users to be addicted to iPhone

Ling Xiaofeng, a member of the Canadian Academy of Engineering: Chinese students lack the spirit of questioning and innovation

Alibaba responds to "splitting the big and medium platforms"; Buffett and Munger: Missing out on Google is a huge mistake; the world's second largest dark web trading platform was destroyed

Is Google giving up on mobile phones?

Apple released new financial report, and was "dragged down" by the Chinese market again

Alibaba’s ten-year AI boss started by “liberating” porn appraisers