ADF - anomaly detection framework
Applicable across a wide range of problems
We are able to detect anomalous user behavior in bank transactions, insurance claims, shopping lists, telecommunication data as well as in computer network, personal, or accounting data. The framework is very well suited for predictive maintenance use cases, monitoring production lines and complex manufacturing and assembly systems through numerous deployed sensors.
Our framework is able to digest small and large datasets
from which it constructs a wide variety of time series and data clusters. A rich algorithmic library utilizing a number of statistical machine learning methods iterates the time series and clusters to find significant outliers which are then reported to the user. The user can give feedback to the system and thus regulate the number of anomalies received for analysis as well as teach the system what is an anomaly and what is not.
Key Performance Indicators
Insight into the performance/security of your systems
Detection of novel types of frauds and threats
Near real-time report of anomalies
As OKSystem we want to constantly innovate and offer our customers a modern product. Together with the Blindspot team, we explored the path to using advanced analytics and extending our solution with elements of artificial intelligence, enabling customers to offer unique insights into the data they already have, simplify and partially automate and modernize their personnel agenda.