Information

Pałczyński, Cezary;Olejnik, Paweł

doi: 10.3390/info16020071pmid: N/A

In this study, anomalies in a fan system were classified using a real measurement setup to simulate mechanical anomalies such as blade detachment or debris accumulation. Data were collected under normal operating conditions and with an added unbalancing mass. Additionally, sensor anomalies were introduced by manipulating accelerometer readings and examining three types: spike, stuck, and dropout. To classify the anomalies, four neural network models—variations in Long Short-Term Memory (LSTM) and Convolutional Neural Network (CNN) were tested. These models incorporated a Continuous Wavelet Transform (CWT) layer. A novel approach for implementing the CWT layer in both LSTM and CNN architectures was proposed, along with a dual-branch input structure featuring two CWT layers using different mother wavelets. The dual-branch configuration with different mother wavelets yielded better accuracy for the simpler LSTM network. Accuracy comparisons were conducted for the 10 best-performing models based on validation set predictions, revealing improved classification performance. The study concluded with a summary of prediction accuracy for both the validation and test sets of data, along with the calculation of average accuracy, demonstrating the effectiveness of the proposed dual-branch neural network structure in classifying anomalies in fan systems.

García, Roberto;Cediel, Ana;Teixidó, Mercè;Gil, Rosa M.

doi: 10.3390/info16020072pmid: N/A

Blockchain technologies provide new opportunities for media copyright management. To provide an overview of the main initiatives in this blockchain application area, we have first reviewed the existing academic literature. The bibliometric analysis of the literature about copyright and blockchain in the Scopus database identifies four main areas of activity, namely “Digital Rights Management”, “Copyright Protection”, “Social Media”, and “Intellectual Property Rights”. However, it also shows that the literature is still scarce and immature in many aspects, which becomes more evident when comparing it to initiatives coming from the industry. Blockchain has been receiving significant inflows of venture capital and crowdfunding, which have boosted its progress in many fields, including its application to media management. Consequently, we have complemented the review with a business perspective. Existing reports about blockchain and media have been studied and consolidated into four prominent business use cases: “Copyright Management”, “Digital Content Scarcity”, “Marketing, Fan Engagement and Fundrising”, and “Disintermediated Distribution”. Moreover, each one has been illustrated through existing businesses already exploring them. Combining the academic and industry perspectives, this review helps researchers identify the current trends in academic research about media copyright management using blockchain technologies, but without losing track of the state of the art in the industry, which in many cases is more advanced, and the business use cases they can connect their research to.

Li, Zhe;Lu, Xinxi

doi: 10.3390/info16020073pmid: N/A

Current applications of large language models (LLMs) in the field of code intelligence face issues related to low tokenization efficiency. This results in longer token sequences for input to source code types, which leads to the waste of contextual resources for large models. Additionally, the existing LLM tokenization technology struggles to ensure the contextual synonymity of variables. To address these problems, we propose a compiler-based compressed input sequence method. We focus on using the compiler’s lexical analyzer for preliminary tokenization of the input statements, followed by tokenization and filtering through the large model’s tokenizer. This approach results in shorter, semantically clearer, and higher-quality embedded token sequences. Then, using a contextual dictionary, the reduced tokens can be restored to their original state in the output statements. The experimental results show that our compressed input sequence method can be run smoothly in code generation scenarios. Compared to the baseline model, the compiler-based tokenization method can reduce the input token count by 33.7%. This study provides new insights for the application of LLMs in the field of code intelligence.

Tormos, Bernardo;Pla, Benjamín;Sánchez-Márquez, Ramón;Carballo, Jose Luis

doi: 10.3390/info16020074pmid: N/A

Industry 4.0, leveraging tools like AI and the massive generation of data, is driving a paradigm shift in maintenance management. Specifically, in the realm of Artificial Intelligence (AI), traditionally “black box” models are now being unveiled through explainable AI techniques, which provide insights into model decision-making processes. This study addresses the underutilization of these techniques alongside On-Board Diagnostics data by maintenance management teams in urban bus fleets for addressing key issues affecting vehicle reliability and maintenance needs. In the context of urban bus fleets, diesel particulate filter regeneration processes frequently operate under suboptimal conditions, accelerating engine oil degradation and increasing maintenance costs. Due to limited documentation on the control system of the filter, the maintenance team faces obstacles in proposing solutions based on a comprehensive understanding of the system’s behavior and control logic. The objective of this study is to analyze and predict the various states during the diesel particulate filter regeneration process using Machine Learning and explainable artificial intelligence techniques. The insights obtained aim to provide the maintenance team with a deeper understanding of the filter’s control logic, enabling them to develop proposals grounded in a comprehensive understanding of the system. This study employs a combination of traditional Machine Learning models, including XGBoost, LightGBM, Random Forest, and Support Vector Machine. The target variable, representing three possible regeneration states, was transformed using a one-vs-rest approach, resulting in three binary classification tasks where each target state was individually classified against all other states. Additionally, explainable AI techniques such as Shapley Additive Explanations, Partial Dependence Plots, and Individual Conditional Expectation were applied to interpret and visualize the conditions influencing each regeneration state. The results successfully associate two states with specific operating conditions and establish operational thresholds for key variables, offering practical guidelines for optimizing the regeneration process.

Deville, Alain;Deville, Yannick

doi: 10.3390/info16020075pmid: N/A

Since a 1932 work from von Neumann, it has been considered that if two statistical mixtures are represented by the same density operator ρ, they should, in fact, be considered as the same mixture. In a 1970 paper, Zeh introduced a thought experiment with neutron spins, and suggested that, in that experiment, the density operator could not tell the whole story. Since then, no consensus has emerged yet, and controversies on the subject still presently develop. In his 1995 book, speaking of the use of the density operator, Peres spoke of a von Neumann postulate. In this paper, keeping the random variable used by von Neumann in his treatment of statistical mixtures, but also considering higher-order moments of this random variable, it is established that the two mixtures imagined by Zeh, with the same ρ, should however be distinguished. We show that the rejection of that postulate, installed on statistical mixtures for historical reasons, does not affect the general use of ρ, e.g., in quantum statistical mechanics, and the von Neumann entropy keeps its own interest and even helps clarifying that confusing consequence of the postulate identified by Peres.

Krütli, David;Hanne, Thomas

doi: 10.3390/info16020076pmid: N/A

In the past few years, generative AI has seen remarkable progress. The emergence of the transformer architecture has facilitated the creation of highly advanced language models that generate text, summarize content, and translate languages with impressive accuracy. Our study introduces a retrieval-augmented generation system tailored to the dynamic needs of facility management. The proposed system aims to provide instant, accurate access to essential information by integrating advanced techniques from natural language processing and information retrieval paradigms. The implementation leverages the Mixtral 8x7B model for multilingual text processing and the Milvus vector database for efficient document storage and retrieval. The dataset used includes documents such as images, operation manuals, inspection results, blueprints, and technical drawings, in various file formats. This diverse dataset reflects the variety of information encountered in construction and facility management. The evaluation involved generating question–answer pairs pertinent to facility management tasks and assessing the system’s performance using metrics such as ROUGE, BLEU, and semantic similarity. The findings suggest that retrieval-augmented generation systems can significantly enhance operational efficiency by reducing the time and effort required to access information while maintaining high security and data privacy standards.

Cadena-Bautista, Ángel;López-Ponce, Francisco F.;Ojeda-Trueba, Sergio Luis;Sierra, Gerardo;Bel-Enguix, Gemma

doi: 10.3390/info16020077pmid: N/A

In this paper various LLMs are tested in a specific domain using a Retrieval-Augmented Generation (RAG) system. The study focuses on the performance and behavior of the models and was conducted in Spanish. A questionnaire based on The Bible, which consists of questions that vary in complexity of reasoning, was created in order to evaluate the reasoning capabilities of each model. The RAG system matches a question with the most similar passage from The Bible and feeds the pair to each LLM. The evaluation aims to determine whether each model can reason solely with the provided information or if it disregards the instructions given and makes use of its pretrained knowledge.

Maulana, Muhammad Mahreza;Suroso, Arif Imam;Nurhadryani, Yani;Seminar, Kudang Boro

doi: 10.3390/info16020078pmid: N/A

Some business license commitments in online single submission (OSS) application currently only consist of an independent statement from the business actor, and there is no time limit for fulfilling the license, especially for low-risk bio-business licenses. Therefore, a system design is needed to monitor the fulfillment of business license commitments and to provide confidence to the public regarding the business provided. The system design started with discussing the importance of smart governance systems, defining the requirements for monitoring the commitment of bio-business licenses, selecting documents, determining important words and document types, and proposing a design. This research produces a prototype of a business license monitoring system. This smart governance system checks the validity of business license documents. This study provides new insights into how smart governance system can be implemented in bio-business licensing. This study also considers the thousands of government applications that currently exist and the desire to provide one solution that contains many functions that can fulfill the many needs of business actors for simple and transparent business licenses.

Blachowicz, Tomasz;Wylezek, Jacek;Sokol, Zbigniew;Bondel, Marcin

doi: 10.3390/info16020079pmid: N/A

The application of modern machine learning methods in industrial settings is a relatively new challenge and remains in the early stages of development. Current computational power enables the processing of vast numbers of production parameters in real time. This article presents a practical analysis of the welding process in a robotic cell using the unsupervised HDBSCAN machine learning algorithm, highlighting its advantages over the classical k-means algorithm. This paper also addresses the problem of predicting and monitoring undesirable situations and proposes the use of the real-time graphical representation of noisy data as a particularly effective solution for managing such issues.

Barbosa-Santillán, Liliana Ibeth;Guzman-Velazquez, Bertha Patricia;Orozco-Aguilera, Ma. Teresa;Flores-Pulido, Leticia

doi: 10.3390/info16020080pmid: N/A

Cyberbullying, which manifests in various forms, is a growing challenge on social media, mainly when it involves threats of violence through images, especially those featuring weapons. This study introduces a computational framework to identify such content using convolutional neural networks of weapon-related images. By integrating artificial intelligence techniques with image analysis, our model detects visual patterns associated with violent threats, creating safer digital environments. The development of this work involved analyzing images depicting scenes with weapons carried by children or adolescents. Images were sourced from social media and spatial repositories. The statistics were processed through a 225-layer convolutional neural network, achieving an 86% accuracy rate in detecting weapons in images featuring children, adolescents, and young adults. The classifier method reached an accuracy of 17.86% with training over only 25 epochs and a recall of 14.2%. Weapon detection is a complex task due to the variability in object exposures and differences in weapon shapes, sizes, orientations, colors, and image capture methods. Segmentation issues and the presence of background objects or people further compound this complexity. Our study demonstrates that convolutional neural networks can effectively detect weapons in images, making them a valuable tool in addressing cyberbullying involving weapon imagery. Detecting such content contributes to creating safer digital environments for young people.