TY - JOUR A1 - Weichert, Dorina A1 - Kister, Alexander A1 - Volbach, Peter A1 - Houben, Sebastian A1 - Trost, Marcus A1 - Wrobel, Stefan T1 - Explainable production planning under partial observability in high-precision manufacturing N2 - Conceptually, high-precision manufacturing is a sequence of production and measurement steps, where both kinds of steps require to use non-deterministic models to represent production and measurement tolerances. This paper demonstrates how to effectively represent these manufacturing processes as Partially Observable Markov Decision Processes (POMDP) and derive an offline strategy with state-of-the-art Monte Carlo Tree Search (MCTS) approaches. In doing so, we face two challenges: a continuous observation space and explainability requirements from the side of the process engineers. As a result, we find that a tradeoff between the quantitative performance of the solution and its explainability is required. In a nutshell, the paper elucidates the entire process of explainable production planning: We design and validate a white-box simulation from expert knowledge, examine state-of-the-art POMDP solvers, and discuss our results from both the perspective of machine learning research and as an illustration for high-precision manufacturing practitioners. KW - Explainability KW - Manufacturing KW - Reinforcement Learning KW - Monte Carlo tree search KW - Partially observable Markov decision process PY - 2023 DO - https://doi.org/10.1016/j.jmsy.2023.08.009 SN - 0278-6125 VL - 70 SP - 514 EP - 524 PB - Elsevier Ltd. CY - Southfield AN - OPUS4-58963 LA - eng AD - Bundesanstalt fuer Materialforschung und -pruefung (BAM), Berlin, Germany ER - TY - CONF A1 - Kister, Alexander T1 - Multi-Agent Neural Rewriter for Vehicle Routing with Limited Disclosure of Costs N2 - We interpret solving the multi-vehicle routing problem as a team Markov game with partially observable costs. For a given set of customers to serve, the playing agents (vehicles) have the common goal to determine the team-optimal agent routes with minimal total cost. Each agent thereby observes only its own cost. Our multi-agent reinforcement learning approach, the so-called multi-agent Neural Rewriter, builds on the single-agent Neural Rewriter to solve the problem by iteratively rewriting solutions. Parallel agent action execution and partial observability require new rewriting rules for the game. We propose the introduction of a so-called pool in the system which serves as a collection point for unvisited nodes. It enables agents to act simultaneously and exchange nodes in a conflict-free manner. We realize limited disclosure of agent-specific costs by only sharing them during learning. During inference, each agents acts decentrally, solely based on its own cost. First empirical results on small problem sizes demonstrate that we reach a performance close to the employed OR-Tools benchmark which operates in the perfect cost information setting. T2 - Gamification and Multiagent Solutions Workshop (ICLR 2022) CY - Online meeting DA - 29.04.2022 KW - Reinforcement learning KW - Deep Learning KW - Vehicle Routing PY - 2022 AN - OPUS4-56687 LA - eng AD - Bundesanstalt fuer Materialforschung und -pruefung (BAM), Berlin, Germany ER - TY - CONF A1 - Weichert, Dorina A1 - Kister, Alexander A1 - Houben, Sebastian A1 - Link, Patrick A1 - Ernis, Gunar ED - Kiyavash, Negar ED - Mooij, Joris M. T1 - Robust Entropy Search for Safe Efficient Bayesian Optimization N2 - The practical use of Bayesian Optimization (BO) in engineering applications imposes special requirements: high sampling efficiency on the one hand and finding a robust solution on the other hand. We address the case of adversarial robustness, where all parameters are controllable during the optimization process, but a subset of them is uncontrollable or even adversely perturbed at the time of application. To this end, we develop an efficient information-based acquisition function that we call Robust Entropy Search (RES). We empirically demonstrate its benefits in experiments on synthetic and real-life data. The results show that RES reliably finds robust optima, outperforming state-of-the-art algorithms. T2 - UAI 2024 CY - Barcelona, Spain DA - 15.07.2024 KW - Bayesian Optimization KW - Gaussian process KW - Active learning PY - 2024 UR - https://nbn-resolving.org/urn:nbn:de:kobv:b43-613523 UR - https://proceedings.mlr.press/v244/weichert24a.html SN - 2640-3498 VL - 244 SP - 3711 EP - 3729 PB - Proceedings of Machine Learning Research AN - OPUS4-61352 LA - eng AD - Bundesanstalt fuer Materialforschung und -pruefung (BAM), Berlin, Germany ER - TY - JOUR A1 - Albakri, Bashar A1 - Turski Silva Diniz, Analice A1 - Benner, Philipp A1 - Muth, Thilo A1 - Nakajima, Shinichi A1 - Favaro, Marco A1 - Kister, Alexander ED - Hillman, Robert T1 - Machine learning-assisted equivalent circuit identification for dielectric spectroscopy of polymers N2 - Polymers have become indispensable across fields of application, and understanding their structure–property relationships and dynamic behaviour is essential for performance optimization. Polymer membranes, particularly ion exchange membranes, play a crucial role in renewable energy conversion technologies, fuel cells, solar energy conversion, and energy storage. In this context, broadband dielectric spectroscopy (BDS) offers a powerful, non-destructive approach to investigate the electrical response and relaxation dynamics of polymers. These properties are investigated by parametrizing the system’s impedance response in terms of a network of circuit elements, i.e. the electrical equivalent circuit (EEC), whose impedance resembles the one of the system under investigation. However, the determination of the EEC from BDS data is challenging due to system complexity, interdependencies of circuit elements, and researcher biases. In this work, we propose a novel approach that incorporates a convolutional neural network (CNN) model to predict the EEC topology. By reducing user bias and enhancing data analysis, this approach aims to make BDS accessible to both experienced users and those with limited expertise. We show that the combination of machine learning and BDS provides valuable insights into the dynamic behaviour of polymer membranes, thus facilitating the design and characterization of tailored polymers for various applications. We also show that our model outperforms state-of-the-art machine learning methods with a top-5 accuracy of around 80% for predicting the circuit topology and a parameter fitting error as low as 0.05%. KW - Polymer membranes KW - Electrochemical impedance spectroscopy KW - Broadband dielectric spectroscopy KW - Deep learning KW - Machine learning KW - Equivalent circuit PY - 2024 UR - https://nbn-resolving.org/urn:nbn:de:kobv:b43-602138 DO - https://doi.org/10.1016/j.electacta.2024.144474 SN - 0013-4686 VL - 496 SP - 1 EP - 13 PB - Elsevier Ltd. AN - OPUS4-60213 LA - eng AD - Bundesanstalt fuer Materialforschung und -pruefung (BAM), Berlin, Germany ER - TY - JOUR A1 - Paul, Nathalie A1 - Karamitsou, Venetia A1 - Giegerich, Clemens A1 - Sadeghi, Afshin A1 - Lücke, Moritz A1 - Wagenhuber, Britta A1 - Kister, Alexander A1 - Rehberg, Markus T1 - Building virtual patients using simulation-based inference N2 - In the context of in silico clinical trials, mechanistic computer models for pathophysiology and pharmacology (here Quantitative Systems Pharmacology models, QSP) can greatly support the decision making for drug candidates and elucidate the (potential) response of patients to existing and novel treatments. These models are built on disease mechanisms and then parametrized using (clinical study) data. Clinical variability among patients is represented by alternative model parameterizations, called virtual patients. Despite the complexity of disease modeling itself, using individual patient data to build these virtual patients is particularly challenging given the high-dimensional, potentially sparse and noisy clinical trial data. In this work, we investigate the applicability of simulation-based inference (SBI), an advanced probabilistic machine learning approach, for virtual patient generation from individual patient data and we develop and evaluate the concept of nearest patient fits (SBI NPF), which further enhances the fitting performance. At the example of rheumatoid arthritis where prediction of treatment response is notoriously difficult, our experiments demonstrate that the SBI approaches can capture large inter-patient variability in clinical data and can compete with standard fitting methods in the field. Moreover, since SBI learns a probability distribution over the virtual patient parametrization, it naturally provides the probability for alternative parametrizations. The learned distributions allow us to generate highly probable alternative virtual patient populations for rheumatoid arthritis, which could potentially enhance the assessment of drug candidates if used for in silico trials. KW - Simulation-based inference KW - Machine learning KW - Artificial intelligence KW - Virtual patients KW - Pathophysiology KW - Pharmacology PY - 2024 UR - https://nbn-resolving.org/urn:nbn:de:kobv:b43-626050 DO - https://doi.org/10.3389/fsysb.2024.1444912 SN - 2674-0702 VL - 4 SP - 1 EP - 11 PB - Frontiers Media SA AN - OPUS4-62605 LA - eng AD - Bundesanstalt fuer Materialforschung und -pruefung (BAM), Berlin, Germany ER - TY - CONF A1 - Paul, Nathalie A1 - Kister, Alexander A1 - Schnellhardt, Thorben A1 - Fetz, Maximilian A1 - Hecker, Dirk A1 - Wirtz, Tim ED - Meo, Rosa ED - Silvestri, Fabrizio T1 - Reinforcement Learning for Segmented Manufacturing N2 - The manufacturing of large components is, in comparison to small components, cost intensive. This is due to the sheer size of the components and the limited scalability in number of produced items. To take advantage of the effects of small component production we segment the large components into smaller parts and schedule the production of these parts on regular-sized machine tools. We propose to apply and adapt recent developments in reinforcement learning in combination with heuristics to efficiently solve the resulting segmentation and assignment problem. In particular, we solve the assignment problem up to a factor of 8 faster and only a few percentages less accurate than a classic solver from operations research. T2 - European Conference on Machine Learning and Principles and Practice of Knowledge Discovery in Databases 2023 CY - Turin, Italy DA - 18.09.2023 KW - Reinforcement Learning KW - Assignment Problem KW - Large component manufacturing PY - 2025 DO - https://doi.org/10.1007/978-3-031-74640-6_38 VL - 1 IS - 1 SP - 470 EP - 485 PB - Springer Cham AN - OPUS4-63031 LA - eng AD - Bundesanstalt fuer Materialforschung und -pruefung (BAM), Berlin, Germany ER - TY - INPR A1 - Alphonse, Amal A1 - Hintermüller, Michael A1 - Kister, Alexander A1 - Lun, Chin Hang A1 - Sirotenko, Clemens T1 - A neural network approach to learning solutions of a class of elliptic variational inequalities N2 - We develop a weak adversarial approach to solving obstacle problems using neural networks. By employing (generalised) regularised gap functions and their properties we rewrite the obstacle problem (which is an elliptic variational inequality) as a minmax problem, providing a natural formulation amenable to learning. Our approach, in contrast to much of the literature, does not require the elliptic operator to be symmetric. We provide an error analysis for suitable discretisations of the continuous problem, estimating in particular the approximation and statistical errors. Parametrising the solution and test function as neural networks, we apply a modified gradient descent ascent algorithm to treat the problem and conclude the paper with various examples and experiments. Our solution algorithm is in particular able to easily handle obstacle problems that feature biactivity (or lack of strict complementarity), a situation that poses difficulty for traditional numerical methods. KW - Obstacle Problems KW - Numerical Analysis KW - Optimization and Control KW - Simulation PY - 2024 UR - https://nbn-resolving.org/urn:nbn:de:kobv:b43-622160 DO - https://doi.org/10.48550/arXiv.2411.18565 SP - 1 EP - 38 AN - OPUS4-62216 LA - eng AD - Bundesanstalt fuer Materialforschung und -pruefung (BAM), Berlin, Germany ER - TY - JOUR A1 - Nowatzky, Yannek A1 - Russo, Francesco Friedrich A1 - Lisec, Jan A1 - Kister, Alexander A1 - Reinert, Knut A1 - Muth, Thilo A1 - Benner, Philipp T1 - FIORA: Local neighborhood-based prediction of compound mass spectra from single fragmentation events N2 - Non-targeted metabolomics holds great promise for advancing precision medicine and biomarker discovery. However, identifying compounds from tandem mass spectra remains a challenging task due to the incomplete nature of spectral reference libraries. Augmenting these libraries with simulated mass spectra can provide the necessary references to resolve unmatched spectra, but generating high-quality data is difficult. In this study, we present FIORA, an open-source graph neural network designed to simulate tandem mass spectra. Our main contribution lies in utilizing the molecular neighborhood of bonds to learn breaking patterns and derive fragment ion probabilities. FIORA not only surpasses state-of-the-art fragmentation algorithms, ICEBERG and CFM-ID, in prediction quality, but also facilitates the prediction of additional features, such as retention time and collision cross section. Utilizing GPU acceleration, FIORA enables rapid validation of putative compound annotations and large-scale expansion of spectral reference libraries with high-quality predictions. KW - Metabolomics KW - Machine Learning KW - Mass spectrometry KW - Bioinformatics PY - 2025 UR - https://nbn-resolving.org/urn:nbn:de:kobv:b43-627585 DO - https://doi.org/10.1038/s41467-025-57422-4 SN - 2041-1723 VL - 16 IS - 1 SP - 1 EP - 17 PB - Springer Science and Business Media LLC AN - OPUS4-62758 LA - eng AD - Bundesanstalt fuer Materialforschung und -pruefung (BAM), Berlin, Germany ER -