Manuel Milling

Manuel Milling

he/him

Deep Learning Researcher and Teacher

Publications

For a comprehensive list of publications, see my Google Scholar. Below is a selected list of papers, which particularly represent my research interests or are particular milestones in my academic journey.

  • Milling, Manuel, et al. “Bringing the discussion of minima sharpness to the audio domain: A filter-normalised evaluation for acoustic scene classification.” ICASSP (2024). (Link): Introduction of a new quantification of sharpness and explorative analysis of the connection between sharpness and generalisation in computer audition, as well as an analysis of hyperparameter influence on sharpness.

  • Milling, Manuel, et al. “Leveraging Sample Difficulty in Computer Audition Analysis.” IEEE Access (2026). (Link): Exploration of curriculum learning-derived sample difficulty measures as an data analysis tool for computer audition to identify mislabelled and noisy recordings, to show distributional differences between easy and difficult samples.

  • Rampp, Simon, and Milling, Manuel, et al. “Does the Definition of Difficulty Matter? Scoring Functions and their Role for Curriculum Learning, accepted at IJCNN (2026).: Systematic analysis of properties, robustness, and alignment between scoring functions for sample difficulty estimation, as well as their role in curriculum learning in computer vision and computer audition.

  • Milling, Manuel, et al. “A frequency analysis of filterbank initialisation and noise augmentation for LEAF.”, Scientific Reports (2026) (Link): Analysis of the learning behaviour of the learnable frontend (LEAF), showing a lack of adaptability in the centre frequency and bandwidth of the filterbank and a strong dependence on initialisation bias, even in audio recordings with very controlled frequency content.

  • Milling, Manuel, et al. “Is Speech the New Blood? Recent Progress in AI-Based Disease Detection From Audio in a Nutshell”, Frontiers in Digital Health (2022) (Link): Mini Review on speech-based disease detection studies utilising machine learning techniques, providing an overview of common methodological approaches, investigated diseases and a position on the future role of speech-based AI as a tool supporting doctors in their decision making.

  • Milling, Manuel, et al. *Evaluating the Impact of Voice Activity Detection on Speech Emotion Recognition for Autistic Children, Frontiers in Computer Science (2022) (Link): Comparison of general and custom (child-specific) voice activity detection and their role on downstream speech emotion recognition on a new dataset focusing on children with autism interacting in an educational setting with a robot (my first first-author paper).