R&D
AFRL & Intuidex Patents Method and Framework for Pattern of Life Analysis
March 20, 2024 – By Intuidex
The Air Force Research Laboratory (AFRL) and Intuidex collaborated on a patent for a theoretical and technical approach to providing real-time situational awareness with Intuidex’s Pattern of Life and Anomaly Detection capabilities (POL/AD).
Pattern of Life is the pattern of activities of an entity (such as a person, a car, a camera, a search term). The research on POL is considered in two primary questions: how to discover the POL of an entity, and to predict the POL of the same entity. Each question entails two sub-questions: how to find the normal pattern (normal activities) and how to pinpoint the abnormal patterns (abnormal activities, or anomalies).
Intuidex investigated representing the POL of an entity in three ways: (1) as numerical data in a temporal sequence, (2) multi-dimensional numeric data in a geospatial context, and (3) categorical data (which includes entities, relations and events extracted from unstructured text). Defining the POL of an entity then becomes a task to discover patterns in these sequences (such as amount of money, or counts of license plate reads (LPR) from an automated license plate reader) and/or patterns of categorical data (such as person names or the vehicular license plate numbers (LPN)).
Pattern of Life and Anomaly Detection can also employ Intuidex’s proprietary Higher-Order Low-Resource Learning™ (HO-LRL™) technology, which supports the analysis of diverse types of data and identifies latent relationships, providing more accurate classification and mapping of entities observed. HO-LRL™ performs this analysis in real-time and is particularly effective when only small fragments of information are available for modeling.
We are very proud to have this patent on the AFRL Patent Wall in Rome, NY! In addition, as we continue to innovate, Intuidex has also developed an enhanced proprietary approach to POL/AD with broader applications across multiple market sectors.
W4S, SDaaS, R&D
Intuidex awarded AFWERX SBIR Phase II - Picosat Pattern of Life, Space Data as a Service
August 4, 2022 – By Intuidex
Intuidex to research and develop low-cost Space Data as a Service (SDaaS) using flocks of attritable low-size, weight, power, and price (SWAP2) Picosats capable of global persistent intelligence, surveillance, and reconnaissance (ISR) and Earth observation.
Using novel hardware solutions, materials, and manufacturing processes, flocks of Picosats are built and launched into orbit equipped with advanced sensors, including multi-spectral optical systems with resolution approaching one meter / pixel. Onboard real-time automated pattern of life and anomaly detection (POL/AD) using Intuidex’s proprietary higher-order low-resource learning (HO-LRL™) technology provides object detection and other valuable data for use in multiple military and civilian applications.
Objectives:
- Create a dual-use technology / commercial offering – Space Data as a Service (SDaaS) – with multiple commercial and military applications. For example, notify a farmer of unusual soil conditions or warn a water treatment plant that a sewage spill has occurred upstream. These customers need actionable information quickly without requiring a data scientist on staff or complex and costly tasking requirements.
- Create a constellation of Picosats returning data on high-speed connections to give a near-real-time picture of the Earth, cislunar space and even Mars! This will give subscribers data they need improve their processes.
- Provide a solution that takes less time, effort and cost to provision and maintain a comparable contemporary solution.
- Increase analysts’ speed and ease in tasking sensors and detecting, identifying and locating military targets.
- Provide a space-based solution that is viable against near-peer adversaries.
Potential dual-use applications include:
- Earth Observations – University Research, Mining/Resource Management, USGS Studies, Land Use Meteorology;
- Internet of Things – Security Research, System Monitoring;
- Planetary Studies – Observation, Scientific Evaluation
Key Differentiators:
- Low size, weight, power and price (SWaP2)
- Mission Assurance-based Design
- Multiple sensor modalities
- Dynamic mobility
- Resilient Comms
- Pattern of Life and Anomaly Detection (POL/AD)
- Higher-Order Low-Resource Learning (HO-LRL™)
- Model Management
R&D
Intuidex awarded AFRL STTR Phase II – PAI Information Extraction
May 6, 2022 – By Intuidex
Intuidex, Inc. is proud to announce the award of a Small Business Technology Transfer (STTR) Phase II contract from the Air Force Reach Lab. The project will focus on further research and development of Intuidex’s Higher-Order Low-Resource Learning™ (HO-LRL™) technology, focusing on information extraction from Publicly Available Information (PAI) in partnership with Carnegie Mellon University.
During the STTR Phase I effort Intuidex, Inc., in partnership with Carnegie Mellon University, focused on research and development (R&D) of deep-learning-based information extraction techniques that leveraged Intuidex’s Deep Learning-based Named Entity Recognition (DL-NER™) framework for Named Entity Extraction (NER) based on Intuidex’s proprietary HO-LRL™ technology. Phase II will continue this effort with the objective to research, design, and develop information extraction theory, algorithms and techniques for extraction from noisy, user-generated text and related PAI based on R&D of low-resource learning approaches to entity resolution across multiple sources coupled with research into linguistically-informed decoders for deep networks, unsupervised decipherment of noisy text, and nondestructive text normalization.
Anticipated benefits include:
- Significant enhancements to deep learning predictive performance of information extraction and entity resolution in multi-source data in low-resource settings;
- Significant improvements in the accurate extraction of meaningful information from PAI;
- Reduced effort and improved throughput in the exploitation of noisy PAI data; and
- Successful transition into the hands of warfighters realistically within one calendar year of completion of the Phase II effort.
Expected commercial applications include the Watchman for Defense™ (W4D™) platform (for Department of Defense, Intelligence Community and Homeland Security applications), the Watchman for Emergency Management™ (W4EM™) platform (for both public- and private-sector emergency management in near-real-time situational awareness applications), and the Watchman for Law Enforcement™ (W4LE™) platform (for law enforcement and private security concerns).
R&D
Intuidex's Higher-Order Low-Resource Learning™ (HO-LRL™) in Use Case Developed with Serco NA Greatly Improves Predictive Analytics
March 9, 2022 – By Intuidex
Intuidex, Inc. partnered with Serco North America (Serco NA) to develop a use case for predictive analytics in low resource environments leveraging Intuidex’s Higher-Order Low-Resource Learning™ (HO-LRL™) to improve performance over traditional predictive models.
This use case will be presented at the Sixth Annual Workshop on Naval Applications of Machine Learning (NAML), March 22-24, 2022.
Poster Schedule Session 1A, Wednesday March 23, 2022: Rapid Exploitation of Human Language in a Low-Resource Environment by Alex Rojas, Serco North America Inc. See the full schedule here.
Presentation Intro Video – courtesy Serco North America
Problem Statement
Data quality, consistency, and integrity remain a common challenge across artificial intelligence (AI) use cases, including in natural language processing (NLP). Most AI algorithms and machine learning methods rely on a large number of training data examples in order to provide higher accuracy predictions.
To better address predictive analytics in low resource environments, Intuidex and Serco NA tested the use of Intuidex’s proprietary HO-LRL™ technology to predict schedule overruns in the delivery of military assets utilizing condition and maintenance planning data. Currently, teams of personnel within the military are required to produce Availability Duration Scorecard (ADS) predictions.
Testing
Previous modeling of predictive analytics performed by Serco NA using a traditional machine learning algorithm produced a baseline solution which provided better results than the human expert-driven ADS approach (14% versus 12.45% mean absolute percent error) but there was room for improvement.
In this test, NLP features were incorporated into a Support Vector Machine (SVM) model enhanced to use HO-LRL™. The HO-LRL™ technique transforms the data to focus on important discriminators for target conditions.
Findings
Metric | HO-LRL™ | Original Implementation |
Fβ=0.25 / Accuracy | >99% | 86% |
Error / MAPE (Mean Absolute | <1% | 14% |
Training Time | <20 seconds (avg. depends on | 1:300 hrs. ratio |
Training Data | 6,500 records (0.2% of original | 3,158,219 records |
Time to Implement | 2 weeks | Few months when data available |
Conclusions
● Using HO-LRL™ yielded > 99% Fβ=0.25, a 13% increase from the original implementation!
● Training time was greatly reduced to <20 seconds (avg. depends on data set size and resources)!
● The amount of data used was only 0.2 % (6500 records) of the original data (3,158,219 records). The smallest amount of data modeled was only 180 samples, 30 with the overrun condition and 150 with non-overrun. HO-LRL™-enhanced SVM beat standard SVM by several points with >99% confidence!
● Take away: What is remarkable is that this represents a >99.99999% reduction in training data and subsequent training time but resulted in significantly increased accuracy!
Benefits of HO-LRL™
HO-LRL™ is part of Intuidex’s Watchman Analytics Suite from which several products, including Watchman for Defense™ (W4D) are derived. W4D provides multi-INT fusion and alerting, including pattern of life and anomaly detection. HO-LRL™ is a data transformation for machine learning in low-resource settings and supports both generative and discriminative learning, including natural language processing (NLP) with deep learning networks and latent embeddings.
HO-LRL™ can be used in a wide array of use cases and applications and provides numerous benefits, such as a much lower threshold of training data, significantly less training time, and greatly improved predictive performance in low-resource settings including real-time streaming data scenarios.