Policies focused on waterbody protection and restoration have been suggested to European Union member countries for some time, but to adopt these policies on a large scale the quality of small water channel maps needs considerable improvement. We developed methods to detect and classify small stream and ditch channels using airborne laser scanning and deep learning. The research questions covered the influence of the resolution of the digital elevation model on channel extraction, the efficacy of different terrain indices to identify channels, the potential advantages of combining indices, and the performance of a U-net model in mapping both ditches and stream channels. Models trained in finer resolutions were more accurate than models trained with coarser resolutions. No single terrain index consistently outperformed all others, but some combinations of indices had higher MCC values. Natural stream channels were not classified to the same extent as ditches. The model trained on the 0.5 m resolution had the most balanced performance using a combination of indices trained using the dataset with both types of channel separately. The deep learning model outperformed traditional mapping methods for ditches, increasing the recall from less than 10% to over 92%, while the recall for natural channels was around 71%. However, despite the successful detection of ditches, the models frequently misclassified streams as ditches. This poses a challenge, as natural channels are protected under land use management practices, while ditches are not.

Reliable information on soil-forming parent materials is crucial for informed decision-making in infrastructure planning, land-use management, environmental assessments, and geohazard mitigation. In the northern landscapes previously affected by glacial processes, these parent materials are predominantly Quaternary deposits. This study explored the potential of machine learning to expedite soil parent material mapping in Sweden. Two Extreme Gradient Boosting models were trained, one using terrain and hydrological indices derived from Light Detection and Ranging data, and the other incorporating additional ancillary map data. Both models were trained on 29,588 soil observations and evaluated against a separate hold-out set of 3500 observations. As a baseline, the existing most detailed maps achieved a Matthews Correlation Coefficient of 0.36. The Extreme Gradient Boosting models achieved higher MCC values of 0.45 and 0.56, respectively. To understand spatial variations in model performance, the second model was evaluated across 28 physiographic regions in Sweden. The results revealed that model performance varied across regions and deposit types, with till and peat exhibiting better performance than sorted sediments. These findings underscore the need for region-specific analyses to optimize the application of machine learning in digital soil mapping. © 2024 The Authors

This article compares novel and existing uncertainty quantification approaches for semantic segmentation used in remote sensing applications. We compare the probability estimates produced by a neural network with Monte Carlo dropout-based approaches, including predictive entropy and mutual information, and conformal prediction-based approaches, including feature conformal prediction (FCP) and a novel approach based on conformal regression. The chosen task focuses on identifying ditches and natural streams based on LiDAR derived digital elevation models. We found that FCP's uncertainty estimates aligned best with the neural network's prediction performance, leading to the lowest Area Under the Sparsification Error curve of 0.09. For finding misclassified instances, the network probability was most suitable, requiring a correction of only 3% of the test instances to achieve a Matthews Correlation Coefficient (MCC) of 0.95. Conformal regression produced the best confident maps, which, at 90% confidence, covered 60% of the area and achieved an MCC of 0.82.

On behalf of the Organizing Committee, it is our pleasure to present the proceedings of the 14th Scandinavian Conference on Artificial Intelligence (SCAI). After a break of almost 10 years, SCAI has been reestablished in a collaboration between the Swedish AI Society (SAIS) and the Norwegian AI Society (NAIS). As its predecessors, SCAI aims to bring together researchers and practitioners from the field of AI to present and discuss ongoing work and future directions. The conference provides a platform for networking among researchers as well as building relationships with practitioners, businesses, and other researchers involved in related fields.

In this paper, we focus on the crucial task of understanding and modeling pedestrian behavior, which is essential for numerous applications. We introduce a semantic representation of pedestrian crossing behavior. The representation is to capture the sub-events within a behavior and the spatial-temporal evolution of interactions between pedestrians and other objects involved in crossing events. We demonstrate its practical application by utilizing it to analyze pedestrian crossing behavior from road user movement data (i.e. trajectories). By constructing a knowledge graph from detailed road user dynamics data using this representation, we enable queries that address safety concerns related to pedestrian crossing behavior, aiding traffic engineers in their work on urban traffic infrastructure design.

Forests worldwide contain unique cultural traces of past human land use. Increased pressure on forest ecosystems and intensive modern forest management methods threaten these ancient monuments and cultural remains. In northern Europe, older forests often contain very old traces, such as millennia-old hunting pits and indigenous Sami hearths. Investigations have repeatedly found that forest owners often fail to protect these cultural remains and that many are damaged by forestry operations. Current maps of hunting pits are incomplete, and the locations of known pits have poor spatial accuracy. This study investigated whether hunting pits can be automatically mapped using national airborne laser data and deep learning. The best model correctly mapped 70% of all the hunting pits in the test data with an F1 score of 0.76. This model can be implemented across northern Scandinavia and could have an immediate effect on the protection of cultural remains.

This study explores the effects of incorporating demonstrations as pre-training of an improved Deep Q-Network (DQN). Inspiration is taken from methods such as Deep Q-learning from Demonstrations (DQfD), but instead of retaining the demonstrations throughout the training, the performance and behavioral effects of the policy when using demonstrations solely as pre-training are studied. A comparative experiment is performed on two game environments, Gymnasium's Car Racing and Atari Space Invaders. While demonstration pre-training in Car Racing shows improved learning efficacy, as indicated by higher evaluation and training rewards, these improvements do not show in Space Invaders, where it instead under-performed. This divergence suggests that the nature of a game's reward structure influences the effectiveness of demonstration pre-training. Interestingly, despite less pronounced quantitative differences, qualitative observations suggested distinctive strategic behaviors, notably in target elimination patterns in Space Invaders. These retained behaviors seem to get forgotten during extended training. The results show that we need to investigate further how exploration functions affect the effectiveness of demonstration pre-training, how behaviors can be retained without explicitly making the agent mimic demonstrations, and how non-optimal demonstrations can be incorporated for more stable learning with demonstrations.

Digital Twins (DTs), enriched with Artificial Intelligence (AI) and Blockchain technology, promise a revolutionary breakthrough in smart asset management and predictive maintenance in the built environment. This study aims to portray a conceptual framework of Blockchain and AIbased DTs and outline its key characteristics, requirements, and system architecture by composing a functional model using IDEF0. Such an approach is expected to enhance predictive maintenance in building facilities, simplify the management and operation of smart built environments, and ultimately deliver valuable outcomes for facility operators, real estate practitioners, and end-users. 

There is a lack of data-driven training instructions for sports shooters, as instruction has commonly been based on subjective assessments. Many studies have correlated body posture and balance to shooting performance in rifle shooting tasks, but have mostly focused on single aspects of postural control. This study has focused on finding relevant rifle shooting factors by examining the entire body over sequences of time. A data collection was performed with 13 human participants carrying out live rifle shooting scenarios while being recorded with multiple body tracking sensors. A pre-processing pipeline produced a novel skeleton sequence representation, which was used to train a transformer model. The predictions from this model could be explained on a per sample basis using the attention mechanism, and visualised in an interactive format for humans to interpret. It was possible to separate the different phases of a shooting scenario from body posture with a high classification accuracy (80%). Shooting performance could be detected to an extent by separating participants using their strong and weak shooting hand. The dataset and pre-processing pipeline, as well as the techniques for generating explainable predictions presented in this study have laid the groundwork for future research in the sports shooting domain.

The integration of blockchain and digital twins (DT) for better building-lifecycle data management has recently received much attention from researchers in the field. In this respect, the adoption of enabling technologies such as artificial intelligence (AI) and machine learning (ML), the Internet of Things (IoT), cloud and edge computing, Big Data analytics, etc., has also been investigated in an abundance of studies. The present review inspects the recent studies to shed light on the foremost among those enabling technologies and their scope, challenges, and integration potential. To this end, 86 scientific papers, recognized and retrieved from the Scopus and Web of Science databases, were reviewed and a thorough bibliometric analysis was performed on them. The obtained results demonstrate the nascency of the research in this field and the necessity of further implementation of practical methods to discover and prove the real potential of these technologies and their fusion. It was also found that the integration of these technologies can be beneficial for addressing the implementation challenges they face individually. In the end, an abstract descriptive model is presented to provide a better understanding of how the technologies can become integrated into a unified system for smartening the built environment.