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.

Predicting future video frames is a challenging task with many downstream applications. Previous work [1], [2] has shown that procedural knowledge enables deep models for complex dynamical settings, however their model ViPro assumed a given ground truth initial symbolic state. We show that this approach led to the model learning a shortcut that does not actually connect the observed environment with the predicted symbolic state, resulting in the inability to estimate states given an observation if previous states are noisy. In this work, we add several improvements to ViPro that enables the model to correctly infer states from observations without providing a full ground truth state in the beginning. We show that this is possible in an unsupervised manner, and extend the original Orbits dataset with a 3D variant to close the gap to real world scenarios.

Using a mono camera, in ADAS, as the primary sensor offers a significant advantage by minimizing system complexity since no extra fusion step is considered. This approach also leverages recent advancements in computer vision and deep learning, which enable high levels of environmental understanding and scene analysis from visual data alone. As such, mono-camera setups hold promise for achieving reliable perception at scale and lead to a growing interest in monocular approaches, particularly for detecting and tracking dynamic objects. However, traditional mono-camera methods are often dependent on auxiliary inputs from GPS or maps, which may be unreliable in complex terrain or areas with poor signal coverage. In this work, we introduce an effective monocular 3D multi-object tracking approach, called MonoSORT3D, which operates without requiring additional auxiliary inputs. Evaluation of our method on the KITTI and MOT17 datasets demonstrates the competitive performance against state-of-the-art methods. Additionally, we provide an in-depth analysis of MonoSORT3D architecture by conducting an ablation study on different components within.

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.

High-definition road maps play a crucial role in the functionality and verification of highly automated driving functions. These contain precise information about the road network, geometry, condition, as well as traffic signs. Despite their importance for the development and evaluation of driving functions, the generation of high-definition maps is still an ongoing research topic. While previous work in this area has primarily focused on the accuracy of road geometry, we present a novel approach for automated large-scale map generation for use in industrial applications. Our proposed method leverages a minimal number of external information about the road to process LiDAR data in segments. These segments are subsequently combined, enabling a flexible and scalable process that achieves high-definition accuracy. Additionally, we showcase the use of the resulting OpenDRIVE in driving function simulation.

Meta-learning has made tremendous progress in recent years and was demonstrated to be particularly suitable in low-resource settings where training data is very limited. However, meta-learning models still require large amounts of training tasks to achieve good generalisation. Since labelled training data may be sparse, self-supervision-based approaches are able to further improve performance on downstream tasks. Although no labelled data is necessary for this training, a large corpus of unlabelled text needs to be available. In this paper, we improve on recent advances in meta-learning for natural language models that allow training on a diverse set of training tasks for few-shot, low-resource target tasks. We introduce a way to generate new training data with the need for neither more supervised nor unsupervised datasets. We evaluate the method on a diverse set of NLP tasks and show that the model decreases in performance when trained on this data without further adjustments. Therefore, we introduce and evaluate two methods for regularising the training process and show that they not only improve performance when used in conjunction with the new training data but also improve average performance when training only on the original data, compared to the baseline.

Recently, leveraging pre-trained Transformer based language models in down stream, task specific models has advanced state of the art results in natural language understanding tasks.However, only a little research has explored the suitability of this approach in low resource settings with less than 1,000 training data points.In this work, we explore fine-tuning methods of BERT -a pre-trained Transformer based language model -by utilizing pool-based active learning to speed up training while keeping the cost of labeling new data constant.Our experimental results on the GLUE data set show an advantage in model performance by maximizing the approximate knowledge gain of the model when querying from the pool of unlabeled data.Finally, we demonstrate and analyze the benefits of freezing layers of the language model during fine-tuning to reduce the number of trainable parameters, making it more suitable for low-resource settings.

One obstacle that so far prevents the introduction of machine learning models primarily in critical areas is the lack of explainability. In this work, a practicable approach of gaining explainability of deep artificial neural networks (NN) using an interpretable surrogate model based on decision trees is presented. Simply fitting a decision tree to a trained NN usually leads to unsatisfactory results in terms of accuracy and fidelity. Using L1-orthogonal regularization during training, however, preserves the accuracy of the NN, while it can be closely approximated by small decision trees. Tests with different data sets confirm that L1-orthogonal regularization yields models of lower complexity and at the same time higher fidelity compared to other regularizers.

Trading-card games are an interesting problem domain for Game AI, as they feature some challenges, such as highly variable game mechanics, that are not encountered in this intensity in many other genres. We present an expert system forming a player-level AI for the digital trading-card game Hearthstone. The bot uses a symbolic approach with a semantic structure, acting as an ontology, to represent both static descriptions of the game mechanics and dynamic game-state memories. Methods are introduced to reduce the amount of expert knowledge, such as popular moves or strategies, represented in the ontology, as the bot should derive such decisions in a symbolic way from its knowledge base. We narrow down the problem domain, selecting the relevant aspects for a play-to-win bot approach and comparing an ontology-driven approach to other approaches such as machine learning and case-based reasoning. On this basis, we describe how the semantic structure is linked with the game-state and how different aspects, such as memories, are encoded. An example illustrates how the bot, at runtime, uses rules and queries on the semantic structure combined with a simple utility system to do reasoning and strategic planning. Finally, an evaluation is presented that was conducted by fielding the bot against the stock “Expert” AI that Hearthstone is shipped with, as well as human opponents of various skill levels in order to assess how well the bot plays. A pseudo Turing test was used to evaluate the believability of the bot's reasoning.

We show that the generator matrix of a generalized concatenated code (GCC code) of order L consists of L submatrices, where the lth submatrix is the Kronecker product of the generator matrices of the lth inner code and the lth outer code. In a similar way we show that the parity-check matrix of a generalized error location code (GEL code) of order L consists of L submatrices, where the lth submatrix is the Gronecker product of the parity-check matrices of the lth inner code and the lth outer code. Then we use these defining matrices to show that for any GCC code there exists an equivalent GEL code and vice versa.