Lukas Rauch

Lukas Rauch, Matthias Aßenmacher, Denis Huseljic et al.

Deep active learning (DAL) seeks to reduce annotation costs by enabling the model to actively query instance annotations from which it expects to learn the most. Despite extensive research, there is currently no standardized evaluation protocol for transformer-based language models in the field of DAL. Diverse experimental settings lead to difficulties in comparing research and deriving recommendations for practitioners. To tackle this challenge, we propose the ActiveGLAE benchmark, a comprehensive collection of data sets and evaluation guidelines for assessing DAL. Our benchmark aims to facilitate and streamline the evaluation process of novel DAL strategies. Additionally, we provide an extensive overview of current practice in DAL with transformer-based language models. We identify three key challenges - data set selection, model training, and DAL settings - that pose difficulties in comparing query strategies. We establish baseline results through an extensive set of experiments as a reference point for evaluating future work. Based on our findings, we provide guidelines for researchers and practitioners.

Lukas Rauch, Raphael Schwinger, Moritz Wirth et al.

We propose a shift towards end-to-end learning in bird sound monitoring by combining self-supervised (SSL) and deep active learning (DAL). Leveraging transformer models, we aim to bypass traditional spectrogram conversions, enabling direct raw audio processing. ActiveBird2Vec is set to generate high-quality bird sound representations through SSL, potentially accelerating the assessment of environmental changes and decision-making processes for wind farms. Additionally, we seek to utilize the wide variety of bird vocalizations through DAL, reducing the reliance on extensively labeled datasets by human experts. We plan to curate a comprehensive set of tasks through Huggingface Datasets, enhancing future comparability and reproducibility of bioacoustic research. A comparative analysis between various transformer models will be conducted to evaluate their proficiency in bird sound recognition tasks. We aim to accelerate the progression of avian bioacoustic research and contribute to more effective conservation strategies.

Ilyass Moummad, Marius Miron, Lukas Rauch et al.

Audio-to-image retrieval offers an interpretable alternative to audio-only classification for bioacoustic species recognition, but learning aligned audio-image representations is challenging due to the scarcity of paired audio-image data. We propose a simple and data-efficient approach that enables audio-to-image retrieval without any audio-image supervision. Our proposed method uses text as a semantic intermediary: we distill the text embedding space of a pretrained image-text model (BioCLIP-2), which encodes rich visual and taxonomic structure, into a pretrained audio-text model (BioLingual) by fine-tuning its audio encoder with a contrastive objective. This distillation transfers visually grounded semantics into the audio representation, inducing emergent alignment between audio and image embeddings without using images during training. We evaluate the resulting model on multiple bioacoustic benchmarks. The distilled audio encoder preserves audio discriminative power while substantially improving audio-text alignment on focal recordings and soundscape datasets. Most importantly, on the SSW60 benchmark, the proposed approach achieves strong audio-to-image retrieval performance exceeding baselines based on zero-shot model combinations or learned mappings between text embeddings, despite not training on paired audio-image data. These results demonstrate that indirect semantic transfer through text is sufficient to induce meaningful audio-image alignment, providing a practical solution for visually grounded species recognition in data-scarce bioacoustic settings.

Denis Huseljic, Marek Herde, Lukas Rauch et al.

Since training deep neural networks takes significant computational resources, extending the training dataset with new data is difficult, as it typically requires complete retraining. Moreover, specific applications do not allow costly retraining due to time or computational constraints. We address this issue by proposing a novel Bayesian update method for deep neural networks by using a last-layer Laplace approximation. Concretely, we leverage second-order optimization techniques on the Gaussian posterior distribution of a Laplace approximation, computing the inverse Hessian matrix in closed form. This way, our method allows for fast and effective updates upon the arrival of new data in a stationary setting. A large-scale evaluation study across different data modalities confirms that our updates are a fast and competitive alternative to costly retraining. Furthermore, we demonstrate its applicability in a deep active learning scenario by using our update to improve existing selection strategies.

Jens Decke, Christian Gruhl, Lukas Rauch et al.

In this experience report, we apply deep active learning to the field of design optimization to reduce the number of computationally expensive numerical simulations. We are interested in optimizing the design of structural components, where the shape is described by a set of parameters. If we can predict the performance based on these parameters and consider only the promising candidates for simulation, there is an enormous potential for saving computing power. We present two selection strategies for self-optimization to reduce the computational cost in multi-objective design optimization problems. Our proposed methodology provides an intuitive approach that is easy to apply, offers significant improvements over random sampling, and circumvents the need for uncertainty estimation. We evaluate our strategies on a large dataset from the domain of fluid dynamics and introduce two new evaluation metrics to determine the model's performance. Findings from our evaluation highlights the effectiveness of our selection strategies in accelerating design optimization. We believe that the introduced method is easily transferable to other self-optimization problems.

Marek Herde, Denis Huseljic, Lukas Rauch et al.

Human annotators typically provide annotated data for training machine learning models, such as neural networks. Yet, human annotations are subject to noise, impairing generalization performances. Methodological research on approaches counteracting noisy annotations requires corresponding datasets for a meaningful empirical evaluation. Consequently, we introduce a novel benchmark dataset, dopanim, consisting of about 15,750 animal images of 15 classes with ground truth labels. For approximately 10,500 of these images, 20 humans provided over 52,000 annotations with an accuracy of circa 67%. Its key attributes include (1) the challenging task of classifying doppelganger animals, (2) human-estimated likelihoods as annotations, and (3) annotator metadata. We benchmark well-known multi-annotator learning approaches using seven variants of this dataset and outline further evaluation use cases such as learning beyond hard class labels and active learning. Our dataset and a comprehensive codebase are publicly available to emulate the data collection process and to reproduce all empirical results.

Denis Huseljic, Paul Hahn, Marek Herde et al.

Deep active learning (AL) seeks to minimize the annotation costs for training deep neural networks. BAIT, a recently proposed AL strategy based on the Fisher Information, has demonstrated impressive performance across various datasets. However, BAIT's high computational and memory requirements hinder its applicability on large-scale classification tasks, resulting in current research neglecting BAIT in their evaluation. This paper introduces two methods to enhance BAIT's computational efficiency and scalability. Notably, we significantly reduce its time complexity by approximating the Fisher Information. In particular, we adapt the original formulation by i) taking the expectation over the most probable classes, and ii) constructing a binary classification task, leading to an alternative likelihood for gradient computations. Consequently, this allows the efficient use of BAIT on large-scale datasets, including ImageNet. Our unified and comprehensive evaluation across a variety of datasets demonstrates that our approximations achieve strong performance with considerably reduced time complexity. Furthermore, we provide an extensive open-source toolbox that implements recent state-of-the-art AL strategies, available at https://github.com/dhuseljic/dal-toolbox.

Houtan Ghaffari, Lukas Rauch, Christoph Scholz et al.

Probing is widely adopted in computer vision to faithfully evaluate self-supervised learning (SSL) embeddings, as fine-tuning may misrepresent their inherent quality. In contrast, audio SSL models still rely on fine-tuning because simple probing fails to unlock their full potential and alters their rankings when competing for SOTA on AudioSet. Hence, a robust and efficient probing mechanism is required to guide the trajectory of audio SSL towards reliable and reproducible methods. We introduce Convex Gated Probing (CGP), a prototype-based method that drastically closes the gap between fine-tuning and probing in audio. CGP efficiently utilizes all frozen layers via a gating mechanism and exposes the location of latent task-relevant information. Guided by CGP, we rework the entire SSL pipeline of current SOTA audio models that use legacy implementations of prior SSL methods. By refining data preprocessing, model architecture, and pre-training recipe, we introduce Better Audio Transformer (BAT), and establish new SOTA on audio benchmarks.

Raphael Schwinger, Ben McEwen, Vincent S. Kather et al.

Passive acoustic monitoring enables large-scale biodiversity assessment, but reliable classification of bioacoustic sounds requires not only high accuracy but also well-calibrated uncertainty estimates to ground decision-making. In bioacoustics, calibration is challenged by overlapping vocalisations, long-tailed species distributions, and distribution shifts between training and deployment data. The calibration of multi-label deep learning classifiers within the domain of bioacoustics has not yet been assessed. We systematically benchmark the calibration of four state-of-the-art multi-label bird sound classifiers on the BirdSet benchmark, evaluating both global, per-dataset and per-class calibration using threshold-free calibration metrics (ECE, MCS) alongside discrimination metrics (cmAP). Model calibration varies significantly across datasets and classes. While Perch v2 and ConvNeXt$_{BS}$ show better global calibration, results vary between datasets. Both models indicate consistent underconfidence, while AudioProtoPNet and BirdMAE are mostly overconfident. Surprisingly, calibration seems to be better for less frequent classes. Using simple post hoc calibration methods we demonstrate a straightforward way to improve calibration. A small labelled calibration set is sufficient to significantly improve calibration with Platt scaling, while global calibration parameters suffer from dataset variability. Our findings highlight the importance of evaluating and improving uncertainty calibration in bioacoustic classifiers.

René Heinrich, Lukas Rauch, Bernhard Sick et al.

Deep learning models have significantly advanced acoustic bird monitoring by being able to recognize numerous bird species based on their vocalizations. However, traditional deep learning models are black boxes that provide no insight into their underlying computations, limiting their usefulness to ornithologists and machine learning engineers. Explainable models could facilitate debugging, knowledge discovery, trust, and interdisciplinary collaboration. This study introduces AudioProtoPNet, an adaptation of the Prototypical Part Network (ProtoPNet) for multi-label bird sound classification. It is an inherently interpretable model that uses a ConvNeXt backbone to extract embeddings, with the classification layer replaced by a prototype learning classifier trained on these embeddings. The classifier learns prototypical patterns of each bird species' vocalizations from spectrograms of training instances. During inference, audio recordings are classified by comparing them to the learned prototypes in the embedding space, providing explanations for the model's decisions and insights into the most informative embeddings of each bird species. The model was trained on the BirdSet training dataset, which consists of 9,734 bird species and over 6,800 hours of recordings. Its performance was evaluated on the seven test datasets of BirdSet, covering different geographical regions. AudioProtoPNet outperformed the state-of-the-art model Perch, achieving an average AUROC of 0.90 and a cmAP of 0.42, with relative improvements of 7.1% and 16.7% over Perch, respectively. These results demonstrate that even for the challenging task of multi-label bird sound classification, it is possible to develop powerful yet inherently interpretable deep learning models that provide valuable insights for ornithologists and machine learning engineers.

Houtan Ghaffari, Lukas Rauch, Paul Devos

Many bioacoustics, neuroscience, and linguistics research utilize birdsongs as proxy models to acquire knowledge in diverse areas. Developing models generally requires precisely annotated data at the level of syllables. Hence, automated and data-efficient methods that reduce annotation costs are in demand. This work presents a lightweight, yet performant neural network architecture for birdsong annotation called Residual-MLP-RNN. Then, it presents a robust three-stage training pipeline for developing reliable deep birdsong syllable detectors with minimal expert labor. The first stage is self-supervised learning from unlabeled data. Two of the most successful pretraining paradigms are explored, namely, masked prediction and online clustering. The second stage is supervised training with effective data augmentations to create a robust model for frame-level syllable detection. The third stage is semi-supervised post-training, which leverages the unlabeled data again. However, unlike the initial phase, this time it is aligned with the downstream task. The performance of this data-efficient approach is demonstrated for the complex song of the Canary in extreme label-scarcity scenarios. Canary has one of the most difficult songs to annotate, which implicitly validates the method for other birds. Finally, the potential of self-supervised embeddings is assessed for linear probing and unsupervised birdsong analysis.

Lukas Rauch, Raphael Schwinger, Moritz Wirth et al.

Deep learning (DL) has greatly advanced audio classification, yet the field is limited by the scarcity of large-scale benchmark datasets that have propelled progress in other domains. While AudioSet is a pivotal step to bridge this gap as a universal-domain dataset, its restricted accessibility and limited range of evaluation use cases challenge its role as the sole resource. Therefore, we introduce BirdSet, a large-scale benchmark dataset for audio classification focusing on avian bioacoustics. BirdSet surpasses AudioSet with over 6,800 recording hours ($\uparrow\!17\%$) from nearly 10,000 classes ($\uparrow\!18\times$) for training and more than 400 hours ($\uparrow\!7\times$) across eight strongly labeled evaluation datasets. It serves as a versatile resource for use cases such as multi-label classification, covariate shift or self-supervised learning. We benchmark six well-known DL models in multi-label classification across three distinct training scenarios and outline further evaluation use cases in audio classification. We host our dataset on Hugging Face for easy accessibility and offer an extensive codebase to reproduce our results.

Lukas Rauch, René Heinrich, Ilyass Moummad et al.

Masked Autoencoders (MAEs) learn rich semantic representations in audio classification through an efficient self-supervised reconstruction task. However, general-purpose models fail to generalize well when applied directly to fine-grained audio domains. Specifically, bird-sound classification requires distinguishing subtle inter-species differences and managing high intra-species acoustic variability, revealing the performance limitations of general-domain Audio-MAEs. This work demonstrates that bridging this domain gap domain gap requires full-pipeline adaptation, not just domain-specific pretraining data. We systematically revisit and adapt the pretraining recipe, fine-tuning methods, and frozen feature utilization to bird sounds using BirdSet, a large-scale bioacoustic dataset comparable to AudioSet. Our resulting Bird-MAE achieves new state-of-the-art results in BirdSet's multi-label classification benchmark. Additionally, we introduce the parameter-efficient prototypical probing, enhancing the utility of frozen MAE representations and closely approaching fine-tuning performance in low-resource settings. Bird-MAE's prototypical probes outperform linear probing by up to 37 percentage points in mean average precision and narrow the gap to fine-tuning across BirdSet downstream tasks. Bird-MAE also demonstrates robust few-shot capabilities with prototypical probing in our newly established few-shot benchmark on BirdSet, highlighting the potential of tailored self-supervised learning pipelines for fine-grained audio domains.

Ilyass Moummad, Kawtar Zaher, Lukas Rauch et al.

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 expensive, scenario-specific training. In this work, we introduce Hashing-Baseline, a strong training-free hashing method leveraging powerful pretrained encoders that produce rich pretrained embeddings. We revisit classical, training-free hashing techniques: principal component analysis, random orthogonal projection, and threshold binarization, to produce a strong baseline for hashing. Our approach combines these techniques with frozen embeddings from state-of-the-art vision and audio encoders to yield competitive retrieval performance without any additional learning or fine-tuning. To demonstrate the generality and effectiveness of this approach, we evaluate it on standard image retrieval benchmarks as well as a newly introduced benchmark for audio hashing.

Raphael Schwinger, Paria Vali Zadeh, Lukas Rauch et al.

Automated bioacoustic analysis is essential for biodiversity monitoring and conservation, requiring advanced deep learning models that can adapt to diverse bioacoustic tasks. This article presents a comprehensive review of large-scale pretrained bioacoustic foundation models and systematically investigates their transferability across multiple bioacoustic classification tasks. We overview bioacoustic representation learning including major pretraining data sources and benchmarks. On this basis, we review bioacoustic foundation models by thoroughly analysing design decisions such as model architecture, pretraining scheme, and training paradigm. Additionally, we evaluate selected foundation models on classification tasks from the BEANS and BirdSet benchmarks, comparing the generalisability of learned representations under both linear and attentive probing strategies. Our comprehensive experimental analysis reveals that BirdMAE, trained on large-scale bird song data with a self-supervised objective, achieves the best performance on the BirdSet benchmark. On BEANS, BEATs$_{NLM}$, the extracted encoder of the NatureLM-audio large audio model, is slightly better. Both transformer-based models require attentive probing to extract the full performance of their representations. ConvNext$_{BS}$ and Perch models trained with supervision on large-scale bird song data remain competitive for passive acoustic monitoring classification tasks of BirdSet in linear probing settings. Training a new linear classifier has clear advantages over evaluating these models without further training. While on BEANS, the baseline model BEATs trained with self-supervision on AudioSet outperforms bird-specific models when evaluated with attentive probing. These findings provide valuable guidance for practitioners selecting appropriate models to adapt them to new bioacoustic classification tasks via probing.

Denis Huseljic, Marek Herde, Lukas Rauch et al.

Existing active learning (AL) strategies capture fundamentally different notions of data value, e.g., uncertainty or representativeness. Consequently, the effectiveness of strategies can vary substantially across datasets, models, and even AL cycles. Committing to a single strategy risks suboptimal performance, as no single strategy dominates throughout the entire AL process. We introduce REFINE, an ensemble AL method that combines multiple strategies without knowing in advance which will perform best. In each AL cycle, REFINE operates in two stages: (1) Progressive filtering iteratively refines the unlabeled pool by considering an ensemble of AL strategies, retaining promising candidates capturing different notions of value. (2) Coverage-based selection then chooses a final batch from this refined pool, ensuring all previously identified notions of value are accounted for. Extensive experiments across 6 classification datasets and 3 foundation models show that REFINE consistently outperforms individual strategies and existing ensemble methods. Notably, progressive filtering serves as a powerful preprocessing step that improves the performance of any individual AL strategy applied to the refined pool, which we demonstrate on an audio spectrogram classification use case. Finally, the ensemble of REFINE can be easily extended with upcoming state-of-the-art AL strategies.

Lukas Rauch, René Heinrich, Houtan Ghaffari et al.

Although probing frozen models has become a standard evaluation paradigm, self-supervised learning in audio defaults to fine-tuning. A key reason is that global pooling creates an information bottleneck causing linear probes to misrepresent the embedding quality: The $\texttt{cls}$-token discards crucial token information about dispersed, localized events in multi-label audio. This weakness is rooted in the mismatch between the pretraining objective (operating globally) and the downstream task (localized events). Across a comprehensive benchmark of 13 datasets and 6 spectrogram-based encoders, we first investigate the global pooling bottleneck. We then introduce binarized prototypical probes: a lightweight and simple pooling method that learns prototypes to perform class-wise information aggregation. Despite its simplicity, our method notably outperforms linear and attentive probing. Our work establishes probing as a competitive and efficient paradigm for evaluating audio SSL models, challenging the reliance on costly fine-tuning.

René Heinrich, Lukas Rauch, Bernhard Sick et al.

Adversarial training is a promising strategy for enhancing model robustness against adversarial attacks. However, its impact on generalization under substantial data distribution shifts in audio classification remains largely unexplored. To address this gap, this work investigates how different adversarial training strategies improve generalization performance and adversarial robustness in audio classification. The study focuses on two model architectures: a conventional convolutional neural network (ConvNeXt) and an inherently interpretable prototype-based model (AudioProtoPNet). The approach is evaluated using a challenging bird sound classification benchmark. This benchmark is characterized by pronounced distribution shifts between training and test data due to varying environmental conditions and recording methods, a common real-world challenge. The investigation explores two adversarial training strategies: one based on output-space attacks that maximize the classification loss function, and another based on embedding-space attacks designed to maximize embedding dissimilarity. These attack types are also used for robustness evaluation. Additionally, for AudioProtoPNet, the study assesses the stability of its learned prototypes under targeted embedding-space attacks. Results show that adversarial training, particularly using output-space attacks, improves clean test data performance by an average of 10.5% relative and simultaneously strengthens the adversarial robustness of the models. These findings, although derived from the bird sound domain, suggest that adversarial training holds potential to enhance robustness against both strong distribution shifts and adversarial attacks in challenging audio classification settings.

Lukas Rauch, Moritz Wirth, Denis Huseljic et al.

The advent of large language models (LLMs) capable of producing general-purpose representations lets us revisit the practicality of deep active learning (AL): By leveraging frozen LLM embeddings, we can mitigate the computational costs of iteratively fine-tuning large backbones. This study establishes a benchmark and systematically investigates the influence of LLM embedding quality on query strategies in deep AL. We employ five top-performing models from the massive text embedding benchmark (MTEB) leaderboard and two baselines for ten diverse text classification tasks. Our findings reveal key insights: First, initializing the labeled pool using diversity-based sampling synergizes with high-quality embeddings, boosting performance in early AL iterations. Second, the choice of the optimal query strategy is sensitive to embedding quality. While the computationally inexpensive Margin sampling can achieve performance spikes on specific datasets, we find that strategies like Badge exhibit greater robustness across tasks. Importantly, their effectiveness is often enhanced when paired with higher-quality embeddings. Our results emphasize the need for context-specific evaluation of AL strategies, as performance heavily depends on embedding quality and the target task.

Lukas Rauch, Thomas Braml

The significant effort required to annotate data for new training datasets hinders computer vision research and machine learning in the construction industry. This work explores adapting standard datasets and the latest transformer model architectures for point cloud semantic segmentation in the context of shell construction sites. Unlike common approaches focused on object segmentation of building interiors and furniture, this study addressed the challenges of segmenting complex structural components in Architecture, Engineering, and Construction (AEC). We establish a baseline through supervised training and a custom validation dataset, evaluate the cross-domain inference with large-scale indoor datasets, and utilize transfer learning to maximize segmentation performance with minimal new data. The findings indicate that with minimal fine-tuning, pre-trained transformer architectures offer an effective strategy for building component segmentation. Our results are promising for automating the annotation of new, previously unseen data when creating larger training resources and for the segmentation of frequently recurring objects.

Lukas Rauch, Denis Huseljic, Moritz Wirth et al.

Passive acoustic monitoring (PAM) in avian bioacoustics enables cost-effective and extensive data collection with minimal disruption to natural habitats. Despite advancements in computational avian bioacoustics, deep learning models continue to encounter challenges in adapting to diverse environments in practical PAM scenarios. This is primarily due to the scarcity of annotations, which requires labor-intensive efforts from human experts. Active learning (AL) reduces annotation cost and speed ups adaption to diverse scenarios by querying the most informative instances for labeling. This paper outlines a deep AL approach, introduces key challenges, and conducts a small-scale pilot study.