Go to JKU Homepage
{{ item.AUTOREN_ZITAT }}:
{{ item.TITEL }}{{ zitatInLang(item) }}
University Assistant
- Science Park 3 - 4th floor - S3 435
- +43 732 2468 4711
- florian.schmid(at)jku.at
Research Interests
- Efficient Network Architectures for Audio Classification, Acoustic Scene Classification and Audio Tagging
- Cross-Model Knowledge Distillation
- General Purpose Audio Representations
- Domain Generalization
Teaching
- UE Artificial Intelligence (WS2022)
- UE Probabilistic Models (WS2022)
Published Work
-
Current Work under review: "Efficient Large-Scale Audio Tagging Via Transformer-To-CNN Knowledge Distillation", opens an external URL in a new window
-
Paper published in the proceedings of the DCASE2022 Workshop: "Knowledge Distillation From Transformers for Low-Complexity Acoustic Scene Classification", opens an external URL in a new window
- Top rank in DCASE 2022 "Low-Complexity Acoustic Scene Classification" Challenge, opens an external URL in a new window and corresponding report "Distilling Knowledge for Low-Complexity Convolutional Neural Networks from a Patchout Audio Transformer, opens an external URL in a new window"
- Master Thesis: "Using Domain Adaptation to Counter Distribution Shift Between Training and Application Domain, opens an external URL in a new window" and Domain Adaptation Framework, opens an external URL in a new window
