Daphne Theodorakopoulos, M.Sc.


30167 Hannover


30167 Hannover
I am a PhD candidate at the AutoML Group. My research topic here is Green AutoML. I am also a researcher at the DFKI Marine Perception in Oldenburg. Our focus is AI for marine perception but also other environmental and sustainability topics. I aim to find a good intersection between the two departments by using Green AutoML to tackle environmental problems.
Research Interests
- Green AutoML
- AI for Sustainability
- Knowledge Graphs
Curriculum Vitae
-
Working Experience
since 2022
Researcher, DFKI Marine Perception2020 - 2022
Teaching Assistant, University of Twente (NL)2019
Internship Big Data & Advanced Analytics, Daimler AG2016 - 2018
Teaching Assistant, University of Hamburg -
Education
2022 - Present
PhD Candidate, Leibniz University Hannover2019-2022
Master of Science, Interaction Technology, University of Twente (NL)2015-2019
Bachelor of Science, Wirtschaftsinformatik, University of Hamburg
Publications
-
(2022): PriorBand: HyperBand + Human Expert Knowledge, 2022 NeurIPS Workshop on Meta Learning (MetaLearn)
-
(2022): Towards Automated Design of Bayesian Optimization via Exploratory Landscape Analysis, 6th Workshop on Meta-Learning at NeurIPS 2022, New Orleans
-
(2022): DeepCAVE: An Interactive Analysis Tool for Automated Machine Learning, ICML Workshop on Adaptive Experimental Design and Active Learning in the Real World (ReALML)
arXiv: 2206.03493 -
(2022): Towards Meta-learned Algorithm Selection using Implicit Fidelity Information, ICML Workshop on Adaptive Experimental Design and Active Learning in the Real World (ReALML)
arXiv: 2206.03130 -
(2022): Automated Dynamic Algorithm Configuration, Computing Research Repository (CoRR) (2022)
DOI: https://doi.org/10.48550/arXiv.2205.13881
arXiv: 2205.13881 -
(2022): POLTER: Policy Trajectory Ensemble Regularization for Unsupervised Reinforcement Learning, CoRR
arXiv: arXiv:2205.11357 -
(2022): Efficient Automated Deep Learning for Time Series Forecasting, Proceedings of the European Conference on Machine Learning (ECML)
arXiv: 2205.05511 -
(2022): piBO: Augmenting Acquisition Functions with User Beliefs for Bayesian Optimization, 10th International Conference on Learning Representations, ICLR'22
arXiv: https://openreview.net/forum?id=MMAeCXIa89 -
(2022): Contextualize Me - The Case for Context in Reinforcement Learning, ArXiv Preprint
arXiv: https://arxiv.org/abs/2202.04500 -
(2022): SMAC3: A Versatile Bayesian Optimization Package for Hyperparameter Optimization, Journal of Machine Learning Research (JMLR) -- MLOSS, Vol. 23, No. 54, pp. 1-9 More Info
-
(2022): Automated Reinforcement Learning (AutoRL): A Survey and Open Problems, Journal of Artificial Intelligence Research (JAIR)
arXiv: 2201.03916 -
(2021): Well-tuned Simple Nets Excel on Tabular Datasets, Advances in Neural Information Processing Systems (NeurIPS 2021)
arXiv: 2106.11189 -
(2021): Explaining Hyperparameter Optimization via Partial Dependence Plots, Proceedings of the international conference on Neural Information Processing Systems (NeurIPS) | File |
arXiv: 2111.04820 -
(2021): Hyperparameters in Contextual RL are Highly Situational, International Workshop on Ecological Theory of RL (at NeurIPS) | File |
-
(2021): Well-tuned Simple Nets Excel on Tabular Datasets, NeurIPS 2021
DOI: https://doi.org/10.48550/arXiv.2106.11189
arXiv: 2106.11189 -
(2021): Searching in the Forest for Local Bayesian Optimization, ECMLPKDD Workshop on Meta-Knowledge Transfe
arXiv: 2111.05834 -
(2021): Maschinelles Lernen in der Prozessplanung
-
(2021): CARL: A Benchmark for Contextual and Adaptive Reinforcement Learning, Workshop on Ecological Theory of Reinforcement Learning, NeurIPS 2021
arXiv: 2110.02102 -
(2021): HPOBench: A Collection of Reproducible Multi-Fidelity Benchmark Problems for HPO, Proceedings of the international conference on Neural Information Processing Systems (NeurIPS) (Datasets and Benchmarks Track)
arXiv: 2109.06716 -
(2021): Bayesian Optimization with a Prior for the Optimum, Proceedings of the European conference on machine learning (ECML)
arXiv: 2006.14608 -
(2022): Developing Open Source Educational Resources for Machine Learning and Data Science, Teaching Machine Learning Workshop at ECML 2022
arXiv: 2107.14330 -
(2021): Hyperparameter Optimization: Foundations, Algorithms, Best Practices and Open Challenges, WIREs Data Mining and Knowledge Discovery
DOI: 10.1002/widm.1484
arXiv: 2107.05847 -
(2021): Self-Paced Context Evaluation for Contextual Reinforcement Learning, Proceedings of the international conference on machine learning (ICML)
arXiv: 2106.05110 -
(2021): TempoRL: Learning When to Act, Proceedings of the international conference on machine learning (ICML) 2021
arXiv: 2106.05262 -
(2021): Towards Explaining Hyperparameter Optimization via Partial Dependence Plots, 8th ICML Workshop on Automated Machine Learning (AutoML) | File |
-
(2021): Learning Heuristic Selection with Dynamic Algorithm Configuration, International Conference on Automated Planning and Scheduling (ICAPS) 2021 | File |
arXiv: 2006.08246 -
(2021): Bag of Baselines for Multi-objective Joint Neural Architecture Search and Hyperparameter Optimization, ICML 2021 Workshop AutoML
arXiv: 2105.01015 -
(2021): DACBench: A Benchmark Library for Dynamic Algorithm Configuration, Proceedings of the international joint conference on AI (IJCAI) 2021
arXiv: 2105.08541 -
(2021): Winning solutions and post-challenge analyses of the ChaLearn AutoDL challenge 2019 | File |
-
(2021): Auto-Pytorch: Multi-Fidelity MetaLearning for Efficient and Robust AutoDL., IEEE Transactions on Pattern Analysis and Machine Intelligence
arXiv: 2006.13799 -
(2020): Using AutoML to Optimize Shape Error Prediction in Milling Processes, Proceedings of the Machining Innovations Conference (MIC) | File |
-
(2019): Best Practices for Scientific Research on Neural Architecture Search, Computing Research Repository (CoRR) (2019)
DOI: https://doi.org/10.48550/arXiv.1908.06756
arXiv: 1908.06756 -
(2020): Learning Step-Size Adaptation in CMA-ES, Proceedings of international PPSN conference 2020 | File |
-
(2020): Towards Self-Paced Context Evaluation for Contextual Reinforcement Learning, International Conference on Machine Learning (ICML) 2020 | File |
-
(2020): Towards TempoRL: Learning When to Act, International Conference on Machine Learning (ICML) 2020 | File |
-
(2022): Auto-Sklearn 2.0: The Next Generation, Journal of Machine Learning Research (JMLR), 23(261):1−61 More Info
arXiv: 2007.04074 -
(2020): Dynamic Algorithm Configuration: Foundation of a New Meta-Algorithmic Framework, European Conference on AI (ECAI) 2020 | File |
-
(2019): Towards Assessing the Impact of Bayesian Optimization’s Own Hyperparameters, Computing Research Repository (CoRR) (2019)
DOI: https://doi.org/10.48550/arXiv.1908.06674
arXiv: 1908.06674 -
(2019): BOAH: A Tool Suite for Multi-Fidelity Bayesian Optimization & Analysis of Hyperparameters, Computing Research Repository (CoRR) (2019)
DOI: https://doi.org/10.48550/arXiv.1908.06756
arXiv: 1908.06756 -
(2019): Towards Automatically-Tuned Deep Neural Networks, Automated Machine Learning
DOI: https://doi.org/10.1007/978-3-030-05318-5
ISBN: 978-3-030-05317-8
ISSN: 2520-131X -
(2018): CAVE: Configuration Assessment, Visualization and Evaluation, Learning and Intelligent Optimization (LION) 2018 | File |
-
(2018): Neural Networks for Predicting Algorithm Runtime Distributions, International Joint Conference on Artificial Intelligence (IJCAI) 2018
DOI: https://doi.org/10.48550/arXiv.1709.07615
arXiv: 1709.07615 -
(2017): Warmstarting of Model-based Algorithm Configuration, Computing Research Repository (CoRR) (2017)
DOI: https://doi.org/10.48550/arXiv.1709.04636
arXiv: 1709.04636 -
(2017): Open Algorithm Selection Challenge 2017 Setup and Scenarios, Open Algorithm Selection Challenge 2017 Setup and Scenarios (OASC) 2017 | File |
-
(2017): Improving local search in a minimum vertex cover solver for classes of networks, IEEE Congress on Evolutionary Computation (CEC) 2017
DOI: 10.1109/CEC.2017.7969507
ISBN: 978-1-5090-4602-7 -
(2017): Efficient Benchmarking of Algorithm Configuration Procedures via Model-Based Surrogates, Computing Research Repository (CoRR) (2017)
DOI: https://doi.org/10.48550/arXiv.1703.10342
arXiv: 1703.10342 -
(2017): Pitfalls and Best Practices in Algorithm Configuration, Computing Research Repository (CoRR) (2017)
DOI: https://doi.org/10.48550/arXiv.1705.06058
arXiv: 1705.06058 -
(2017): Automatic Construction of Parallel Portfolios via Algorithm Configuration, Artif. Intell. (2017)
DOI: https://doi.org/10.1016/j.artint.2016.05.004 -
(2017): Efficient Parameter Importance Analysis via Ablation with Surrogates, Association for the Advancement of Artificial Intelligence (AAAI) 2017
DOI: https://doi.org/10.1609/aaai.v31i1.10657 -
(2016): An Empirical Study of Per-Instance Algorithm Scheduling, Learning and Intelligent Optimization (LION) 2016
DOI: https://doi.org/10.1007/978-3-319-50349-3_20
ISBN: 978-3-319-50348-6 -
(2016): A case study of algorithm selection for the traveling thief problem, Computing Research Repository (CoRR) (2016)
arXiv: 1609.00462 -
(2016): The Configurable SAT Solver Challenge (CSSC), Computing Research Repository (CoRR) (2015)
DOI: https://doi.org/10.48550/arXiv.1505.01221
arXiv: 1505.01221 -
(2016): SpyBug: Automated Bug Detection in the Configuration Space of SAT Solvers, SAT 2016
DOI: 10.1007/978-3-319-40970-2_36 -
(2015): AutoFolio: An Automatically Configured Algorithm Selector, Journal of Artificial Intelligence Research (2015)
DOI: https://doi.org/10.1613/jair.4726 -
(2015): AutoFolio: Algorithm Configuration for Algorithm Selection, Association for the Advancement of Artificial Intelligence (AAAI) Workshop: Algorithm Configuration 2015 | File |
-
(2015): ASlib: A Benchmark Library for Algorithm Selection, Computing Research Repository (CoRR) (2015)
DOI: https://doi.org/10.48550/arXiv.1506.02465
arXiv: 1506.02465 -
(2015): From Sequential Algorithm Selection to Parallel Portfolio Selection, Learning and Intelligent Optimization (LION) 2015
DOI: 10.1007/978-3-319-19084-6_1 -
(2014): claspfolio 2: Advances in Algorithm Selection for Answer Set Programming, Theory and Practice of Logic Programming
DOI: 10.1017/S1471068414000210
arXiv: 1405.1520 -
(2014): AClib: a Benchmark Library for Algorithm Configuration, Learning and Intelligent Optimization (LION 8)
DOI: https://doi.org/10.1007/978-3-319-09584-4
ISBN: 978-3-319-09583-7
ISSN: 0302-9743 -
(2013): Robust Benchmark Set Selection for Boolean Constraint Solvers, Learning and Intelligent Optimization (LION) 2013
DOI: https://doi.org/10.1007/978-3-642-44973-4_16
ISBN: 978-3-642-44972-7 -
(2012): Surviving Solver Sensitivity: An ASP Practitioner’s Guide, International Conference on Logic Programming (ICLP (Technical Communications)) 2012
DOI: 10.4230/LIPIcs.ICLP.2012.164
ISBN: 978-3-939897-43-9
ISSN: 1868-8969 -
(2012): aspeed: ASP-based Solver Scheduling, International Conference on Logic Programming (ICLP (Technical Communications)) 2012 | File |
ISBN: 978-3-939897-43-9
ISSN: 1868-8969 -
(2012): Quantifying Homogeneity of Instance Sets for Algorithm Configuration, Learning and Intelligent Optimization (LION) 2012
DOI: https://doi.org/10.1007/978-3-642-34413-8_14
ISBN: 978-3-642-34412-1 -
(2012): Algorithm Configuration for Portfolio-based Parallel SAT-Solving, Proceedings of the ECAI-12 Workshop on Combining Constraint Solving with MIning and Learning | File |
-
(2011): A Portfolio Solver for Answer Set Programming: Preliminary Report, Logic Programming and Non-Monotonic Reasoning (LPNMR) 2011
DOI: https://doi.org/10.1007/978-3-642-20895-9_40
ISBN: 978-3-642-20894-2 -
(2011): Potassco: The Potsdam Answer Set Solving Collection, AI Commun. (2011)
DOI: 10.3233/AIC-2011-0491 -
(2010): Centurio, a General Game Player: Parallel, Java- and ASP-based, Künstliche Intell. (2011)
DOI: https://doi.org/10.1007/s13218-010-0077-4 -
(2013): Ricochet Robots: A transverse ASP benchmark, Logic Programming and Non-Monotonic Reasoning (LPNMR) 2013
DOI: https://doi.org/10.1007/978-3-642-40564-8_35
ISBN: 978-3-642-40563-1 -
(2012): clasp, claspfolio, aspeed: Three Solvers from the Answer Set Solving Collection Potassco | File |
-
(2017): What can we learn from algorithm selection data? (Breakout Session Report), Dagstuhl Reports
-
(2022): Learning Domain-Independent Policies for Open List Selection, Proceedings of the 3rd ICAPS workshop on Bridging the Gap Between AI Planning and Reinforcement Learning (PRL), pp. 1-9 | File |
-
(2020): Dynamic Algorithm Configuration: Foundation of a New Meta-Algorithmic Framework, European Conference on Artificial Intelligence (ECAI) 2020
DOI: 10.3233/FAIA200122 -
(2018): Practical Automated Machine Learning for the AutoML Challenge 2018, International Conference on Machine Learning (ICML) 2018 AutoML Workshop | File |
-
(2017): AutoFolio: An Automatically Configured Algorithm Selector (Extended Abstract), International Joint Conference on Artificial Intelligence (IJCAI) 2017
DOI: https://doi.org/10.24963/ijcai.2017/715 -
(2022): PI is back! Switching Acquisition Functions in Bayesian Optimization, 2022 NeurIPS Workshop on Gaussian Processes, Spatiotemporal Modeling, and Decision-making Systems
arXiv: 2211.01455
Projects
-
Dynamic Algorithm ConfigurationDa die Konfigurationen während der Laufzeit in Abhängigkeit vom aktuellen Zustand des Algorithmus ausgewählt werden sollten, kann es als ein Problem des Reinforcement Learning (RL) betrachtet werden, bei dem ein Agent in jedem Zeitschritt die zu verwendende Konfiguration auf der Grundlage der Leistung im letzten Schritt und des aktuellen Zustands des Algorithmus auswählt. Dies ermöglicht uns einerseits den Einsatz leistungsfähiger RL-Methoden, andererseits bringt RL auch eine Reihe von Herausforderungen mit sich, wie Instabilität, Rauschen und Ineffizienz bei der Abtastung, die bei Anwendungen wie DAC angegangen werden müssen. Daher umfasst die Forschung zu DAC auch die Forschung zu zuverlässigem, interpretierbarem, allgemeinem und schnellem Reinforcement Learning.Led by: Prof. Dr. Marius LindauerYear: 2019Funding: DFGDuration: 2019-2023
-
Dynamic Algorithm ConfigurationAs configurations should be chosen during runtime depending on the current algorithm state, it can be viewed as a reinforcement learning (RL) problem where at each timestep an agent selects the configuration to use based on the performance in the last step and the current state of the algorithm. This enables us to use powerful RL methods on one hand; on the other, RL also brings a set of challenges like instability, noise and sample inefficiency that need to be addressed in applications such as DAC. Therefore research on DAC also includes research on reliable, interpretable, general and fast reinforcement learning.Led by: Prof. Dr. Marius LindauerYear: 2019Funding: DFGDuration: 2019-2023
-
CoyPu: Cognitive Economy Intelligence Plattform für die Resilienz wirtschaftlicher ÖkosystemeNaturkatastrophen, Pandemien, Finanzkrisen, politische Krisen und Angebotsknappeheiten oder Nachfrageschocks propagieren sich durch offensichtliche und latente Handelsbeziehungen durch das globale ökonomische System. Dies ist eine Konsequenz der kontinuierlichen Globalisierung mit der einhergehenden Arbeitsteilung. Ziel dieses Projektes ist es diese Verbindungen offenzulegen und kaskadierende Risiken vorherzusagen um damit Unternehmen die Möglichkeit einzuräumen vorausschauend agieren zu können.Led by: Prof. Marius Lindauer and Prof. Maria Esther-Vidal (L3S/LUH)Team:Year: 2021Funding: Innovationswettbewerb Künstliche Intelligenz (BMWK)Duration: 2021-2024
-
Leibniz AI AcademyThe Leibniz AI Academy aims to develop and establish a trans-curricular and interdisciplinary micro-degree program at the Leibniz Universität Hannover (LUH), in which students from different courses of study acquire competencies in the field of Artificial IntelligenceLed by: Prof. Dr. Marius Lindauer, Prof. Dr. Ralph Ewert, Prof. Dr. Johannes KrugelYear: 2021Funding: Bundesministerium für Bildung und Forschung (BMBF)Duration: 2021 - 2024
-
ERC Starting Grant: Interactive and Explainable Human-Centered AutoMLTrust and interactivity are key factors in the future development and use of automated machine learning (AutoML), supporting developers and researchers in determining powerful task-specific machine learning pipelines, including pre-processing, predictive algorithm, their hyperparameters and--if applicable--the architecture design of deep neural networks. Although AutoML is ready for its prime time after it achieved impressive results in several machine learning (ML) applications and its efficiency improved by several orders of magnitudes in recent years, democratization of machine learning via AutoML is still not achieved. In contrast to previously purely automation-centered approaches, ixAutoML is designed with human users at its heart in several stages. First of all, the foundation of trustful use of AutoML will be based on explanations of its results and processes. Therefore, we aim for: (i) Explaining static effects of design decisions in ML pipelines optimized by state-of-the-art AutoML systems. (ii) Explaining dynamic AutoML policies for temporal aspects of dynamically adapted hyperparameters while ML models are trained. These explanations will be the base for allowing interactions, bringing the best of two worlds together: human intuition and generalization capabilities for complex systems, and efficiency of systematic optimization approaches for AutoML. Concretely, we aim for: (iii) Enabling interactions between humans and AutoML by taking human's latent knowledge into account and learning when to interact. (iv) Building first ixAutoML prototypes and showing its efficiency in the context of Industry 4.0. Perfectly aligned with the EU's AI strategy and recent efforts on interpretability in the ML community, we strongly believe that this timely human-centered ixAutoML will have a substantial impact on the democratization of machine learning.Led by: Prof. Dr. Marius LindauerTeam:Year: 2022Funding: EUDuration: 2022-2027
-
KISSKI: AI Service CenterThe central approach for the KISSKI project is the research on AI methods and their provision with the goal of enabling a highly available AI service center for critical and sensitive infrastructures with a focus on the fields of medicine and energy. Due to their relevance to society as a whole, medicine and the energy industry are among the future fields of application-oriented AI research in Germany. Beyond the technological developments, artificial intelligence (AI) has the potential to make a significant contribution to social progress. This is particularly true in areas where digitization processes are increasingly gaining ground and complexity is high. For both medicine and the energy industry, the pressure to innovate, but also the potential, is immense due to the availability of more and more distributed information based on a multitude of new sensors and actuators. The increasing complexity of the tasks as well as the availability of very large data sets offer a high potential for the application of AI methods in both topics.Led by: Prof. Dr. Marius LindauerTeam:Year: 2022Funding: BMBFDuration: 2022-2025