Analyzing action scenes in soccer is a challenging task due to the complex and dynamic nature of the game, as well as the interactions between players. This article provides a comprehensive overview of this task, divided into action recognition, spotting key moments, and identifying actions in both time and space (spatio-temporal action localization) in soccer. We explore publicly available data sources and metrics used to evaluate models’ performance. The article reviews recent state-of-the-art methods that leverage deep learning techniques and traditional approaches. Our analysis begins with methods based on feature engineering, followed by an exploration of various deep learning techniques. This includes using Convolutional Neural Networks (CNNs) for visual information processing, Recurrent Neural Networks (RNNs) for analyzing temporal sequences, and transformer architectures to effectively capture context. In particular, we focus on the specifics of multimodal data, illustrating the potential for improved model accuracy and robustness. This includes an exploration of methods that integrate information from multiple sources, such as video and audio data, and methods that represent a single data source through multiple analytical lenses, offering a richer, more nuanced understanding of soccer actions (e.g., using a graph representation of players). Finally, the article highlights some of the open research questions and future directions in the field of soccer action analysis, especially the potential for multimodal methods to advance this field. Overall, this survey provides a valuable resource for researchers interested in the field of analyzing action scenes in soccer.

The LHCb (Large Hadron Collider beauty) experiment is one of the main research areas carried out by the European Organization for Nuclear Research CERN in Geneva. The presented paper describes a novel procedure for the detection of long-lived particles provided for the needs of the above experiment. A problem that is investigated mainly relies on the determination of tracks (trajectories) of such particles based on observations obtained from measurement modules placed far apart; thus, such kind of detection is alternatively named a reconstruction. The main intention of the algorithm presented here constitutes the elimination of erroneously created tracks, the so-called ghost-tracks. The presented concept is based on elements of mathematical statistics (nonparametric estimation) as well as computational intelligence (neural networks, fuzzy approach). In order to measure the obtained effectiveness, an appropriate performance index which does not require information if the given track is real or only a ghost (as it occurs in practice) is defined. Finally, the number of ghost-tracks is reduced threefold in comparison to prior methods used. Illustrative interpretations make the presented concept easy to adopt in similar problems connected with detection/reconstruction of tracks/trajectories of elementary physical particles.

In recent years, significant efforts have been made to address graph node classification tasks by applying graph neural networks and methods based on label propagation.Despite the progress achieved by these approaches, their success often hinges on complex architectures and algorithms, sometimes leading to the oversight of crucial technical details.In designing artificial neural networks, one crucial aspect of the innovative approach is suggesting a novel neural architecture.Currently used architectures have mostly been developed manually by human experts, which is a time-consuming and error-prone process.That is why the adoption of more sophisticated semi-automatic methods, such as Neural Architecture Search, has become commonplace.This paper introduces and assesses an evolutionary-based approach for the design of graph convolutional neural networks in the context of node classification.Our approach aims to systematically define the graph convolutional networks parameter space, drawing inspiration from recent research on design principles.By doing so, our method seeks to strike a balance between achieving satisfactory performance and optimizing memory and computation resources, thus offering a more efficient alternative to conventional approaches from the neural architecture search area.

Misinformation is a significant societal issue with potentially severe consequences. It appears in text, image, audio, and video modalities, encompassing various categories such as unimodal deception (fact-conflicting, AI-generated & offensive content) and cross-modal inconsistencies. However, current detection approaches often focus on text and image, overlooking the growing prevalence of misinformation in audio and video content. Moreover, these methods typically tend to address only one or two types of misinformation, failing to address all categories simultaneously. These detectors are also usually designed to make judgments without providing explanations, reducing transparency and limiting their broader applicability. To address these issues, we propose MDAM3, a Misinformation Detection and Analysis Framework for Multitype Multimodal Media. MDAM3 analyzes each input in internal detection and examines relationships across modalities to identify inconsistencies. It utilizes web resources and integrates Large Vision-Language Models (LVLMs) to deliver accurate detection results along with detailed analysis. To evaluate MDAM3, we curate MDAM3-DB, a specialized multitype multimodal misinformation dataset. A user study is conducted to explore MDAM3's usability, interpretability, and effectiveness. We hope this research contributes to advancing misinformation detection methodologies and provides valuable insights for developing robust multimodal analysis tools.

The aim of this paper is to propose a new image data set for assessing the quality of solutions to machine learning tasks, in particular, deep neural networks. The data set is derived from X-ray images of wheat grains, in which three species, Kama, Rosa, and Canadian, are distinguished. In this paper, the structure of the data is presented in detail and ten pretrained deep neural networks are applied to identify individual wheat species. The Seeds Image Data Set, due to its compact nature, can compete with well-known and quite frequently used object recognition benchmark tasks (CIFAR-10, CIFAR-100, SVHN, and ImageNet, etc.). The compactness of the data set is based on a relatively small number of data instances, which shortens the rather time-consuming computing process. The proposed data set will be made available in a public repository, and the results presented will provide a starting point for other competing solutions for exploratory data analysis in the broad sense.

This study examines whether LLMs can be utilized in metaheuristic tuning through selection of appropriate parameters.Instances of two optimization problems, Travelling Salesman and Graph Coloring, were solved with GA, ACO, PSO, and SA.Experiment involved running these heuristic optimizers with parameter values advised by LLMs.A round of feedback was performed through feeding LLMs with prompts that included initial parameters, average performance, and population variance, where applicable.The results show LLMs exhibit the ability to comprehend the non-trivial task of tuning metaheuristics' parameters.Additionally, feedback runs often outperform results achieved by initial setups, yielding a new application of LLMs.

With the development of various generative models, misinformation in news media becomes more deceptive and easier to create, posing a significant problem. However, existing datasets for misinformation study often have limited modalities, constrained sources, and a narrow range of topics. These limitations make it difficult to train models that can effectively combat real-world misinformation. To address this, we propose a comprehensive, large-scale Multimodal Misinformation dataset for Media Authenticity Analysis ( M 3 A ), featuring broad sources and fine-grained annotations for topics and sentiments. To curate M 3 A , we collect genuine news content from 60 renowned news outlets worldwide and generate fake samples using multiple techniques. These include altering named entities in texts, swapping modalities between samples, creating new modalities, and misrepresenting movie content as news. M 3 A contains 708K genuine news samples and over 6M fake news samples, spanning text, images, audio, and video. M 3 A provides detailed multi-class labels, crucial for various misinformation detection tasks, including out-of-context detection and deepfake detection. For each task, we offer extensive benchmarks using state-of-the-art models, aiming to enhance the development of robust misinformation detection systems. • We present M 3 A , a large-scale multimodal misinformation dataset with diverse news samples. • M 3 A includes texts, images, audio, and videos from multiple reputable news outlets. • M 3 A addresses limitations in existing datasets in misinformation generation methods and scale. • We provide multi-class annotations in M 3 A for various key tasks in misinformation detection. • We propose benchmarks for M 3 A using state-of-the-art models and out-of-distribution testing.

The complexity of industry-grade event-based datalakes grows dynamically each passing hour. Companies actively gather behavioral information on their customers, recording multiple types of events, such as clicks, likes, page views, card transactions, add-to-basket, or purchase events. In response to this, the Synerise Monad platform has been proposed. The primary focus of Monad is to produce Universal Behavioral Representations (UBRs) - large vectors encapsulating the behavioral patterns of each user. UBRs do not lose knowledge about individual events, in contrast to aggregated features or averaged embeddings. They are based on award-winning algorithms developed at Synerise - Cleora and EMDE - and allow to process real-life datasets composed of billions of events in record time. In this paper, we introduce a new aspect of Monad: private foundation models for behavioral data, trained on top of UBRs. The foundation models are trained in purely self-supervised manner and allow to exploit general knowledge about human behavior, which proves especially useful when multiple downstream models must be trained and time constraints are tight, or when labeled data is scarce. Experimental results show that the Monad foundation models can cut training time in half and require 3x less data to reach optimal results, often achieving state-of-the-art results.

Product matching corresponds to the task of matching identical products across different data sources. It typically employs available product features which, apart from being multimodal, i.e., comprised of various data types, might be non-homogeneous and incomplete. The paper shows that pre-trained, multilingual Transformer models, after fine-tuning, are suitable for solving the product matching problem using textual features both in English and Polish languages. We tested multilingual mBERT and XLM-RoBERTa models in English on Web Data Commons - training dataset and gold standard for large-scale product matching. The obtained results show that these models perform similarly to the latest solutions tested on this set, and in some cases, the results were even better.Additionally, we prepared a new dataset – ProductMatch.pl – that is entirely in Polish and based on offers in selected categories obtained from several online stores for the research purpose. It is the first open dataset for product matching tasks in Polish, which allows comparing the effectiveness of the pre-trained models. Thus, we also showed the baseline results obtained by the fine-tuned mBERT and XLM-RoBERTa models on the Polish datasets.

Deep convolutional neural networks have strong data structure mining ability and can be successfully applied to facilitate the feature detection process when applied to porous media analysis. One of the encountered limitations is the lack of large data when applying deep learning algorithms. This work proposes a novel approach to overcome this problem by generating highly representative datasets based on real CT-images. The method uses an original morphological concept for the determination of pore size distribution to provide input data that is to be fed to a designed neural network. The generated data was employed for the classification of soil aggregates that differ in their pore size distribution. The image classification results were achieved by exploiting well-known pre-trained deep learning models: VGG-16, ResNet50, InceptionV3, Dense-Net121, and MobileNet. We applied k-fold cross-validation for$k$equal to five to validate the results. The average accuracy for the validating data achieved for the MobileNet, VGG-16, InceptionV3 and DenseNet121 ranged from 90% to 95% and were 5% higher compared to ResNet50. The deep learning approach has demonstrated great promise for analyzing very complex and irregular structures based on micro-CT images.

Graph Neural Networks (GNNs) are a family of graph networks inspired by mechanisms existing between nodes on a graph. In recent years there has been an increased interest in GNN and their derivatives, i.e., Graph Attention Networks (GAT), Graph Convolutional Networks (GCN), and Graph Recurrent Networks (GRN). An increase in their usability in computer vision is also observed. The number of GNN applications in this field continues to expand; it includes video analysis and understanding, action and behavior recognition, computational photography, image and video synthesis from zero or few shots, and many more. This contribution aims to collect papers published about GNN-based approaches towards computer vision. They are described and summarized from three perspectives. Firstly, we investigate the architectures of Graph Neural Networks and their derivatives used in this area to provide accurate and explainable recommendations for the ensuing investigations. As for the other aspect, we also present datasets used in these works. Finally, using graph analysis, we also examine relations between GNN-based studies in computer vision and potential sources of inspiration identified outside of this field.

Recent years brought an increasing interest in the application of machine learning algorithms in e-commerce, om-nichannel marketing, and the sales industry. It is not only to the algorithmic advances but also to data availability, representing transactions, users, and background product information. Finding products related in different ways, i.e., substitutes and complements is essential for users' recommendations at the vendor's site and for the vendor - to perform efficient assortment optimization. The paper introduces a novel method for finding products' substitutes and complements based on the graph embedding Cleora algorithm. We also provide its experimental evaluation with regards to the state-of-the-art Shopper algorithm, studying the relevance of recommendations with surveys from industry experts. It is concluded that the new approach presented here offers suitable choices of recommended products, requiring a minimal amount of additional information. The algorithm can be used in various enterprises, effectively identifying substitute and complementary product options.

This work presents a comparative analysis of probabilistic neural network training methods applied to achieve best performance in various classification tasks. Two result from classical mathematical methods based on the theory of kernel density estimators: the plug-in method and cross-validation procedure. The other two methods are more advanced: a metaheuristic algorithm of particle swarm optimization, and a procedure based on reinforcement learning. Ten data sets, regarded in eleven classification problems, taken from the UCI repository are used for the numerical analysis. A comparative analysis of probabilistic neural network learning methods leads to interesting conclusions. Although it does not allow for unambiguous selection of the best learning method, it provides a possibility of choosing a method that is adequate for the given conditions. The description of this is included in the work.

Recent growth of metaheuristic search strategies has brought a huge progress in the domain of computational optimization. The breakthrough started since the well-known Particle Swarm Optimization algorithm had been introduced and examined. Optimization technique presented in this contribution mimics the process of flower pollination. It is build on the foundation of the first technique of this kind—known as Flower Pollination Algorithm (FPA). In this paper, its simplified and improved version, obtained after extensive performance testing, is presented. It is based on only one natural phenomena—called flower constancy—the natural mechanism allowing pollen carrying insects to remember the positions of the best pollen sources. Modified FPA, named as Biotic Flower Pollination Algorithm (BFPA) and relying solely on biotic pollinators, outperforms original FPA, which itself proved to be very effective approach. The paper first presents a short description of original FPA and the changes leading to Biotic Flower Pollination Algorithm. It also discusses performance of the modified algorithm on a full set of CEC17 benchmark functions. Furthermore, in that aspect, the comparison between BFPA and other optimization algorithms is also given. Finally, brief exemplary application of modified algorithm in the field of probabilistic modeling, related to physics and engineering, is also presented.

The paper provides an insight into the novel meta-heuristic of the Crow Search Algorithm (CSA) as used for continuous optimization tasks. The presented procedure is inspired by the social behaviour of crows. Based on established CEC 2017 benchmark tasks instances, the paper concentrates on performed experimental parameter studies and on a comparison with the existing Particle Swarm Optimization strategy. Building on the test experiences, sets of internal parameters have been formulated which constitute recommendations for other numerical calculations. Finally, some concluding remarks on possible algorithm extensions are given.

Recommender systems (or recommendation systems) represent a typical decision support system, aimed at supplying users optimal set of options, according to their interests and preferences. Typically user is being presented a list of choices obtained on the basis of algorithmic procedure, which typically employs available knowledge concerning users preferences and characteristics of options at hand. Though usually the most influential factor for the choice of recommendations is presented to the user, no relations between recommended options are frequently being explored.The aim of this paper is to address this issue and introduce novel method of generating descriptive explanations of recom-mender systems result via fuzzy linguistic summaries. Along with the sketch of the method its pilot implementation for movies recommendation is being provided. The discussion of obtained results is followed by areas of possible improvements and plans for further research.

This paper describes the use of a new swarm-based metaheuristic, namely Krill Herd Algorithm (KHA), in computer gaming.In this work, KHA is employed to find a bots movement strategy in a computer racing game.The complete algorithm is implemented using a Unity Engine in C# language.Herein, the triggering of the metaheuristic optimization task was conducted by the way of a KHA internal parameter investigation.In this approach, the goal of the race (the KHA evaluation function) for both the human and computer player is to finish a lap in the shortest time possible.

Dividing a dataset into disjoint groups of homogeneous structure, known as data clustering, constitutes an important problem of data analysis.It can be solved with broad range of methods employing statistical approaches or heuristic procedures.The latter often include mechanisms known from nature as they are known to serve as useful components of effective optimizers.The paper investigates the possibility of using novel nature-inspired technique -Grasshopper Optimization Algorithm (GOA) -to generate accurate data clusterings.As a quality measure of produced solutions internal clustering validation measure of Calinski-Harabasz index is being employed.This paper provides description of proposed algorithm along with its experimental evaluation for a set of benchmark instances.Over a course of our study it was established that clustering based on GOA is characterized by high accuracy -when compared with standard K-means procedure.

Task of clustering, that is data division into homogeneous groups represents one of the elementary problems of contemporary data mining. Cluster analysis can be approached through variety of methods based on statistical inference or heuristic techniques. Recently algorithms employing novel meta-heuristics are of special interest — as they can effectively tackle the problem under consideration which is known to be NP-hard. The paper studies the application of nature-inspired Flower Pollination Algorithm for clustering with internal measure of Calinski-Harabasz index being used as optimization criterion. Along with algorithm's description its performance is being evaluated over a set of benchmark instances and compared with the one of well-known K-means procedure. It is concluded that the application of introduced technique brings very promising outcomes. The discussion of obtained results is followed by areas of possible improvements and plans for further research.

This paper describes a new approach to metaheuristic-based data clustering by means of Krill Herd Algorithm (KHA).In this work, KHA is used to find centres of the cluster groups.Moreover, the number of clusters is set up at the beginning of the procedure, and during the subsequent iterations of the optimization algorithm, particular solutions are evaluated by selected validity criteria.The proposed clustering algorithm has been numerically verified using twelve data sets taken from the UCI Machine Learning Repository.Additionally, all cases of clustering were compared with the most popular method of k-means, through the Rand Index being applied as a validity measure.

In recent times, several new metaheuristic algorithms based on natural phenomena have been made available to researchers. One of these is that of the Krill Herd Algorithm (KHA) procedure. It contains many interesting mechanisms. The purpose of this article is to compare the KHA optimization algorithm used for learning an artificial neural network (ANN), with other heuristic methods and with more conventional procedures. The proposed ANN training method has been verified for the classification task. For that purpose benchmark examples drawn from the UCI Machine Learning Repository were employed with Classification Error and Sum of Square Errors being used as evaluation criteria. It has been concluded that the application of KHA offers promising performance—both in terms of aforementioned metrics, as well as time needed for ANN training.

Particle swarm optimization constitutes currently one of the most important nature-inspired metaheuristics, used successfully for both combinatorial and continuous problems.Its popularity has stimulated the emergence of various variants of swarm-inspired techniques, based in part on the concept of pairwise communication of numerous swarm members solving optimization problem in hand.This paper overviews some examples of such techniques, namely Fully Informed Particle Swarm Optimization (FIPSO), Firefly Algorithm (FA) and Glowworm Swarm Optimization (GSO).It underlines similarities and differences among them and studies their practical features.Performance of those algorithms is also evaluated over a set of benchmark instances.Finally, some concluding remarks regarding the choice of suitable problem-oriented optimization technique along with areas of possible improvements are given as well.