The misdirected approach of open source algorithms

Variations in Earth's orbit and axial tilt induce climatic changes on its surface, which are recorded in sedimentary deposits. These cyclical variations are driven by the main secular frequencies of the solar system. Analyzing these cycles in sedimentary records can help establish an astronomical time scale for the geological record by correlating geological proxies with computed variations in insolation on the Earth's surface, in accordance with the laws of celestial mechanics. A critical aspect of this analysis is the estimation of the sedimentary deposition rate, which determines the time-depth transfer function, relating geological depth to relative or absolute time.We propose a novel approach for constructing astronomical time scales for geological stratigraphic records. The AstroGeoFit method establishes a time-depth transfer function throughout the record, accommodating variable sedimentation rates, and extracts the primary astronomical signal from the geological sequence. This is achieved using a genetic algorithm that adapts to a wide range of sedimentation rate variations. This statistical analysis enables the reconstruction of an astronomical signal (e.g., eccentricity and/or precession) purely from the stratigraphic sequence with minimal personal bias. When this template is correlated with an astronomical solution, an absolute time scale is obtained for the entire record. In addition, we show that quantitative uncertainties can be estimated at each stage of the AstroGeoFit process with the Bayesian approach.The publication of the AstroGeoFit approach will be associated to an open source python package that fully implement the AstroGeoFit algorithm.Ref: J. Laskar, N. Hoang, N. Hara, Y. Wu, A.Sultanov, M. Sinnesael, T. Westerhold, P. Bujons, AstroGeoFit. A Genetic Algorithm and Bayesian approach for  the Astronomical Calibration of the Geological Timescale, 2025, in revision. 

The Open/ADF tool allows the evaluation of derivatives of functions defined by a Fortran program. The derivative evaluation is performed by a Fortran code resulting from the analysis and transformation of the original program that defines the function of interest. Open/ADF has been designed with a particular emphasis on modularity, flexibility, and the use of open source components. While the code transformation follows the basic principles of automatic differentiation, the tool implements new algorithmic approaches at various levels, for example, for basic block preaccumulation and call graph reversal. Unlike most other automatic differentiation tools, Open/ADF uses components provided by the Open/AD framework, which supports a comparatively easy extension of the code transformations in a language-independent fashion. It uses code analysis results implemented in the OpenAnalysis component. The interface to the language-independent transformation engine is an XML-based format, specified through an XML schema. The implemented transformation algorithms allow efficient derivative computations using locally optimized cross-country sequences of vertex, edge, and face elimination steps. Specifically, for the generation of adjoint codes, Open/ADF supports various code reversal schemes with hierarchical checkpointing at the subroutine level. As an example from geophysical fluid dynamics, a nonlinear time-dependent scalable, yet simple, barotropic ocean model is considered. OpenAD/F's reverse mode is applied to compute sensitivities of some of the model's transport properties with respect to gridded fields such as bottom topography as independent (control) variables.

<p>Shoreline variability is a key factor in coastal morphodynamic studies. Beaches act as natural buffers to wave energy, protecting the areas behind them from damage and flooding. In the last decade, remote sensing techniques (video monitoring, shore-based radar, airborne LIDAR, AUVs) are widely applied in coastal studies and several algorithms for shoreline detection have been developed to extract the so called Satellite Derived Shorelines (SDS). Multispectral satellites provide images that cover large areas with high spatial and temporal resolution allowing to perform a near real-time analysis of shorelines worldwide. The main techniques applied to EO-derived images are either manual shoreline detection or image-processing techniques. There are several open source algorithms (e.g. SHOREX and CoastSat) for shoreline detection at sub-pixel level, using available free open-source multispectral images (Landsat and Sentinel constellations). Both algorithms use the three visible bands, the near infrared band, and the short-wave infrared band.</p><p>In this study we tested the performance of the CoastSat algorithm on two different microtidal beaches of the Italian Adriatic coast (Emilia-Romagna and Marche Regions): Punta Marina (PM) and Sirolo (SIR). While PM is a typical intermediate fine sandy beach, SIR is a mixed coarse sand-gravel reflective one. Their mean foreshore slopes are respectively 0.09 and 0.16. At PM, SDS were compared with RTK-DGPS surveyed shorelines measured following the upper limit of the swash zone. The surveys were coincident with Landsat-5, Landsat-7 and Sentinel-2 satellite overpasses on 26/05/2011, 21/01/2020 and 13/02/2020. In the SIR beach case, the SDS were compared with those obtained by a video monitoring station, after manual mapping on variance images on 09/05/2010, 18/04/2011 and 29/06/2011, coincident with Landsat-5 and Landsat-7 overpasses. CoastSat detects the shoreline by classifying the pixels images into four categories (water, white-water, sand and other land features) using a Multilayer Perceptron. As the default settings may not be suitable for every beach, due to different luminosity conditions and sand colour, we specifically trained the classifier with PM and SIR images. The influence on the identification of the SDS shorelines by the run-up extent and beach state was evaluated.</p><p>The obtained RMSE ranges between ~ 6.5 and 14 m at both sites, comparable to the values found by CoastSat developers, indicating that the shoreline is effectively obtained at sub-pixel level. Our results suggest that in the SIR case, the magnitude of the errors can be correlated with the hydrodynamic conditions, as they increase in pair with the run-up extension. This could be explained by the fact that on a reflective beach, with coarser sediments, waves break on the beachface and the water percolates delimiting a clear shoreline, with a distinguishable edge. This correlation was not found in PM, suggesting a bad performance in sand-water classification when the classifier has to deal with a wider swash zone with saturated sand.</p><p>The research received funding from the EU H2020 program under grant agreement 101004211-ECFAS Project.</p>

The German National Library (DNB) began developing solutions for automatic subject cataloguing 15 years ago. The main reason for this was the huge and ever-growing number of digital media works that needed to be indexed. Today, the DNB uses open source algorithms and frameworks to assign various types of thematic meta information in this way. This practice paper provides a deeper insight into automatic subject cataloguing at the DNB. We look at the data and vocabularies used as well as at the different methods and approaches. The vocabulary for classification is based on the Dewey Decimal Classification (DDC). For verbal subject indexing we use the German Integrated Authority File (GND). The use case of automatic classification is divided into the assignment of DDC Subject Categories and DDC Short Numbers. Due to the large size of the GND vocabulary, the use case of automatic indexing is an extreme multi-label classification (XMLC) problem. A brief report is given about the construction and the performance of our models. Based on these use cases, we present some implementation aspects of our “subject cataloguing machine” EMa, the environment for automatic subject cataloguing in productive use. We point out the basic feature set and provide a high-level introduction of the productive EMa system. The modular design of the EMa software architecture with the open source software Annif as a central toolkit is described. The development of EMa is an ongoing task at the DNB. It requires continuous development and maintenance, technological and human resources. Applied research activities in the DNB's AI project are closely related to the EMa ensuring that relevant scientific findings get integrated into its development.

PurposeDistal radius fractures (DRFs) are often initially assessed by junior doctors under time constraints, with limited supervision, risking significant consequences if missed. Convolutional Neural Networks (CNNs) can aid in diagnosing fractures. This study aims to internally and externally validate an open source algorithm for the detection and localization of DRFs.MethodsPatients from a level 1 trauma center from Adelaide, Australia that presented between 2016 and 2020 with wrist trauma were retrospectively included. Radiographs were reviewed confirming the presence or absence of a fracture, as well as annotating radius, ulna, and fracture location. An internal validation dataset from the same hospital was created. An external validation set was created with two other level 1 trauma centers, from Groningen and Rotterdam, the Netherlands. Three surgeons reviewed both sets for DRFs.ResultsThe algorithm was trained on 659 radiographs. The internal validation set included 190 patients, showing an accuracy of 87% and an AUC of 0.93 for DRF detection. The external validation set consisted of 188 patients, with an accuracy and AUC were 82% and 0.88 respectively. Radial and ulnar bone segmentation on the internal validation was excellent with an AP50 of 99 and 98, but moderate for fracture segmentation with an AP50 of 29. For external validation the AP50 was 92, 89 and 25 for radius, ulna, and fracture respectively.ConclusionThis open-source algorithm effectively detects DRFs with high accuracy and localizes them with moderate accuracy. It can assist clinicians in diagnosing suspected DRFs and is the first radiograph-based CNN externally validated on patients from multiple hospitals.

Currently, password-based authentication is the most widespread identity verification method for web pages access. However it presents security issues due to the growth of attacks based on spywares and phishing. The main purpose of both techniques is the digital identity theft, that is, stealing users' passwords in an unnoticed way. In order to counter this type of attack, many financial institutions have adopted strong authentication, a technique that employs a simultaneous use of different authentication factors. By synergistically combining the advantages of distinct factors, such arrangement results in the mutual mitigation of the vulnerabilities of each one, yielding an architecturally safer identity verification method. This work presents the design and implementation of an authentication device, which combines passwordbased and object-based authenticators. Its main distinguishing features are the reduced cost and the use of open sourced cryptographic algorithms. Open source algorithms have their security widely and independently verified, a characteristic that helps increase the system's reliability, since third parties may check the source code running on the device.

The Penn State Taxonomic Tags group, with representatives from Information Technology, Business Administration, and the Penn State Libraries, was formed to examine whether a taxonomic set of tags, systematically applied across the university's Web pages, could (a) make finding specific pages easier from among the University's greater than 500,000 Web pages, (b) simplify Web content management tasks and (c) prove useful over time as search engines continue to evolve and despite whether open source or commercial (and often, proprietary) search algorithms are employed. The University has had broad experience with several search engines, and currently holds a University-wide license for the Google Appliance. The Tags Group has developed recommendations that it believes will address issues found in the current environment and yet remain useful during and after what it expects will be the increasing adoption of Content Management Systems across Penn State University.

Template-Protection-Techniken stellen eine wichtige Ergänzung zu biometrischen Systemen, um das gewünschte Sicherheitsniveau zu erreichen und die Anforderungen an den Schutz der Privatsphäre zu erfüllen. Diese Techniken ermöglichen Erneuerung und Widerruf der Referenzendaten in biometrischen Systemen. In dieser Arbeit wird ein systematisches Evaluierungsframework entwickelt, das sich an die wesentlichen Kriterien und Anforderungen an biometrische Template-Protection-Verfahren festhält. Seine Anwendbarkeit wird durch die Analyse der verschiedenen Algorithmen demonstriert. Die in dieser Arbeit präsentierte Evaluierung ist grundlegend für eine vollständige Analyse. Darüber hinaus ermöglicht es einen Nachweis der Sicherheit und Einhaltung der Privatsphäre. Es schafft eine Basis für Zertifizierung und Benchmark der Template-Protection-Verfahren.

In this work, we tested Simultaneous localization and mapping (SLAM) about mobile robots in indoor environment, where all experiments were conducted based on the Robot Operating System (ROS). The urban search and rescue(USAR) environment was build in the ROS simulation tool Gazebo, and our car was used to test hector SLAM in Gazebo. The rplidar A1 single-line lidar was used for 2D laser scan matching data acquisition in the practical experiments and the indoor map was built by using the open source algorithms gmapping, karto SLAM, hector SLAM software package for indoor SLAM, which can get the indoor grid maps in ROS graphical tool RVIZ. The experimental results of the three open source algorithms show that the mobile robot for simultaneous localization and mapping (SLAM) is feasible, and high-precision grid maps can be constructed.

Time series analysis comprises a wide repertoire of methods for extracting information from data sets. Despite great advances in time series analysis, identifying and quantifying the strength of nonlinear temporal correlations remain a challenge. We have recently proposed a new method based on training a machine learning algorithm to predict the temporal correlation parameter, , of flicker noise (FN) time series. The algorithm is trained using as input features the probabilities of ordinal patterns computed from FN time series, , generated with different values of . Then, the ordinal probabilities computed from the time series of interest, , are used as input features to the trained algorithm and that returns a value, , that contains meaningful information about the temporal correlations present in . We have also shown that the difference, , of the permutation entropy (PE) of the time series of interest, , and the PE of a FN time series generated with , , allows the identification of the underlying determinism in . Here, we apply our methodology to different datasets and analyze how and correlate with well-known quantifiers of chaos and complexity. We also discuss the limitations for identifying determinism in highly chaotic time series and in periodic time series contaminated by noise. The open source algorithm is available on Github.

Machine learning language models are combinations of algorithms and neural networks designed for text processing composed in natural language (Natural Language Processing, NLP). 
 In 2020, the largest language model from the artificial intelligence research company OpenAI, GPT-3, was released, the maximum number of parameters of which reaches 175 billion. The parameterization of the model increased by more than 100 times made it possible to improve the quality of generated texts to a level that is hard to distinguish from human-written texts. It is noteworthy that this model was trained on a training dataset mainly collected from open sources on the Internet, the volume of which is estimated at 570 GB. 
 This article discusses the problem of memorizing critical information, in particular, personal data of individual, at the stage of training large language models (GPT-2/3 and derivatives), and also describes an algorithmic approach to solving this problem, which consists in additional preprocessing training dataset and refinement of the model inference in the context of generating pseudo-personal data and embedding into the results of work on the tasks of summarization, text generation, formation of answers to questions and others from the field of seq2seq.

Color and pattern are integral parts of the visual characteristics of camouflage. These means, taking into account the experience of military operations during the Russian-Ukrainian war, can significantly increase the survivability and safety of personnel, weapons and military equipment, by eliminating the characteristic unmasking signs of these military facilities and hiding them on vegetative, desert-steppe, snowy and urbanized areas background.
 The paper considers the first stage in the design of camouflage means of concealment - the identification of the characteristic colors of the area. The identification of characteristic colors is proposed to be carried out using clustering related to unsupervised machine learning methods. The number of clusters determines the number of colors that will be displayed on the masking surface.
 It was determined that it is advisable to analyze terrain images stored in the digital JPEG format, and the colors are represented in the RGB additive color model.
 When conducting research, such a clustering method for image analysis as k-means was used, which has an advantage over other clustering methods in ease of implementation, unpretentiousness in resources and sufficient computational speed. Other clustering methods, such as hierarchical or density-based, have not proven to be suitable for image clustering. The comparison was made with the most common clustering methods: c-means, DBSCAN, OPTICS, agglomerative, spectral biclustering, etc.
 Various algorithmic approaches to choosing the number of clusters were tested, according to the results of the experiments, the “elbow” method was chosen as the most optimal one.
 Mathematical algorithms were taken from open sources, their implementation was carried out using common software libraries for machine learning of the Python programming language.
 The results of the work made it possible to choose mathematical algorithms for determining the number of colors of camouflage means of concealment. This will allow to analyze the terrain of all natural zones of Ukraine and design effective camouflage coverings for the Armed Forces of Ukraine.

Marine domain awareness is important for crisis management in case of accidents but also as a surveillance tool against illicit discharge activities. Copernicus Sentinel-1 data are used for the detection of spills as they are provided free and in near real time. The availability of data and the recently developed cloud infrastructures give the opportunity for developing open source algorithms and techniques for oil spill detection. In this study a simple semi-automatic system for oil spill detection is presented for Sentinel-1 images. After image segmentation, the Orfeo Toolbox SVM vector classifier is trained based on the features extracted and separates the objects to possible oil spills and lookalikes. The method has been tested using three images depicting the evolution of an oil spill that resulted from a ship collision in the Mediterranean, offshore Corsica, on the 7th October 2018. Results were sound but further development is needed especially on algorithm training and feature extraction. All the software packages used are free and open source.

Smart charging assigns charging capacities between vehicles in limited charging infrastructures. Smart charging solutions are becoming widespread with increasing numbers of commercial offerings. However, commercial solutions are intransparent regarding algorithm implementation. In contrast, open source solutions are transparent and open for collaborative development and scientific research. This article presents an open source package with a smart charging algorithm. The algorithm schedules heterogeneous fleets of vehicles for charging while ensuring a fair share for each vehicle. We present implementation aspects of the smart charging algorithm including data structures and representational state transfer application programming interfaces. Additionally, the smart charging algorithm was validated in a one-year field test. Experimental results of the field test show elcetronic vehicles (EVs) at six charging stations can be scheduled for charging when the grid connection only allows two EVs to charge concurrently. Runtime measurements demonstrate the smart charging algorithm is applicable in real-time and scales to large fleet sizes.

ObjectiveDiabetes Mellitus (DM) has reached epidemic levels globally. A contributing factor to the development of DM is high blood glucose (hyperglycemia). One complication associated with DM is a decreased angiogenesis. The Matrigel tube formation assay (TFA) is the most widely utilized in vitro assay designed to assess angiogeneic factors and conditions. In spite of the widespread use of Matrigel TFAs, quantification is labor-intensive and subjective, often limiting experiential design and interpretation of results. This study describes the development and validation of an open source software tool for high throughput, morphometric analysis of TFA images and the validation of an in vitro hyperglycemic model of DM.Approach and ResultsEndothelial cells mimic angiogenesis when placed onto a Matrigel coated surface by forming tube-like structures. The goal of this study was to develop an open-source software algorithm requiring minimal user input (Pipeline v1.3) to automatically quantify tubular metrics from TFA images. Using Pipeline, the ability of endothelial cells to form tubes was assessed after culture in normal or high glucose for 1 or 2 weeks. A significant decrease in the total tube length and number of branch points was found when comparing groups treated with high glucose for 2 weeks versus normal glucose or 1 week of high glucose.ConclusionsUsing Pipeline, it was determined that hyperglycemia inhibits formation of endothelial tubes in vitro. Analysis using Pipeline was more accurate and significantly faster than manual analysis. The Pipeline algorithm was shown to have additional applications, such as detection of retinal vasculature.