Using AI to Save Missing and Exploited Children – Learn More at SXSW

Co-Authored by Lisa Thee, Artificial Intelligence and Analytics Solution Owner at Intel

Intel co-founder Robert Noyce said “Optimism is an essential ingredient of innovation” – and we’re extremely optimistic about artificial intelligence (AI). We’re already using AI to not only solve engineering and business problems, but also to make life better for mankind. That’s why we’re excited to share about the work we’re doing with the National Center for Missing and Exploited Children (NCMEC) to use AI to protect and rescue children.

Technology Challenges of Finding Missing and Exploited Children

NCMEC is a nonprofit organization that serves as a national clearing house for reports of missing children and child sexual exploitation, and they have exactly the kind of problems that AI can help solve. They must search tens of millions of images in an effort to identify children and suspects; to associate images with network addresses or domain names; and to distinguish new images from known ones. There are about 460,000 images of missing children in the NCMEC database, and last year NCMEC received more than 8 million reports of possible child sexual exploitation. Their work is essential to helping law enforcement agencies intervene and rescue endangered children – but investigating all of these reports in a timely manner with a staff of 25 analysts is impossible for a human to do without the assistance of technology.

The image recognition challenges are particularly significant. They must be able to match an image of a child to another image where the age, hair style and color, setting, lighting, and other factors are different. They also want to be able to recognize other elements of a photo like furniture or bedding, so they can relate multiple images of different children taken in the same location. While image recognition has advanced considerably in the past few years, this is much more nuanced than distinguishing a cat from a dog. With our partners, we have set out to develop a customized deep learning solution to help automate this complex image analysis.

Technology Community Coming Together to Solve These Challenges

Intel is working with collaborators like Microsoft and Google to accelerate image search tools that NCMEC can apply in all parts of their operation. Intel is working with NCMEC on an enterprise data management vision to integrate their existing data silos and create a scalable storage architecture that lets them apply common tools across many applications. We are helping them automate and streamline the analytics pipeline their analysts apply to investigate the more than 8 million tips per year pouring in through their CyberTipline (largely fed by cloud service providers themselves using AI to identify child pornography in their networks).

Intel AI and Deep Learning can help identify victims and suspects involved in child exploitation through image recognition and matching.
Bob Rogers, Intel, Marita Rodriguez, NCMEC, Lisa Davis, Intel, Rachel Tyler, Intel, John Shehan, NCMEC, Kathleen Crowe, Intel, Michelle DeLaune, NMEC, Lisa Thee, Intel, John Clark, NCMEC, Laura, Horton, Intel

See for Yourself How We’re Using AI to Find Missing and Exploited Children

To raise awareness for both the NCMEC mission and the AI technology we’re applying to it, we’re at South by Southwest* (SXSW*) in Austin, TX, along with our industry collaborators, demonstrating how we use deep learning to analyze children’s images. It’s a great chance to see the technology and understand how it goes beyond existing image search tools to solve problems in business and society that we have never been able to tackle before.

Here’s where you can see us at SXSW:

We’ll also be at Strata + Hadoop World in San Jose, CA in March, demonstrating the technology in Intel booth #917. Come by and see us if you’re there.

Robert Noyce also said, “Don't be encumbered by history, just go out and do something wonderful.” We think he’d be pleased with how we’re applying AI for good.

 

*Third-party marks and brands may be the property of their respective owners.