JHJulia Herbinger

Julia Herbinger

*This is an algorithmically generated author record based on publicly available information.

Current Organizations / Institutes

The organizations/institutes from the scholar's latest article

Previous Organizations / Institutes

The organizations / institutes from the scholar's other articles in the past ten years


Julia Herbinger has a total of 26 co-authors. Here are the top co-authors by number of shared publications:


Latest Publications

  • Open Access

    Predicting the seed microbiome using phylogeny-driven machine learning

    by Julia Herbinger, Dinesh Kumar Ramakrishnan, Jannik Reißfelder, Majharulislam Babor, Marina Höhne, Ahmed Abdelfattah

    Background: The composition of the seed-associated bacterial microbiome can reflect host evolutionary relationships, a pattern consistent with phylosymbiosis. While machine learning offers new opportunities to predict microbial community composition, existing models often require prior microbial profiles or environmental variables, limiting their application to unsampled hosts. Here, we tested whether plant nuclear internal transcribed spacer (ITS) sequences, used as a marker of host relatedness, can predict species-level seed-associated bacterial communities using 16S rRNA data from 61 plant species. Results: We introduced customized machine learning models that use sequence-based Hamming distances to capture plant host relatedness. Among the tested models, the Hamming Distance-based k-Nearest Neighbor model (HD-KNN) achieved the highest overall predictive accuracy, yielding an average Jensen-Shannon divergence (JSD) of 0.276 between observed and predicted microbiome profiles. HD-KNN performed particularly well within densely sampled host groups, including Brassicaceae and Poaceae , where closely related reference species were available. In contrast, Hamming Distance-based Gaussian Process Regression (HD-GPR) showed slightly better performance for phylogenetically isolated species, suggesting that model performance depends on host representation within the training dataset. Conclusions: Our framework demonstrates that plant nuclear ITS-derived host relatedness carries a partial predictive signal for seed-associated bacterial microbiome composition. These results provide a foundation for low-input predictive modelling of seed-associated bacteria and may help prioritise microbiome predictions for unsampled plant species when closely related reference species are available. However, our conclusions are strictly limited to seed-associated bacterial communities and should not be directly generalized to fungal communities or other plant compartments, such as the rhizosphere or phyllosphere, which may be shaped by different environmental filtering mechanisms.

  • Leveraging Model-Based Trees as Interpretable Surrogate Models for Model Distillation

    by Julia Herbinger, Susanne Dandl, Fiona K. Ewald, Sofia Loibl, Giuseppe Casalicchio


  • Open Access

    Stratiform and Convective Rain Classification Using Machine Learning Models and Micro Rain Radar

    by Wael Ghada, Enric Casellas, Julia Herbinger, Albert Garcia-Benadí, Ludwig Bothmann, Nicole Estrella, Joan Bech, Annette Menzel

    Rain type classification into convective and stratiform is an essential step required to improve quantitative precipitation estimations by remote sensing instruments. Previous studies with Micro Rain Radar (MRR) measurements and subjective rules have been performed to classify rain events. However, automating this process by using machine learning (ML) models provides the advantages of fast and reliable classification with the possibility to classify rain minute by minute. A total of 20,979 min of rain data measured by an MRR at Das in northeast Spain were used to build seven types of ML models for stratiform and convective rain type classification. The proposed classification models use a set of 22 parameters that summarize the reflectivity, the Doppler velocity, and the spectral width (SW) above and below the so-called separation level (SL). This level is defined as the level with the highest increase in Doppler velocity and corresponds with the bright band in stratiform rain. A pre-classification of the rain type for each minute based on the rain microstructure provided by the collocated disdrometer was performed. Our results indicate that complex ML models, particularly tree-based ensembles such as xgboost and random forest which capture the interactions of different features, perform better than simpler models. Applying methods from the field of interpretable ML, we identified reflectivity at the lowest layer and the average spectral width in the layers below SL as the most important features. High reflectivity and low SW values indicate a higher probability of convective rain.


  • Open Access

    Grouped feature importance and combined features effect plot

    by Quay Au, Julia Herbinger, Clemens Stachl, Bernd Bischl, Giuseppe Casalicchio

    Interpretable machine learning has become a very active area of research due to the rising popularity of machine learning algorithms and their inherently challenging interpretability. Most work in this area has been focused on the interpretation of single features in a model. However, for researchers and practitioners, it is often equally important to quantify the importance or visualize the effect of feature groups. To address this research gap, we provide a comprehensive overview of how existing model-agnostic techniques can be defined for feature groups to assess the grouped feature importance, focusing on permutation-based, refitting, and Shapley-based methods. We also introduce an importance-based sequential procedure that identifies a stable and well-performing combination of features in the grouped feature space. Furthermore, we introduce the combined features effect plot, which is a technique to visualize the effect of a group of features based on a sparse, interpretable linear combination of features. We used simulation studies and real data examples to analyze, compare, and discuss these methods.


  • Top Cited Papers
    Open Access

    General Pitfalls of Model-Agnostic Interpretation Methods for Machine Learning Models

    by Christoph Molnar, Gunnar König, Julia Herbinger, Timo Freiesleben, Susanne Dandl, Christian A. Scholbeck, Giuseppe Casalicchio, Moritz Grosse-Wentrup, Bernd Bischl

    An increasing number of model-agnostic interpretation techniques for machine learning (ML) models such as partial dependence plots (PDP), permutation feature importance (PFI) and Shapley values provide insightful model interpretations, but can lead to wrong conclusions if applied incorrectly. We highlight many general pitfalls of ML model interpretation, such as using interpretation techniques in the wrong context, interpreting models that do not generalize well, ignoring feature dependencies, interactions, uncertainty estimates and issues in high-dimensional settings, or making unjustified causal interpretations, and illustrate them with examples. We focus on pitfalls for global methods that describe the average model behavior, but many pitfalls also apply to local methods that explain individual predictions. Our paper addresses ML practitioners by raising awareness of pitfalls and identifying solutions for correct model interpretation, but also addresses ML researchers by discussing open issues for further research.


  • Open Access

    Portfolio optimization with optimal expected utility risk measures

    by S. Geissel, H. Graf, J. Herbinger, F. T. Seifried

    The purpose of this article is to evaluate optimal expected utility risk measures (OEU) in a risk-constrained portfolio optimization context where the expected portfolio return is maximized. We compare the portfolio optimization with OEU constraint to a portfolio selection model using value at risk as constraint. The former is a coherent risk measure for utility functions with constant relative risk aversion and allows individual specifications to the investor’s risk attitude and time preference. In a case study with three indices, we investigate how these theoretical differences influence the performance of the portfolio selection strategies. A copula approach with univariate ARMA-GARCH models is used in a rolling forecast to simulate monthly future returns and calculate the derived measures for the optimization. The results of this study illustrate that both optimization strategies perform considerably better than an equally weighted portfolio and a buy and hold portfolio. Moreover, our results illustrate that portfolio optimization with OEU constraint experiences individualized effects, e.g., less risk-averse investors lose more portfolio value in the financial crises but outperform their more risk-averse counterparts in bull markets.