Partners Can Drive Cisco's Deep Fusion Reasoning Engine Into Smart Cities, Health Care, Manufacturing

The Cisco system is in early development and executives say pilot programs in Europe are promising for partners willing to develop capabilities around artificial intelligence and customer business outcomes.

Crosstown Traffic

Cisco Systems took partners on a journey last week into a future where artificial intelligence can analyze and manage something as complex and unpredictable as a busy city intersection.

id
unit-1659132512259
type
Sponsored post

A system called Deep Fusion Reasoning Engine is in early development at Cisco, and executives say pilot programs in Europe point to promise for partners willing to develop capabilities around AI and customer business outcomes.

Ruba Borno, vice president of growth initiatives and chief of staff to CEO Chuck Robbins, and Sean Curtis, Cisco director of technology experiences, demonstrated the system in a keynote this week at Cisco's Partner Connection Week conference.

From a nearly indecipherable web of code drawn from video data, the system identifies objects with great accuracy, determines patterns, predicts interactions and makes recommendations. Borno and Curtis demonstrated the system using a city street as an example, but said the AI improvements the system makes will be applicable across a wide range of industries from health care to manufacturing.

Cisco will rely on partners' customer knowledge to drive the system into the market, and Borno said significant opportunities await partners with code development expertise.

Screening Process

The Deep Fusion Reasoning Engine uses video analytics to track things. In the example used by Borno and Curtis, those things were pedestrians and vehicles on and around a busy European street. AI has given video analytics systems the ability to identify objects, Curtis said, and Cisco has been able to increase the accuracy of those IDs to 95 percent. "But we need to be able to go beyond identifying objects," Curtis said. "We need to be able to predict business outcomes."

Goal Setting

To get to the point where the system can predict outcomes, users set goals. In this case, the goal is finding and prevent hazards on the street. Those hazards can take many forms: pedestrians jaywalking, for example, or vehicles moving erratically or at an unsafe speed. The system can identify objects in the roadway, but it can also learn to prioritize living things and predict interactions between vehicles and pedestrians, Curtis said.

In Action

The system breaks down the information coming in from the video analytics into categories: things it's seeing, things it's seeing but doesn't have corresponding data about, and things it's seeing that it can correlate to rules that have been set. The system is built on open-source code called non-axiomatic reasoning system developed by Temple University. The system asks itself questions and then "reasons" to come up with answers, Curtis said.i

Plain Language

The result of all the analytics, categorization and learning is a dense web of text all but unintelligible to human eyes. Over time, though, the system provides information in easy-to-understand terms. In the city street example, it uses the visual information being gathered to draw conclusions about risk factors, dividing the action on the street into high, medium and low risk. Jaywalking, for example, is categorized as a high-risk incident. From there, the system can make recommendations, like adjusting the timing of crosswalks to better accommodate pedestrians.

Where Partners Come In

While Cisco has the brainpower and wherewithal to create and develop the Deep Fusion Reasoning Engine, the company will be relying on partners to bring the system to market. "You know the customers' business best," Curtis said to partners during a keynote at the Partner Connection Week conference. "This could be applied to smart cities, health care, manufacturing -- the opportunities are endless." Borno listed several more potential use cases for the system, including infrastructure performance and security. "This is a great code development opportunity for many of our partners," she said.