Pandora Camille (AI and ML Module)
We developed Pandora Camille to introduce next generation cognitive approach to cybersecurity.
Artificial Intelligence and Machine Learning Module was built into the system overall architecture to learn customer specific cybersecurity landscape (devices, data, behavioral patterns). Thanks to Pandora hyper-scalability Camille is learning rapidly and eliminates the noise of false positive - it is able to embrace input of up to 20 billion sessions daily coming from millions of end devices communicating massive data online. Camille is using:
- ANN Models – tensorflow/GPU
- Gradient Boosting – XGBoost/GPU(Pascal)
- Other not computationally intensive – scikit learn
Pandora Camille is based on signature-less technology and is working its way into the future to provide protection from known and yet not be detected cybersecurity threats.
- Signature-less technology
- End-to-end anomaly analysis for both inbound and outbound traffic
- Connectivity agnostictelecommunication operator independent, 5G ready
- Endpoint independency focus on behavior and data patterns for any device in mobile network
- Cognitive and industry focusedable to learn and adapt to specific customer cybersecurity landscape
- Hyperscalableprocesses up to 20 billion sessions daily, designed in container approach
- Bi-modular and simultaneous data processingreal-time and batch workloads
- Cloud native with Microsoft Azure Platform
- Private cloud readydedicated hosted solution in customer private cloud