Intel Labs and iOmniscient Collaborate on Machine Learning Anomaly Detection

Highlights

  • Intel Labs is collaborating with iOmniscient on research using machine learning anomaly detection algorithms, starting with identifying objects on escalators.

  • Intel Labs’ cutting edge probabilistic anomaly detection algorithm will be integrated in several iOmniscient surveillance solutions for field trial and benchmarking.

author-image

投稿者:

Intel Labs and iOmniscient are collaborating on research and development of machine learning (ML) anomaly detection algorithms. Through this complementary research effort, Intel Labs gains access to iOmniscient’s high quality and practical data sets, while iOmniscient may use the state-of-the-art algorithms developed by Intel Labs.

Through its ongoing work on probabilistic computing and related algorithms, Intel Labs has developed cutting edge technologies in anomaly detection. An integral part of developing new algorithms in ML and artificial intelligence (AI) is access to high quality data sets and deployment environments for training, testing and evaluating the algorithms. iOmniscient, an AI technology leader in providing video analytics and automated surveillance solutions, has extensive data sets suitable for ML and AI research and development, plus a vast ecosystem of partners for field trial and benchmarking.

“We value Intel's ongoing commitment to staying ahead of the market. We're excited to collaborate on research that will combine our unique AI technologies,” said Dr. Rustom Kanga, co-founder and CEO of iOmniscient. The company’s solutions are powered by Intel architecture.

Intel Labs’ probabilistic anomaly detection algorithm will be integrated in several iOmniscient surveillance solutions for field trial and benchmarking. The two companies have started collaboration on anomaly detection on escalators — a project that could improve the safety of escalators and reduce their down time. The system will be trained with labeled images or video streams of escalators that detect and identify normal objects as well as abnormal objects, and the algorithm will classify the objects into different classes. For example, anomalies could include a young child left alone on an escalator, an animal riding on an escalator, or objects that potentially could damage the escalator.

“We’re pleased to develop a relationship with iOmniscient to provide field trial opportunities for ML technologies developed at Intel Labs, especially in the area of visual analytics, probabilistic computing and multimodal sensory analytics,” said Pranav Mehta, vice president of Intel Labs and director of Systems and Software Research..

In the future, the companies plan to collaborate on research and development of technologies using multimodal ML and AI, and detection and identification using low quality sensory signals.