Experiments

with

Graphs

Discover our growing inventory of prototypes and pilots, which showcase graphs' critical importance and versatility in representing reality.

 

Our collaborations focus on diverse aspects of tangible physical systems and abstract conceptual frameworks, targeting real-life use cases and applications of Graph Theory, Knowledge Graphs, and Graph Neural Networks.

Developing RAG applications with Knowledge Graphs

Cybersecurity Knowledge Graphs with RAG: Discovering CSF v2.0 Compliance Use-Cases

Brain Networks - Applications in Graph Theory

EEG Biomarkers - Graph Algorithms supporting the detection and measurement of brain states.

Threat Detection with Graph Neural Networks

A GNN based framework for Kubernetes Security Agents leveraging MITRE Att&ck

LLM with Knowledge Graphs (KG) - RAGs

Exploring Knowledge Graph (KG) supported Retrieval Augmented Generation (RAG)

Cybersecurity use cases and POC: Compliance Assistant leveraging concept associations across frameworks targeting NIST CSF versions 2.0 and 1.1. to implement RAG for generating personalized assessment questionnaires.

Developing Corpus with LLMs

GenAI/LLM support to build large corpus of self-assessment questionnaires and cross-standard, cross-version mapping documentation leveraging publicly available Open-Source cybersecurity resources.

Building a Cybersecurity Knowledge Graph

A framework/standard-agnostic, bottom-up approach to identify a semantic network of fundamental notions/concepts, and the relations that capture a multitude of ways they associate and affect each other.

Graph Theory and Neural Pathways

EEG Pain Biomarkers

Exploring Chronic and Acute Pain Pathways: A minimally effective way of measuring the perception of pain, relying on non-invasive EEG recordings, and a computational model that can take advantage of the Category Theory (CT).

Graph Neural Net (GNN) Inference

 

OSS Agents, Models & Intelligence 

Continuous processing pipelines for OSINT feeds, real-time telemetry, and emulated threat states, to keep GNN models learning and adapting to changing landscape:

  • Consolidated Feed Crawlers (OSINT, Vulnerability Databases)
  • Ethereal (K8S) Agents equipped with On-Line ML, real-time processing pipelines, learning from detected attack patterns
  • Secure, cloud-agnostic  GNN Agents with inference capabilities targeting core security scenarios