With the growing importance of privacy regulations such as the General Data Protection Regulation, anonymizing visual data is becoming increasingly relevant across institutions. However, anonymization can negatively affect the performance of Computer Vision systems that rely on visual features, such as Content-Based Image Retrieval (CBIR). Despite this, the impact of anonymization on CBIR has not been systematically studied. This work addresses this gap, motivated by the DOKIQ project, an artificial intelligence-based system for document verification actively used by the State Criminal Police Office Baden-Württemberg. We propose a simple evaluation framework: retrieval results after anonymization should match those obtained before anonymization as closely as possible. To this end, we systematically assess the impact of anonymization using two public datasets and the internal DOKIQ dataset. Our experiments span three anonymization methods, four anonymization degrees, and four training strategies, all based on the state of the art backbone Self-Distillation with No Labels (DINO)v2. Our results reveal a pronounced retrieval bias in favor of models trained on original data, which produce the most similar retrievals after anonymization. The findings of this paper offer practical insights for developing privacy-compliant CBIR systems while preserving performance.

Document forgery poses a growing threat to legal, economic, and governmental processes, requiring increasingly sophisticated verification mechanisms. One approach involves the use of plausibility checks, rule-based procedures that assess the correctness and internal consistency of data, to detect anomalies or signs of manipulation. Although these verification procedures are essential for ensuring data integrity, existing plausibility checks are manually implemented by software engineers, which is time-consuming. Recent advances in code generation with large language models (LLMs) offer new potential for automating and scaling the generation of these checks. However, adapting LLMs to the specific requirements of an unknown domain remains a significant challenge. This work investigates the extent to which LLMs, adapted on domain-specific code and data through different fine-tuning strategies, can generate rule-based plausibility checks for forgery detection on constrained hardware resources. We fine-tune open-source LLMs, Llama 3.1 8B and OpenCoder 8B, on structured datasets derived from real-world application scenarios and evaluate the generated plausibility checks on previously unseen forgery patterns. The results demonstrate that the models are capable of generating executable and effective verification procedures. This also highlights the potential of LLMs as scalable tools to support human decision-making in security-sensitive contexts where comprehensibility is required.

The steadily increasing utilization of data-driven methods and approaches in areas that handle sensitive personal information such as in law enforcement mandates an ever increasing effort in these institutions to comply with data protection guidelines. In this work, we present a system for automatically anonymizing images of scanned documents, reducing manual effort while ensuring data protection compliance. Our method considers the viability of further forensic processing after anonymization by minimizing automatically redacted areas by combining automatic detection of sensitive regions with knowledge from a manually anonymized reference document. Using a self-supervised image model for instance retrieval of the reference document, our approach requires only one anonymized example to efficiently redact all documents of the same type, significantly reducing processing time. We show that our approach outperforms both a purely automatic redaction system and also a naive copy-paste scheme of the reference anonymization to other documents on a hand-crafted dataset of ground truth redactions.

Many recent advances in artificial intelligence remain difficult to apply in real games due to their black-box nature. These systems often lack transparency and control, making them difficult to integrate into games where player experience and narrative coherence are important. Unpredictable or inexplicable agent behavior can confuse players and frustrate developers. My dissertation explores how program synthesis can address this issue by generating interpretable, controllable representations of agent behavior. Instead of relying on black-box neural network policies, symbolic programs are extracted or generated that capture agent logic in a readable and editable form. Several methods are explored: synthesizing functional programs to imitate and explain game agents, adapting logical program policies to multi-agent settings, and evaluating large language models for code generation across domains such as simplified Atari games, Baba is You, and also tabletop games. An open research question is whether the created conceptual program libraries are transferable across different game domains. My research aims to bridge the gap between AI capabilities and game development needs by making agent behavior transparent, explainable, and adaptable for developers.

Large language models (LLMs) have shown impressive capabilities in generating program code, opening exciting opportunities for applying program synthesis to games. In this work, we explore the potential of LLMs to directly synthesize usable code for a wide range of gaming applications, focusing on two programming languages, Python and Java. We use an evolutionary hill-climbing algorithm, where the mutations and seeds of the initial programs are controlled by LLMs. For Python, the framework covers various game-related tasks, including five miniature versions of Atari games, ten levels of Baba is You, an environment inspired by Asteroids, and a maze generation task. For Java, the framework contains 12 games from the TAG tabletop games framework. Across 29 tasks, we evaluated 12 language models for Python and 8 for Java. Our findings suggest that the performance of LLMs depends more on the task than on model size. While larger models generate more executable programs, these do not always result in higher-quality solutions but are much more expensive. No model has a clear advantage, although on any specific task, one model may be better. Trying many models on a problem and using the best results across them is more reliable than using just one.

This paper presents Tone, a wordless, interactive music learning app designed for educational use in schools. It was developed through an iterative, human-centered design process in collaboration with music educators and universities. As an hybrid between app and game, Tone combines puzzle-based ear and rhythm training with creative composition tasks. The app is built to be instantly accessible-without requiring language, prior music knowledge, or technical setup. Early feedback from educators and learners shows strong engagement and usability, alongside valuable critique regarding content diversity and notation style. This demo showcases Tone's current prototype and outlines future directions for inclusive, engaging music learning.

The balancing process for game levels in competitive two-player contexts involves a lot of manual work and testing, particularly for non-symmetrical game levels. In this work, we frame game balancing as a procedural content generation task and propose an architecture for automatically balancing of tile-based levels within the PCGRL framework (procedural content generation via reinforcement learning). Our architecture is divided into three parts: (1) a level generator, (2) a balancing agent, and (3) a reward modeling simulation. Through repeated simulations, the balancing agent receives rewards for adjusting the level towards a given balancing objective, such as equal win rates for all players. To this end, we propose new swap-based representations to improve the robustness of playability, thereby enabling agents to balance game levels more effectively and quickly compared to traditional PCGRL. By analyzing the agent's swapping behavior, we can infer which tile types have the most impact on the balance. We validate our approach in the Neural MMO (NMMO) environment in a competitive two-player scenario. In this extended conference paper, we present improved results, explore the applicability of the method to various forms of balancing beyond equal balancing, compare the performance to another search-based approach, and discuss the application of existing fairness metrics to game balancing.

Knowing the printer model used to print a given document may provide a crucial lead towards identifying counterfeits or conversely verifying the validity of a real document. Inkjet printers produce probabilistic droplet patterns that appear to be distinct for each printer model and as such we investigate the utilization of droplet characteristics including frequency domain features extracted from printed document scans for the classification of the underlying printer model. We collect and publish a dataset of high resolution document scans and show that our extracted features are informative enough to enable a neural network to distinguish not only the printer manufacturer, but also individual printer models.

Despite tremendous progress, machine learning and deep learning still suffer from incomprehensible predictions. Incomprehensibility, however, is not an option for the application of (deep) reinforcement learning in the real world, as unpredictable actions can seriously harm the individuals involved. In this work, we propose a genetic programming framework to generate explanations for the decision-making process of already trained agents by imitating them with programs. Programs are interpretable and can be executed to generate explanations of why the agent chooses a particular action. Furthermore, we conduct an ablation study, which investigates how extending the domain-specific language by using library learning changes the performance of the method. We compare our results with the previous state of the art for this problem and show that we are comparable in performance but require much less hardware resources and computation time.

Understanding the interactions of agents trained with deep reinforcement learning is crucial for deploying agents in games or the real world. In the former, unreasonable actions confuse players. In the latter, that effect is even more significant, as unexpected behavior cause accidents with potentially grave and long-lasting consequences for the involved individuals. In this work, we propose using program synthesis to imitate reinforcement learning policies after seeing a trajectory of the action sequence. Programs have the advantage that they are inherently interpretable and verifiable for correctness. We adapt the state-of-the-art program synthesis system DreamCoder for learning concepts in grid-based environments, specifically, a navigation task and two miniature versions of Atari games, Space Invaders and Asterix. By inspecting the generated libraries, we can make inferences about the concepts the black-box agent has learned and better understand the agent's behavior. We achieve the same by visualizing the agent's decision-making process for the imitated sequences. We evaluate our approach with different types of program synthesizers based on a search-only method, a neural-guided search, and a language model fine-tuned on code.

The balancing process for game levels in a competitive two-player context involves a lot of manual work and testing, particularly in non-symmetrical game levels. In this paper, we propose an architecture for automated balancing of tile-based levels within the recently introduced PCGRL framework (procedural content generation via reinforcement learning). Our architecture is divided into three parts: (1) a level generator, (2) a balancing agent and, (3) a reward modeling simulation. By playing the level in a simulation repeatedly, the balancing agent is rewarded for modifying it towards the same win rates for all players. To this end, we introduce a novel family of swap-based representations to increase robustness towards playability. We show that this approach is capable to teach an agent how to alter a level for balancing better and faster than plain PCGRL. In addition, by analyzing the agent's swapping behavior, we can draw conclusions about which tile types influence the balancing most. We test and show our results using the Neural MMO (NMMO) environment in a competitive two-player setting.

Generating explanations for the behavior of artificial or living agents is still an underexplored problem. We propose a new framework to make agent behavior explainable based on program synthesis. Programs have the advantage that they are inherently interpretable and can be verified for correctness. In detail, we use neural program synthesis with a language model in combination with library learning. In addition, we introduce a domain-agnostic curriculum that is necessary for the system to learn at all. We compare our methods with traditional and state-of-the-art program synthesis systems and justify the necessity of the curriculum and library learning module in ablation studies, followed by an analysis of the extracted libraries.

This work shows that it is possible to learn Logical Program Policies (LPP) for the Multi-Agent Reinforcement Learning environment called Level-based Foraging. Policies are derived from decision trees that use a logical combination of small feature detector programs generated by a probabilistic context-free grammar. We show that these created programmatic policies allow to explain decisions of all agents in the environment and also can be used without retraining on all other environments. This is possible by creating not just one program for each agent, but several, by sampling learned programs and selecting the best programs according to the approximate reward in the specific environment the program was learned on. By using LPP to represent the policies for agents, the policies are interpretable and also extrapolate to unseen environments