Pontimax Technologies LLC announced its Dynamic Meta-Inferencing (DM-I) Agent software, with innovative capabilities for meta-specification-based, generative predicate dynamic reasoning, utilizing hybrid Machine Learning (ML)-based Inferencing and Spatiotemporal Fact Pattern Inferencing.
Ray Keating, Managing Member of Pontimax Technologies, described the DM-I Agent as providing “a capability for “Inferencing Goal Events” to be defined with “Predicate Fact Patterns” (PFP) Meta-Specifications. The Inferencing Goal Event’s specified set of Predicate Fact Patterns is dynamically generated, invoked and evaluated using the DM-I Agent’s Inferencing Engine.” Mr. Keating noted that Predicate Fact Patterns can be either Machine Learning (ML) based or a “Spatiotemporal Fact Pattern”. The former might be an object-recognition ML inferencing; the latter could be Movements-at-a-Location for a specified period. The key, however, is that, unlike the current “stand-alone” use of ML inferences, Pontimax’s DM-I Agent uniquely supports combining multiple inferencing predicates. In fact, there is no specific limit on the number of the Predicate Fact Patterns for the specified Inferencing Goal Event. Moreover, the DM-I Agent is completely driven by the configured Inferencing Goal(s) and their associated Fact Pattern Meta-Specifications, so the DM-I Agent provides a “No-Code” user environment.”
CIO INFLUENCE: Nextira Selected by Ansys Technology Partner Program to Support Customers Implementing Ansys Gateway Powered by AWS
He provided an example from Pontimax’s yrConnected Personal Safety & Wellness system: the ability to infer an Immobility Condition at a Residence Location. He stated, “if your Inferencing Goal is to infer that an individual is immobile at a location within their residence, maybe due to a fall, you could accomplish that by leveraging the DM-I Agent‘s built-in maintaining changes of the individual’s location status. The DM-I Agent has a “Triggered Event” capability to launch Event Predicates from changes in Location State such as the subject transiting to a specific location. To determine if an Immobility condition exists, two Fact Pattern predicates are considered: has the individual been at the current location greater than a user-specified threshold duration? If so, is there minimum or no movements at that current location for the calculated threshold duration? If so, an Immobility Condition can be inferred and text messages sent to a Caregiver and E911“.
“Let’s look at how, in a military domain, the DM-I Agent could provide operators on an infiltration mission with an “Information Advantage”. Assume that there are motion sensors placed along the ingress route and a tasked overhead video sensor platform. During mission planning, an “Anti-Ambush” Inferencing Goal Event with two Predicates is specified to the DM-I Agent: a Motion Occurrences Pattern Predicate and an Object Recognition Predicate, along with sensor data sources and video sensor addresses. The DM-I Agent will apply its built-in Spatiotemporal Pattern recognition abilities to determine if the Motion Occurrences Predicate Fact Pattern has occurred. If so, the “chained” Object Recognition predicate is launched. If the ML recognition probability exceeds the inferencing threshold, the “Anti-Ambush” Goal Event is inferred and its user-specified Consequent Actions (CA) are launched. Besides recording the Event’s Predicate inferencing data, a Threat Alert message would certainly go to the Mission Lead. This military domain Use Case illustrates the ability of Pontimax’s DM-I Agent to provide agile “Sensor-to-Shooter” and “Data-to-Decision” capabilities essential to delivering real tactical Information Advantage.
CIO INFLUENCE: CIO Influence Interview with Pete Lilley, Vice President and GM at Instaclustr
[To share your insights with us, please write to sghosh@martechseries.com]
