Please download the CNSM PDF timetable here:
Resilmesh Project Introduction by Martin Husák, Masaryk University, Czech Republic
ResilMesh is an Innovation Action project funded by the European Union, dedicated to revolutionizing cybersecurity practices. At its core, ResilMesh endeavors to develop a cutting-edge security orchestration and analytics toolset grounded in cyber situational awareness (CSA). This initiative aims to equip organizations with the capabilities needed for real-time defense of essential business functions in an era marked by dispersed, heterogeneous cyber systems.
AI-based networks and network management from Cisco under the hood
Keynote by Martin Diviš and Dominik Soukup, Cisco, Czech Republic
Computer networks have to be flexible enough to connect, monitor, and secure everything we need. Over the past decades, we have phased several trends that added more demands on computer networks and AI is not an exception. AI brings demand for more powerful computational resources that must be interconnected and provides a great opportunity for higher intelligence based on the huge amount of data samples we are getting. During this talk we will demonstrate how Cisco is building and segment networks for AI era, how we use AI methods in network management solutions and why Splunk is now part of Cisco family.
For many years machine learning help optimized resources in network management. It then help to better plan network topologies, bandwidth control, BGP routes, and finally security. Although security defenses is implemented in servers and devices, much of the attacks are simulated and conducted in the network. With the advent of LLMs much technology was dedicated to better explain, to better understand and validate routers configurations, graph analysis, and lately to help the incident responders. However, in security LLMs have a much needed role: to produce better attacks, to produce better detections and to produce better humans.
We believe that using AI to catch criminals is fundamentally different from other AI applications due to a unique combination of constraints, e.g. data privacy, compliance, explainability, class imbalance and first and foremost the fact that we are engaging with an intelligent and evasive adversary. We will share the experience gathered along our journey at Czech Technical University, Cognitive Security, Cisco and Resistant AI, where we have deployed machine learning methods to catch criminals across many domains. Starting with network traffic, progressing through web proxy logs and encrypted traffic to combination of all of the above with large-scale malware analysis. We are now focusing on digital fraud and financial crime detection at Resistant AI. During our talk, we will present 10 hypothesis we have validated across many years and domains.
Recent years have seen remarkable advancements in Artificial Intelligence (AI) and Machine Learning (ML), leading to their widespread adoption across various sectors of society. The field of communication networks is no exception to this trend. As a research community, we are actively exploring innovative strategies that leverage AI/ML techniques to address long-standing or emerging problems such as traffic classification, network diagnosis, and intent-based networking. In this context, programmable networks have emerged as a promising framework for executing low-latency, high-throughput AI/ML computations. In this talk, we will examine the synergistic relationship between AI/ML and network programmability in addressing operation and management challenges. We will explore the progress we have made, the lessons we have learned, and the obstacles we face in developing novel AI/ML-enabled management solutions within programmable networks. Additionally, we will highlight ongoing initiatives and identify opportunities for collaboration within our community to further advance this field.