Für das Paper ?Fitting the Puzzle: Towards Source Traffic Modeling for Mobile Instant Messaging" ist Prof. Dr. Michael Seufert, Lehrstuhl für Vernetzte Eingebettete Systeme und Kommunikationssysteme, Fakult?t für Angewandte Informatik an der Universit?t Augsburg, gemeinsam mit einem Team der Julius-Maximilians-Universit?t Würzburg auf der 15. Internationalen “Conference on Network of the Future” mit dem Best-Paper-Award ausgezeichnet worden.

Today our paper “Agree to Disagree: Exploring Consensus of XAI Methods for ML-based NIDS” was presented at the 1st Workshop on Network Security Operations (NecSecOr). This paper examines the effectiveness and consensus of various explainable AI (XAI) methods in enhancing the interpretability of machine learning-based Network Intrusion Detection Systems (ML-NIDS), finding that while some methods align closely, others diverge, underscoring the need for careful selection to build trust in real-world cybersecurity applications.

Today our paper “Certainly Uncertain: Demystifying ML Uncertainty for Active Learning in Network Monitoring Tasks” got presented at the 20th International Conference on Network and Service Management (CSNM). This paper explores the use of Active Learning (AL) to enhance Machine Learning (ML) models in network monitoring by incorporating expert input, aiming to increase model trust, adaptability, and performance, with a comprehensive evaluation of uncertainty-based AL approaches across various datasets and scenarios.

Our latest research paper in ACM TOMM focuses on how video streaming systems can better utilize available bandwidth to provide users with an improved Quality of Experience (QoE).

Our paper “(Not) The Sum of Its Parts: Relating Individual Video and Browsing Stimuli to Web Session QoE” got presented at the 16th International Conference on Quality of Multimedia Experience (QoMEX). This paper investigates the Quality of Experience (QoE) in web sessions that combine both web browsing and video streaming stimuli, addressing the gap in understanding session-level QoE and proposing models to estimate it based on individual stimuli.

Unser Paper ?QoEXplainer: Mediating Explainable Quality of Experience Models with Large Language Models“ wurde auf der 16th International Conference on Quality of Multimedia Experience (QoMEX) vorgestellt. Das Papier stellt QoEXplainer vor, ein Dashboard, das gro?e Sprachmodelle und die Verwendung von Mediatoren verwendet, um erkl?rbare, datengesteuerte Quality of Experience (QoE) Modelle zu veranschaulichen und den Benutzern zu helfen, die Beziehungen zwischen den Modellen durch eine interaktive Chatbot-Schnittstelle zu verstehen.

Our paper “Sitting, Chatting, Waiting: Influence of Loading Times on Mobile Instant Messaging QoE” got presented at the 16th International Conference on Quality of Multimedia Experience (QoMEX). The paper examines the relationship between loading times and user experience (QoE) in mobile instant messaging applications and shows that longer loading times reduce user acceptance and satisfaction, although they do not directly influence QoE ratings.

Prof. Dr. Michael Seufert, Inhaber des Lehrstuhls für Vernetzte Eingebettete Systeme und Kommunikationssysteme, hat den diesj?hrigen CNOM Young Professional Award des Institute of Electrical and Electronics Engineers Communications Society Technical Committee on Network Operation and Management gewonnen.

?

Our paper “HALIDS: a Hardware-Assisted Machine Learning IDS for in-Network Monitoring” was published in the 8th Network Traffic Measurement and Analysis (TMA) Conference.?The paper presents HALIDS, a prototype of a Machine Learning-driven Intrusion Detection System that enables network devices to autonomously make security decisions using in-band and off-band traffic analysis, ultimately aiming to enhance network security through faster processing and intelligent decision-making.

Our paper “The Missing Link in Network Intrusion Detection: Taking AI/ML Research Efforts to Users” was published in IEEE Access. The paper focuses on the challenges faced in adopting Artificial Intelligence (AI) and Machine Learning (ML) within Intrusion Detection Systems (IDS). It identifies barriers to implementation, such as the lack of explainability, usability, and privacy considerations that hinder trust among non-expert users. The authors employ a user-centric approach by examining IDS research through the lens of various stakeholders, deriving realistic personas, and proposing design guidelines and hypotheses to enhance practical adoption of AI/ML-based IDS solutions.

Prof. Dr. Michael Seufert was invited by Deutschlandfunk to talk about our new system for real-time evaluation of the quality of Internet data streams. The interview appeared in the program “Forschung aktuell - Computer und Kommunikation” and was broadcast on May 11, 2024.

Our latest paper is a good starting point for newcomers to the fields of artificial intelligence and communication networks. It provides a comprehensive overview for researchers investigating the application of machine learning to optimize communication networks and the use of these networks to improve training processes in machine learning.