activity
20242026
collaborators

9 papers

cs.SD2026

Unmute the Patch Tokens: Rethinking Probing in Multi-Label Audio Classification

Lukas Rauch, René Heinrich, Houtan Ghaffari +4

Although probing frozen models has become a standard evaluation paradigm, self-supervised learning in audio defaults to fine-tuning when pursuing state-of-the-art on AudioSet. A ke…

cs.LG2026

Cleaning the Pool: Progressive Filtering of Unlabeled Pools in Deep Active Learning

Denis Huseljic, Marek Herde, Lukas Rauch +2

Existing active learning (AL) strategies capture fundamentally different notions of data value, e.g., uncertainty or representativeness. Consequently, the effectiveness of strategi…

cs.LG2026

Efficient Bayesian Updates for Deep Active Learning via Laplace Approximations

Denis Huseljic, Marek Herde, Lukas Rauch +5

Deep active learning (AL) selects batches of instances for annotation to avoid retraining deep neural networks (DNNs) after each new label. Employing a naive top- selection can…

cs.LG2026

Hashing-Baseline: Rethinking Hashing in the Age of Pretrained Models

Ilyass Moummad, Kawtar Zaher, Lukas Rauch +1

Information retrieval with compact binary embeddings, also referred to as hashing, is crucial for scalable fast search applications, yet state-of-the-art hashing methods require ex…

cs.LG2025

Can Masked Autoencoders Also Listen to Birds?

Lukas Rauch, René Heinrich, Ilyass Moummad +3

Masked Autoencoders (MAEs) learn rich semantic representations in audio classification through an efficient self-supervised reconstruction task. However, general-purpose models fai…

cs.LG2025

Adversarial Training Improves Generalization Under Distribution Shifts in Bird Sound Classification

René Heinrich, René Heinrich, Lukas Rauch +5

Adversarial training is a promising strategy for enhancing robustness against adversarial attacks, but its impact on generalization under substantial distribution shifts in audio c…