Georg Moser
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
Research Subjects
Georg Moser has a total of 70 co-authors. Here are the top co-authors by number of shared publications:
- Martin Avanzini
Institute of Computer Science University of Innsbruck Austria
7 shared publications Martin Avanzini
6 shared publications- Martin Avanzini
Centre Inria d'Université Côte d'Azur, Sophia Antipolis, France
5 shared publications - Florian Zuleger
TU Wien
4 shared publications - Tobias Hell
Data Lab Hell GmbH, Zirl, Austria
4 shared publications - Albert Nössig
Department of Computer Science, University of Innsbruck, Innsbruck, Austria
3 shared publications
Latest Publications
- Open Access
- 7 October 2026
- Published by Springer Nature
- p. 530-557
Inferring Empirical Sound Resource Bounds via Symbolic Execution and Linear Programming
- 7 October 2026
- Published by Springer Nature
- p. 530-557
Open AccessInferring Empirical Sound Resource Bounds via Symbolic Execution and Linear Programming
- Open Access
- 22 August 2026
- Published by Association for Computing Machinery (ACM)
- p. 114-146
A Cost-Aware Probability Monad for Liquid Haskell
- 22 August 2026
- Published by Association for Computing Machinery (ACM)
- p. 114-146
Open AccessA Cost-Aware Probability Monad for Liquid Haskell
- Open Access
- 24 July 2026
- Published by Springer Nature
- p. 100-122
Automated Amortised Analysis of Skew Heaps and Leftist Heaps
We study the fully automated amortised analysis of purely functional data structures like skew heaps , as well as weight - and rank-biased leftist heaps. For that we generalise earlier works on automated amortised resource analysis by developing a type inference based approach with a generic type system. This allows for modular reasoning and the inference of precise and optimal cost bounds. More specifically, we extend the work on the ATLAS system by Leutgeb et al. which was developed to cover the analysis of splay trees and some closely related data structures. To enable the analysis of skew heaps, however, and the even more challenging (amortised) analysis of leftist heaps, we have developed a range of new techniques for type-based automated analysis. By introducing a generic type system we allow for arbitrary (classes of) potential functions, compared to the use of hard-coded potential functions in ATLAS, which we have implemented in Haskell in an entirely modular way. We have also greatly enhanced the existing type inference algorithm by extensions in multiple directions, including path-sensitive reasoning, data structure invariants, and template parameters for piecewise defined potential functions. We show how our newly developed system supports the use of all known potential functions for analysing skew heaps and leftist heaps, confirming the known bounds.
- 24 July 2026
- Published by Springer Nature
- p. 100-122
Open AccessAutomated Amortised Analysis of Skew Heaps and Leftist Heaps
We study the fully automated amortised analysis of purely functional data structures like skew heaps , as well as weight - and rank-biased leftist heaps. For that we generalise earlier works on automated amortised resource analysis by developing a type inference based approach with a generic type system. This allows for modular reasoning and the inference of precise and optimal cost bounds. More specifically, we extend the work on the ATLAS system by Leutgeb et al. which was developed to cover the analysis of splay trees and some closely related data structures. To enable the analysis of skew heaps, however, and the even more challenging (amortised) analysis of leftist heaps, we have developed a range of new techniques for type-based automated analysis. By introducing a generic type system we allow for arbitrary (classes of) potential functions, compared to the use of hard-coded potential functions in ATLAS, which we have implemented in Haskell in an entirely modular way. We have also greatly enhanced the existing type inference algorithm by extensions in multiple directions, including path-sensitive reasoning, data structure invariants, and template parameters for piecewise defined potential functions. We show how our newly developed system supports the use of all known potential functions for analysing skew heaps and leftist heaps, confirming the known bounds.
- Open Access
- 20 December 2025
- Published by Centre pour la Communication Scientifique Directe (CCSD) in Electronic Notes in Theoretical Informatics and Computer Science
- Vol. ume 5 - Pr
Modular abstract syntax trees (MAST): substitution tensors with second-class sorts
We adapt Fiore, Plotkin, and Turi's treatment of abstract syntax with binding, substitution, and holes to account for languages with second-class sorts. These situations include programming calculi such as the Call-by-Value lambda-calculus (CBV) and Levy's Call-by-Push-Value (CBPV). Prohibiting second-class sorts from appearing in variable contexts changes the characterisation of the abstract syntax from monoids in monoidal categories to actions in actegories. We reproduce much of the development through bicategorical arguments. We apply the resulting theory by proving substitution lemmata for varieties of CBV.
- 20 December 2025
- Published by Centre pour la Communication Scientifique Directe (CCSD) in Electronic Notes in Theoretical Informatics and Computer Science
- Vol. ume 5 - Pr
Open AccessModular abstract syntax trees (MAST): substitution tensors with second-class sorts
We adapt Fiore, Plotkin, and Turi's treatment of abstract syntax with binding, substitution, and holes to account for languages with second-class sorts. These situations include programming calculi such as the Call-by-Value lambda-calculus (CBV) and Levy's Call-by-Push-Value (CBPV). Prohibiting second-class sorts from appearing in variable contexts changes the characterisation of the abstract syntax from monoids in monoidal categories to actions in actegories. We reproduce much of the development through bicategorical arguments. We apply the resulting theory by proving substitution lemmata for varieties of CBV.
- Open Access
- 15 September 2025
- Published by Springer Nature
- p. 228-246
Data-Driven Runtime Complexity Analysis
We establish a data-driven method for the assessment of the runtime complexity of first-order term rewrite systems (TRSs for short). The fully automated complexity analysis of TRSs has a long tradition in rewriting and numerous sophisticated static analysis methods have been developed. The recent success in machine learning motivates the quest for data-driven analysis techniques, which, while unsound in principle, can potentially return insightful upper bounds on the runtime complexity where traditional (static) techniques fail. We present the first such technique based on bottom-up rule unfolding, akin to a variant of backward narrowing. Further, we employ a dedicated notion of data fitting that is fine-tuned to the estimation of asymptotic complexities. We provide ample experimental data indicating the viability of the approach.
- 15 September 2025
- Published by Springer Nature
- p. 228-246
Open AccessData-Driven Runtime Complexity Analysis
We establish a data-driven method for the assessment of the runtime complexity of first-order term rewrite systems (TRSs for short). The fully automated complexity analysis of TRSs has a long tradition in rewriting and numerous sophisticated static analysis methods have been developed. The recent success in machine learning motivates the quest for data-driven analysis techniques, which, while unsound in principle, can potentially return insightful upper bounds on the runtime complexity where traditional (static) techniques fail. We present the first such technique based on bottom-up rule unfolding, akin to a variant of backward narrowing. Further, we employ a dedicated notion of data fitting that is fine-tuned to the estimation of asymptotic complexities. We provide ample experimental data indicating the viability of the approach.
- Open Access
To Zip Through the Cost Analysis of Probabilistic Programs
Probabilistic programming and the formal analysis of probabilistic algorithms are active areas of research, driven by the widespread use of randomness to improve performance. While functional correctness has seen substantial progress, automated reasoning about expected runtime remains comparatively limited. In this work, we address this challenge by introducing a refinement-typed probability monad in Liquid Haskell. Our monad enables automated reasoning about expected values and costs by encoding probabilistic behaviour directly in types. Initially defined for discrete distributions over finite support, it is extended to support infinite distributions via an axiomatic approach. By leveraging Liquid Haskell's SMT-based refinement type checking, our framework provides a high degree of automation. We evaluate our approach through four case studies: meldable heaps, coupon collector, randomised quicksort, and zip trees. The first two demonstrate automation with minimal annotation overhead. The latter two showcase how our monad integrates with interactive proofs, including the first formal verification of the expected runtime of zip trees.
Open AccessTo Zip Through the Cost Analysis of Probabilistic Programs
Probabilistic programming and the formal analysis of probabilistic algorithms are active areas of research, driven by the widespread use of randomness to improve performance. While functional correctness has seen substantial progress, automated reasoning about expected runtime remains comparatively limited. In this work, we address this challenge by introducing a refinement-typed probability monad in Liquid Haskell. Our monad enables automated reasoning about expected values and costs by encoding probabilistic behaviour directly in types. Initially defined for discrete distributions over finite support, it is extended to support infinite distributions via an axiomatic approach. By leveraging Liquid Haskell's SMT-based refinement type checking, our framework provides a high degree of automation. We evaluate our approach through four case studies: meldable heaps, coupon collector, randomised quicksort, and zip trees. The first two demonstrate automation with minimal annotation overhead. The latter two showcase how our monad integrates with interactive proofs, including the first formal verification of the expected runtime of zip trees.
- Open Access
- 18 January 2025
- Published by Springer Nature in Neural Computing and Applications
- Vol. 37 (31) , 25663-25694
Average reward adjusted discounted reinforcement learning
In this paper, we provide a novel view upon reinforcement learning (RL for short). In particular, we are interested in applications of RL in use cases, where average rewards may be nonzero. While RL methodologies have been extensively researched upon, this particular application area has only received scarce attention in the Literature. In part, our motivation stems from applications in Operation Research (OR for short), where it is typically the case that rewards are profit derived. Similar use cases can be found in more general applications in economics. Based on a principled study of the mathematical background of discounted reinforcement learning we establish a novel adaptation of standard RL, dubbed Average Reward Adjusted Discounted Reinforcement Learning (ARAL for short). Our approach stems from revisiting the Laurent Series expansion of the discounted state value and a subsequent reformulation of the target function guiding the learning process. While the theoretical advance is arguably incremental, we provide ample experimental evidence that the thus obtained novel RL methodology compares favorable to well-established techniques like Q-learning or R-learning.
- 18 January 2025
- Published by Springer Nature in Neural Computing and Applications
- Vol. 37 (31) , 25663-25694
Open AccessAverage reward adjusted discounted reinforcement learning
In this paper, we provide a novel view upon reinforcement learning (RL for short). In particular, we are interested in applications of RL in use cases, where average rewards may be nonzero. While RL methodologies have been extensively researched upon, this particular application area has only received scarce attention in the Literature. In part, our motivation stems from applications in Operation Research (OR for short), where it is typically the case that rewards are profit derived. Similar use cases can be found in more general applications in economics. Based on a principled study of the mathematical background of discounted reinforcement learning we establish a novel adaptation of standard RL, dubbed Average Reward Adjusted Discounted Reinforcement Learning (ARAL for short). Our approach stems from revisiting the Laurent Series expansion of the discounted state value and a subsequent reformulation of the target function guiding the learning process. While the theoretical advance is arguably incremental, we provide ample experimental evidence that the thus obtained novel RL methodology compares favorable to well-established techniques like Q-learning or R-learning.
- Open Access
- 1 January 2025
- Published by Association for Computational Linguistics (ACL)
- p. 739-748
LaTeXMT: Machine Translation for LaTeX Documents
Calvin Hoy, Samuel Frontull, Georg Moser. Proceedings of the 2025 Conference on Empirical Methods in Natural Language Processing: System Demonstrations. 2025.
- 1 January 2025
- Published by Association for Computational Linguistics (ACL)
- p. 739-748
Open AccessLaTeXMT: Machine Translation for LaTeX Documents
Calvin Hoy, Samuel Frontull, Georg Moser. Proceedings of the 2025 Conference on Empirical Methods in Natural Language Processing: System Demonstrations. 2025.
- Open Access
- 3 July 2024
- Published by Springer Nature in Machine Learning
- Vol. 113 (10) , 7479-7508
Rule learning by modularity
In this paper, we present a modular methodology that combines state-of-the-art methods in (stochastic) machine learning with well-established methods in inductive logic programming (ILP) and rule induction to provide efficient and scalable algorithms for the classification of vast data sets. By construction, these classifications are based on the synthesis of simple rules, thus providing direct explanations of the obtained classifications. Apart from evaluating our approach on the common large scale data sets MNIST, Fashion-MNIST and IMDB, we present novel results on explainable classifications of dental bills. The latter case study stems from an industrial collaboration with Allianz Private Krankenversicherung which is an insurance company offering diverse services in Germany.
- 3 July 2024
- Published by Springer Nature in Machine Learning
- Vol. 113 (10) , 7479-7508
Open AccessRule learning by modularity
In this paper, we present a modular methodology that combines state-of-the-art methods in (stochastic) machine learning with well-established methods in inductive logic programming (ILP) and rule induction to provide efficient and scalable algorithms for the classification of vast data sets. By construction, these classifications are based on the synthesis of simple rules, thus providing direct explanations of the obtained classifications. Apart from evaluating our approach on the common large scale data sets MNIST, Fashion-MNIST and IMDB, we present novel results on explainable classifications of dental bills. The latter case study stems from an industrial collaboration with Allianz Private Krankenversicherung which is an insurance company offering diverse services in Germany.
- 21 June 2024
- Published by Springer Nature
- p. 155-169
A Voting Approach for Explainable Classification with Rule Learning
- 21 June 2024
- Published by Springer Nature
- p. 155-169
A Voting Approach for Explainable Classification with Rule Learning