Probabilistic space–time prisms (PSTPs) offer a method for calculating probabilities of interaction between two or more individuals. Previously, PSTP methods have been used to calculate probabilities of being colocated or meeting in space. However, other types of interaction are ecologically important to organisms, as interaction can be physical or proximal. In this paper, we develop protocols for using PSTPs to calculate proximal and visual interactions, defined using both distance-based intervals and visibility analysis. First, PSTPs for multiple individuals with overlapping tracking intervals are generated. Second, equations are used to quantify the probabilities of proximity or visibility between pairs of locations in their respective space–time disks. Probabilities are calculated for each pair of potential locations for the individuals and then can be summarized by time step or aggregated across different temporal scales. We demonstrate the approach using tracking data for 11 pairs, or dyads, of ducks tracked over hour-long time periods. The results offer insights into how to visualize and interpret proximal and visual interactions of animals.

Amid the rapid proliferation of thousands of new websites daily, distinguishing safe ones from potentially harmful ones has become an increasingly complex task. These websites often collect user data, and, without adequate cybersecurity measures such as the efficient detection and classification of malicious URLs, users’ sensitive information could be compromised. This study aims to develop models based on machine learning algorithms for the efficient identification and classification of malicious URLs, contributing to enhanced cybersecurity. Within this context, this study leverages support vector machines (SVMs), random forests (RFs), decision trees (DTs), and k-nearest neighbors (KNNs) in combination with Bayesian optimization to accurately classify URLs. To improve computational efficiency, instance selection methods are employed, including data reduction based on locality-sensitive hashing (DRLSH), border point extraction based on locality-sensitive hashing (BPLSH), and random selection. The results show the effectiveness of RFs in delivering high precision, recall, and F1 scores, with SVMs also providing competitive performance at the expense of increased training time. The results also emphasize the substantial impact of the instance selection method on the performance of these models, indicating its significance in the machine learning pipeline for malicious URL classification

Weather forecasting has always been challenging due to the atmosphere’s complex and dynamic nature. Weather conditions such as rain, clouds, clear skies, and sunniness are influenced by several factors, including temperature, pressure, humidity, wind speed, and direction. Physical and complex models are currently used to determine weather conditions, but they have their limitations, particularly in terms of computing time. In recent years, supervised machine learning methods have shown great potential in predicting weather events accurately. These methods use historical weather data to train a model, which can then be used to predict future weather conditions. This study enhances weather forecasting by employing four supervised machine learning techniques—artificial neural networks (ANNs), support vector machines (SVMs), random forest (RF), and k-nearest neighbors (KNN)—on three distinct datasets obtained from the Weatherstack database. These datasets, with varying temporal spans and uncertainty levels in their input features, are used to train and evaluate the methods. The results show that the ANN has superior performance across all datasets. Furthermore, when compared to Weatherstack’s weather prediction model, all methods demonstrate significant improvements. Interestingly, our models show variance in performance across different datasets, particularly those with predicted rather than observed input features, underscoring the complexities of handling data uncertainty. The study provides valuable insights into the use of supervised machine learning techniques for weather forecasting and contributes to the development of more precise prediction models.

Rooftop photovoltaic panels (RPVs) are being increasingly used in urban areas as a promising means of achieving energy sustainability. Determining proper layouts of RPVs that make the best use of rooftop areas is of importance as they have a considerable impact on the RPVs performance in efficiently producing energy. In this study, a new spatial methodology for automatically determining the proper layouts of RPVs is proposed. It aims to both extract planar rooftop segments and identify feasible layouts with the highest number of RPVs in highly irradiated areas. It leverages digital surface models (DSMs) to consider roof shapes and occlusions in placing RPVs. The innovations of the work are twofold: (a) a new method for plane segmentation, and (b) a new method for optimally placing RPVs based on metaheuristic optimization, which best utilizes the limited rooftop areas. The proposed methodology is evaluated on two test sites that differ in urban morphology, building size, and spatial resolution. The results show that the plane segmentation method can accurately extract planar segments, achieving 88.7% and 99.5% precision in the test sites. In addition, the results indicate that complex rooftops are adequately handled for placing RPVs, and overestimation of solar energy potential is avoided if detailed analysis based on panel placement is employed.

Rooftop photovoltaics have been acknowledged as a critical component in cities’ efforts to reduce their reliance on fossil fuels and move towards energy sustainability. Identifying rooftop areas suitable for installing rooftop photovoltaics-referred to as utilizable areas-is essential for effective energy planning and developing policies related to renewable energies. Utilizable areas are greatly affected by the size, shape, superstructures of rooftops, and shadow effects. This study estimates utilizable areas and solar energy potential of rooftops by considering the mentioned factors. First, rooftops are extracted from LiDAR data by training PointNet++, a neural network architecture for processing 3D point clouds. The second step involves extracting planar segments of rooftops using a combination of clustering and region growing. Finally, utilizable areas of planar segments are identified by removing areas that do not have a suitable size and do not receive sufficient solar irradiation. Additionally, in this step, areas reserved for accessibility to photovoltaics are removed. According to the experimental results, the methods have a high success rate in rooftop extraction, plane segmentation, and, consequently, estimating utilizable areas for photovoltaics.

Rooftop solar energy has long been regarded as a promising solution to cities’ growing energy demand and environmental problems. A reliable estimate of rooftop solar energy facilitates the deployment of photovoltaics and helps formulate renewable-related policies. This reliable estimate underpins the necessity of accurately pinpointing the areas utilizable for mounting photovoltaics. The size, shape, and superstructures of rooftops as well as shadow effects are the important factors that have a considerable impact on utilizable areas. In this study, the utilizable areas and solar energy potential of rooftops are estimated by considering the mentioned factors using a three-step methodology. The first step involves training PointNet++, a deep network for object detection in point clouds, to recognize rooftops in LiDAR data. Second, planar segments of rooftops are extracted using clustering. Finally, areas that receive sufficient solar irradiation, have an appropriate size, and fulfill photovoltaic installation requirements are identified using morphological operations and predefined thresholds. The obtained results show high accuracy for rooftop extraction (93%) and plane segmentation (99%). Moreover, the spatially detailed analysis indicates that 17% of rooftop areas are usable for photovoltaics.

The considerable potential of rooftop photovoltaics (RPVs) for alleviating the high energy demand of cities has made them a proven technology in local energy networks. Identification of rooftop areas suitable for installing RPVs is of importance for energy planning. Having these suitable areas referred to as utilizable areas greatly assists in a reliable estimate of RPVs energy production. Within such a context, this research aims to propose a spatially detailed methodology that involves (a) automatic extraction of buildings footprint, (b) automatic segmentation of roof faces, and (c) automatic identification of utilizable areas of roof faces for solar infrastructure installation. Specifically, the innovations of this work are a new method for roof face segmentation and a new method for the identification of utilizable rooftop areas. The proposed methodology only requires digital surface models (DSMs) as input, and it is independent of other auxiliary spatial data to become more functional. A part of downtown Gothenburg composed of vegetation and high-rise buildings with complex shapes was selected to demonstrate the methodology performance. According to the experimental results, the proposed methodology has a high success rate in building extraction (about 95% correctness and completeness) and roof face segmentation (about 85% completeness and correctness). Additionally, the results suggest that the effects of roof occlusions and roof superstructures are satisfactorily considered in the identification of utilizable rooftop areas. Thus, the methodology is practically effective and relevant for the detailed RPVs assessments in arbitrary urban regions where only DSMs are accessible.

In cities where land availability is limited, rooftop photovoltaic panels (RPVs) offer high potential for satisfying concentrated urban energy demand by using only rooftop areas. However, accurate estimation of RPVs potential in relation to their spatial distribution is indispensable for successful energy planning. Classification, plane segmentation, and spatial analysis are three important aspects in this context. Classification enables extracting rooftops and allows for estimating solar energy potential based on existing training samples. Plane segmentation helps to characterize rooftops by extracting their planar patches. Additionally, spatial analyses enable the identification of rooftop utilizable areas for placing RPVs. This dissertation aims to address some issues associated with these three aspects, particularly (a) training support vector machines (SVMs) in large datasets, (b) plane segmentation of rooftops, and (c) identification of utilizable areas for RPVs. SVMs are among the most potent classifiers and have a solid theoretical foundation. However, they have high time complexity in their training phase, making them inapplicable in large datasets. Two new instance selection methods were proposed to accelerate the training phase of SVMs. The methods are based on locality-sensitive hashing and are capable of handling large datasets. As an application, they were incorporated into a rooftop extraction procedure, followed by plane segmentation. Plane segmentation of rooftops for the purpose of solar energy potential estimation should have a low risk of overlooking superstructures, which play an essential role in the placement of RPVs. Two new methods for plane segmentation in high-resolution digital surface models were thus developed. They have an acceptable level of accuracy and can successfully extract planar segments by considering superstructures. Not all areas of planar segments are utilizable for mounting RPVs, and some factors may further limit their useability. Two spatial methods for identifying RPV-utilizable areas were developed in this realm. They scrutinize extracted planar segments by considering panel installation regulations, solar irradiation, roof geometry, and occlusion, which are necessary for a realistic assessment of RPVs potential. All six proposed methods in this thesis were thoroughly evaluated, and the experimental results show that they can successfully achieve the objectives for which they were designed.

I städer där marktillgången är begränsad erbjuder takmonterade solpaneler (eng. rooftop photovoltaic panels) ett attraktivt alternativ för att tillfredsställa höga energibehov. Noggrann värdering av deras potantial f förhållande till spatial utbredning och variation är dock oumbärlig för framgångsrik energi planering. För detta krävs klassificering och segmentering av plana ytor samt spatial analys. Klassificering möjliggör extrahering av hustak och uppskattning av solenergipotentialen baserat på befintliga träningsprov. Segmentering i plan hjälper till att karakterisera hustaken genom extrahering av deras plana segment och spatial analys möjliggör identifiering av användbara takytor för placering av takmonterade solpaneler. Denna avhandlings syfte ät att adressera olika problem associerade med dessa; särskilt: (a) träning av stödvektormaskiner (eng. support vector machines) för stora datamängder, (b) segmentering i plan av hustakspunkter och (c) identifiering av lämpliga ytor för placering av takmonterade solpaneler. Stödvektormaskiner tillhör de mest kraftfulla klassificeringsmetoderna och vilar på en solid teoretisk grund. Men på grund av högkomplexitet under träningsfasen är de tidskrävande, vilket gör dem olämpliga för stora datamängder. Två nya initiala urvalsmetoder (eng. instance selection methods) för data föresås för att påskynda träningsfasen i stödvektormaskiner. Metoderna är baserade på lokalitetskänslig hashning och kan hantera stora datamängder. De inkorpooreras i en applikation i form av extrahering av takyta föjlt av segmentering i plan. Segmentering av hustak för uppskattning av solenergipotential bör inkludera låg risk att förbise överbyggnader, som spelar en viktig roll vid placering av takmonterade solpaneler. Två nya metoder för segmentering i plan för högupplösta digitala ytmodeller har därför utvecklats. De har en acceptabel nivå av noggrannhet och kan framgångsrikt extrahera plana segment genom att ta hänsyn till överbyggnader. Alla ytor med extraherade plana segment är dock inte användbara för montering av takmonterade solpaneler, samtidigt som andra faktorer ytterligare kan begränsa ytornas användbarhet. Två spatiala metoder för att identifiera användbara takmonterade solpanelytor har utvecklats för detta ändamål. De granskar extraherade plana segment genom att ta hänsyn till regler för panelinstallationer, solinstrålning, takgeometri och ocklusion, vilket är nödvändigt för en realistisk bedömning av potentialen av takmonterade solpaneler. Samtliga sex föreslagna metoder i denna studie har utvärderats noggrant och de experimentella resultaten visar att de framgångsrikt kan uppnå de mål som de utformades för.

Support vector machines (SVMs) are powerful classifiers that have high computational complexity in the training phase, which can limit their applicability to large datasets. An effective approach to address this limitation is to select a small subset of the most representative training samples such that desirable results can be obtained. In this study, a novel instance selection method called border point extraction based on locality-sensitive hashing (BPLSH) is designed. BPLSH preserves instances that are near the decision boundaries and eliminates nonessential ones. The performance of BPLSH is benchmarked against four approaches on different classification problems. The experimental results indicate that BPLSH outperforms the other methods in terms of classification accuracy, preservation rate, and execution time. The source code of BPLSH can be found in https://github.com/mohaslani/BPLSH. 

Training support vector machines (SVMs) for pixel-based feature extraction purposes from aerial images requires selecting representative pixels (instances) as a training dataset. In this research, locality-sensitive hashing (LSH) is adopted for developing a new instance selection method which is referred to as DR.LSH. The intuition of DR.LSH rests on rapidly finding similar and redundant training samples and excluding them from the original dataset. The simple idea of this method alongside its linear computational complexity make it expeditious in coping with massive training data (millions of pixels). DR.LSH is benchmarked against two recently proposed methods on a dataset for building extraction with 23,750,000 samples obtained from the fusion of aerial images and point clouds. The results reveal that DR.LSH outperforms them in terms of both preservation rate and maintaining the generalization ability (classification loss). The source code of DR.LSH can be found in https://github.com/mohaslani/DR.LSH.