Search Results (181)

Machine learning techniques, particularly decision trees, have been extensively utilized in Network-based Intrusion Detection Systems (NIDSs) due to their transparent, rule-based structures that enable straightforward interpretation. However, this transparency presents privacy risks, as decision trees may inadvertently expose sensitive information such as network configurations or IP addresses. In our previous work, we introduced a sensitive pruning-based decision tree to mitigate these risks within a limited dataset and basic pruning framework. In this extended study, three privacy-preserving pruning strategies are proposed: standard sensitive pruning, which conceals specific sensitive attribute values; optimistic sensitive pruning, which further simplifies the decision tree when the sensitive splits are minimal; and pessimistic sensitive pruning, which aggressively removes entire subtrees to maximize privacy protection. The methods are implemented using the J48 (Weka C4.5 package) decision tree algorithm and are rigorously validated across three full-scale NIDS datasets: GureKDDCup, UNSW-NB15, and CIDDS-001. To ensure a realistic evaluation of time-dependent intrusion patterns, a rolling-origin resampling scheme is employed in place of conventional cross-validation. Additionally, IP address truncation and port bilateral classification are incorporated to further enhance privacy preservation. Experimental results demonstrate that the proposed pruning strategies effectively reduce the exposure of sensitive information, significantly simplify decision tree structures, and incur only minimal reductions in classification accuracy. These findings reaffirm that privacy protection can be successfully integrated into decision tree models without severely compromising detection performance. To further support the proposed pruning strategies, this study also includes a comprehensive review of decision tree post-pruning techniques. Full article

Perennial wall rocket (Diplotaxis tenuifolia L.—DC.) exhibits genotype-dependent responses to biostimulant applications, which have not yet been deeply investigated. A two-year greenhouse factorial experiment was carried out to assess the interactions between five cultivars (Mars, Naples, Tricia, Venice, and Olivetta), three biostimulant formulations (Cystoseira tamariscifolia L. extract; a commercial legume-derived protein hydrolysate, “Dynamic”; and Spirulina platensis extract) plus an untreated control, and three crop cycles (autumn, autumn–winter, and winter) on leaf yield and dry matter, organic acids, colorimetric parameters, hydrophilic and lipophilic antioxidant activities, nitrate concentration, nitrogen use efficiency, and mineral composition, using a split plot design with three replicates. Protein hydrolysate significantly enhanced yield and nitrogen use efficiency in Mars (+26%), Naples (+25.6%), Tricia (+25%), and Olivetta (+26%) compared to the control, while Spirulina platensis increased the mentioned parameters only in Venice (+36.2%). Nitrate accumulation was reduced by biostimulant application just in Venice, indicating genotype-dependent nitrogen metabolism responses. The findings of the present research demonstrate that the biostimulant efficacy in perennial wall rocket is mainly ruled by genotypic factors, and the appropriate combinations between the two mentioned experimental factors allow for optimization of leaf yield and quality while maintaining nitrate concentration under the regulation thresholds. Full article

In cybersecurity, constructing an accurate knowledge graph is vital for discovering key entities and relationships in security incidents buried in vast unstructured threat reports. Traditional knowledge-graph construction pipelines based on handcrafted rules or conventional machine learning models falter when the data scale and linguistic variety grow. GraphRAG, a retrieval-augmented generation (RAG) framework that splits documents into fixed-length chunks and then retrieves the most relevant ones for generation, offers a scalable alternative yet still suffers from fragmentation and semantic gaps that erode graph integrity. To resolve these issues, this paper proposes SC-LKM, a cybersecurity knowledge-graph construction method that couples the GraphRAG backbone with hierarchical semantic chunking. SC-LKM applies semantic chunking to build a cybersecurity knowledge graph that avoids the fragmentation and inconsistency seen in prior work. The semantic chunking method first respects the native document hierarchy and then refines boundaries with topic similarity and named-entity continuity, maintaining logical coherence while limiting information loss during the fine-grained processing of unstructured text. SC-LKM further integrates the semantic comprehension capacity of Qwen2.5-14B-Instruct, markedly boosting extraction accuracy and reasoning quality. Experimental results show that SC-LKM surpasses baseline systems in entity-recognition coverage, topology density, and semantic consistency. Full article

An abstract set of one-dimensional (spinor-type) elements randomly oriented on a plane is introduced as a basic subgeometric object. Endowing the set with the binary operations of multiplication and invertible addition sequentially yields a specific semi-group (for which an original Cayley table is given) and a generic algebraic system which is shown to generate, apart from algebras of real and complex numbers, the associative hypercomplex algebras of dual numbers, split-complex numbers, and quaternions. The units of all these algebras turn out to be composed of basic 1D elements, thus ensuring the automatic fulfillment of multiplication rules (once postulated). From the standpoint of a three-dimensional space defined by a vector quaternion triad, the condition of a standard (unit) length of 1D basis elements is considered; it is shown that fulfillment of this condition provides an equation mathematically equivalent to the main equation of quantum mechanics. The similarities and differences of the proposed logical scheme with other approaches that involve abstract subgeometric objects are discussed. Full article

Although Singing Voice Synthesis (SVS) has revolutionized audio content creation, global linguistic diversity remains challenging. Current SVS research shows scant exploration of cross-lingual generalization, as fragmented, language-specific phoneme encodings (e.g., Pinyin, ARPA) hinder unified phonetic modeling. To address this challenge, we built a four-language dataset based on GTSinger’s speech data, using the International Phonetic Alphabet (IPA) for consistent phonetic representation and applying precise segmentation and calibration for improved quality. In particular, we propose a novel method of decomposing IPA phonemes into letters and diacritics, enabling the model to deeply learn the underlying rules of pronunciation and achieve better generalization. A dynamic IPA adaptation strategy further enables the application of learned phonetic representations to unseen languages. Based on VISinger2, we introduce Transinger, an innovative cross-lingual synthesis framework. Transinger achieves breakthroughs in phoneme representation learning by precisely modeling pronunciation, which effectively enables compositional generalization to unseen languages. It also integrates Conformer and RVQ techniques to optimize information extraction and generation, achieving outstanding cross-lingual synthesis performance. Objective and subjective experiments have confirmed that Transinger significantly outperforms state-of-the-art singing synthesis methods in terms of cross-lingual generalization. These results demonstrate that multilingual aligned representations can markedly enhance model learning efficacy and robustness, even for languages not seen during training. Moreover, the integration of a strategy that splits IPA phonemes into letters and diacritics allows the model to learn pronunciation more effectively, resulting in a qualitative improvement in generalization. Full article

Explainable artificial intelligence (XAI) is essential for fostering trust, transparency, and accountability in machine learning systems, particularly when applied in high-stakes domains. This paper introduces a novel XAI system designed for classification tasks on tabular data, which offers a balance between performance and interpretability. The proposed method, GlassBoost, first trains an XGBoost model on a given dataset and then computes gain scores, quantifying the average improvement in the model’s loss function contributed by each feature during tree splits. Based on these scores, a subset of significant features is selected. A shallow decision tree is then trained using the top d features with the highest gain scores, where d is significantly smaller than the total number of original features. This model compression yields a transparent, IF–THEN rule-based decision process that remains faithful to the original high-performing model. To evaluate the system, we apply it to an anomaly detection task in the context of intrusion detection systems (IDSs), using a dataset containing traffic features from both malicious and normal activities. Results show that our method achieves high accuracy, precision, and recall while providing a clear and interpretable explanation of its decision-making. We further validate its explainability using SHAP, a well-established approach in the field of XAI. Comparative analysis demonstrates that GlassBoost outperforms SHAP in terms of precision, recall, and accuracy, with more balanced performance across the three metrics. Likewise, our review of literature findings indicate that Glassboost outperforms many other XAI models while retaining computational efficiency. In one of our configurations, GlassBoost achieved accuracy of 0.9868, recall of 0.9792, and precision of 0.9843 using only eight features within a tree structure of a maximum depth of four. Full article

Developing high-efficiency photoelectrodes plays an important role in the photoelectrocatalytic generation of hydrogen peroxide (H₂O₂) in the photoelectrochemical (PEC) water splitting field. In this work, an innovative strategy was proposed, the synergistic photocatalytic production of H₂O₂ using a bidirectional photoanode–photocathode coupling system under visible-light irradiation. Fe₂O₃-Ti, as the photoanode, which was built by way of Fe₂O₃ loaded on Ti-mesh using the hydrothermal-calcination method, was investigated in terms of the suitability of its properties for PEC H₂O₂ production after optimization of the bias voltage, the type of electrolyte solution, and the concentration of the electrolyte. Afterwards, a H-type double-electrode coupling system with an Fe₂O₃-Ti photoanode and a WO₃@Co₂SnO₄ photocathode was established for the bidirectional synergistic production of H₂O₂ under visible-light irradiation. The yield of H₂O₂ reached 919.56 μmol·L⁻¹·h⁻¹ in 2 h over −0.7 V with 1 mol·L⁻¹ of KHCO₃ as the anolyte and 0.1 mol·L⁻¹ Na₂SO₄ as the catholyte (pH = 3). It was inferred that H₂O₂ production on the WO₃@Co₂SnO₄ photocathode was in line with the 2e^- oxygen reduction reaction (ORR) principle, and on the Fe₂O₃-Ti photoanode was in line with the 2e^- water oxidation reaction (WOR) rule, or it was indirectly promoted by the electrolyte solution KHCO₃. This work provides an innovative idea and a reference for anode–cathode double coupling systems for the bidirectional production of H₂O₂. Full article

Split air conditioning units are crucial for ensuring the thermal comfort of buildings. Conventional scheduling or pre-timed system activities result in high consumption and wasted energy. This study proposes an intelligent control system for automatic setpoint adjustment in an educational building based on real-time indoor and outdoor environmental and room occupancy data. Principal component analysis was used to identify energy consumption components in ramp-up and steady-state power usage scenarios. K-means clustering was then used to categorize environmental scenarios and occupancy patterns to identify operational states, predict power consumption and environmental variables, and generate fuzzy inference system rules. The application of rough set theory eliminated rule redundancy by at least 99.27% and enhanced computational speed by 96.40%. After testing using real historical data from an uncontrolled environment and occupant thermal comfort satisfaction surveys reflecting a range of ACU setpoints, the enhanced inference system achieved daily average power savings of 25.56% and a steady-state power period at 63.24% of the ACU operating time, as compared to conventional and variable setpoint operations. The proposed technique provides a basis for dynamic and data-driven decision-making, enabling sustainable energy management in smart building applications. Full article

An accurate understanding of the tensile mechanical behavior of shale rock is essential for optimizing shale gas drilling and hydraulic fracturing operations. However, the mechanical behavior of shale is significantly influenced by its anisotropy. Therefore, this study investigated the tensile mechanical behavior of layered shale by combining acoustic emission (AE) monitoring with two testing methods: the Brazilian splitting test (BST) and a novel direct tensile test (DTT). The impact of anisotropy on the tensile mechanical behavior and failure modes of layered shale under different test methods was evaluated. Additionally, seven anisotropic tensile strength criteria were compared and validated using the experimental results. The results show that: (1) As the loading angle (β) increased, the tensile strength measured by both BST and DTT increased. Both methods exhibited maximum tensile strength at β = 90° and minimum tensile strength at β = 0°. The anisotropy ratios for BST and DTT were 1.52 and 2.36, respectively, indicating the significant influence of the loading angle on tensile strength. (2) The AE results indicated that both DTT and BST specimens exhibited brittle failure characteristics. However, the DTT specimens demonstrated more pronounced progressive failure behavior, with failure modes categorized into four types: tensile failure across the bedding plane, shear failure along the bedding plane, and two types of tensile–shear mixed failure. In contrast, the BST specimens primarily exhibited tensile–shear mixed failure, except for tensile failure along the bedding plane at β = 0° and tensile failure across the bedding plane at β = 90°. (3) Neither of the two test methods could fully eliminate the influence of anisotropy, but three anisotropic tensile criteria, the Lee–Pietruszczak criterion, the critical plane approach criterion, and the anisotropic mode I fracture toughness criterion based on the stress–strain transformation rule demonstrated high accuracy in predicting tensile strength. Furthermore, in alignment with previous studies, the indirect tensile strength of various rock types was found to range between one and three times the direct tensile strength, and a linear correlation between the two variables was established, with a coefficient of approximately 1.11. Full article

Computer-assisted grading plays an important role in an educational context, mainly by reducing the workload of teachers in manual scoring. While electronic choice surveys have long been used in many web applications, automatic scoring of open-ended responses remains an interesting research problem in natural language processing. In this article, we propose a new hybrid text-processing method for scoring students’ responses based on word splitting and preprocessing, which will then combine textual algorithms with a set of artificial neural network classifiers and a set of heuristic decision rules. This concept has been implemented in the interactive e-test system operating in the local computer network of the Institute of Applied Computer Science at the Lodz University of Technology. The dataset is acquired as questions, reference answers, and students’ answers generated on the basis of exams conducted at our institute in the years 2015–2022 for more than a thousand students. This article extends our previous research and includes comparative tests. The proposed method achieves excellent results and outperforms the previous approaches. The obtained precision is equal to 1, and the recall measure is 0.97 for the final results. Full article

Aiming at the problem that asphalt pavement materials in plateau areas are vulnerable to freeze-thaw damage, research was carried out on asphalt pavements of representative road sections, and the temperature within the pavement structure was monitored using buried sensors. Based on this, an indoor test method for high-frequency freeze-thaw was established, and UV, thermo-oxygen-aging and high-frequency freeze-thaw tests were combined. The effects of aging and maximum aggregate particle size on the resistance of asphalt mixtures to high-frequency freeze-thaw were investigated using the splitting strength ratio, mass-loss rate and void-ratio changes by employing the newly made RS-type modified asphalt in the laboratory. At the same time, the high-frequency freeze-thaw resistance of the asphalt mixture was compared with that of the SS/SMA-13 asphalt mixture on the top layer of a representative road section. The results show that UV aging at 180 h followed by thermal-oxygen aging at 120 h has the greatest impact on the asphalt mixture; in this condition, the high-frequency freeze-thaw-cycle asphalt mixture with freeze-thaw damage is affected by the rule of change of the third-degree polynomial. In the plateau environment conditions, compared with the original pavement material (SS-type modified asphalt), the RS-type modified asphalt has better anti-aging properties, adhesion properties and elasticity performance. Full article

All characterizations of the Shannon entropy include the so-called chain rule, a formula on a hierarchically structured probability distribution, which is based on at least two elementary distributions. We show that the chain rule can be split into two natural components, the well-known additivity of the entropy in case of cross-products and a variant of the chain rule that involves only a single elementary distribution. The latter is given as a proportionality relation and, hence, allows a vague interpretation as self-similarity, hence intrinsic property of the Shannon entropy. Analogous characterizations are given for the Rényi entropy and its limits, the min-entropy and the Hartley entropy. Full article

The power split plug-in hybrid electric bus (PHEB) boasts the capability for concurrent decoupling of rotation speed and torque, emerging as the key technology for energy conservation. The optimization of energy management strategies (EMSs) and powertrain parameters for PHEB contributes to bolstering vehicle performance and fuel economy. This paper revolves around optimizing fuel economy in PHEBs by proposing an optimization algorithm for the combination of a multi-layer rule-based energy management strategy (MRB-EMS) and powertrain parameters, with the former incorporating intelligent algorithms alongside deterministic rules. It commences by establishing a double-planetary-gear power split model for PHEBs, followed by parameter matching for powertrain components in adherence to relevant standards. Moving on, this paper plunges into the operational modes of the PHEB and assesses the system efficiency under each mode. The MRB-EMS is devised, with the battery’s State of Charge (SOC) serving as the hard constraint in the outer layer and the Charge Depletion and Charge Sustaining (CDCS) strategy forming the inner layer. To address the issue of suboptimal adaptive performance within the inner layer, an enhancement is introduced through the integration of optimization algorithms, culminating in the formulation of the enhanced MRB (MRB-II)-EMS. The fuel consumption of MRB-II-EMS and CDCS, under China City Bus Circle (CCBC) and synthetic driving cycle, decreased by 12.02% and 10.35% respectively, and the battery life loss decreased by 33.33% and 31.64%, with significant effects. Subsequent to this, a combined multi-layer powertrain optimization method based on Genetic Algorithm-Optimal Adaptive Control of Motor Efficiency-Particle Swarm Optimization (GOP) is proposed. In parallel with solving the optimal powertrain parameters, this method allows for the synchronous optimization of the Electric Driving (ED) mode and the Shutdown Charge Hold (SCH) mode within the MRB strategy. As evidenced by the results, the proposed optimization method is tailored for the EMSs and powertrain parameters. After optimization, fuel consumption was reduced by 9.04% and 18.11%, and battery life loss was decreased by 3.19% and 7.42% under the CCBC and synthetic driving cycle, which demonstrates a substantial elevation in the fuel economy and battery protection capabilities of PHEB. Full article

The use of decision trees for obtaining and representing clustering solutions is advantageous, due to their interpretability property. We propose a method called Decision Trees for Axis Unimodal Clustering (DTAUC), which constructs unsupervised binary decision trees for clustering by exploiting the concept of unimodality. Unimodality is a key property indicating the grouping behavior of data around a single density mode. Our approach is based on the notion of an axis unimodal cluster: a cluster where all features are unimodal, i.e., the set of values of each feature is unimodal as decided by a unimodality test. The proposed method follows the typical top-down splitting paradigm for building axis-aligned decision trees and aims to partition the initial dataset into axis unimodal clusters by applying thresholding on multimodal features. To determine the decision rule at each node, we propose a criterion that combines unimodality and separation. The method automatically terminates when all clusters are axis unimodal. Unlike typical decision tree methods, DTAUC does not require user-defined hyperparameters, such as maximum tree depth or the minimum number of points per leaf, except for the significance level of the unimodality test. Comparative experimental results on various synthetic and real datasets indicate the effectiveness of our method. Full article

(1) Background: Urinary tract infection (UTI) is a leading cause of emergency department visits and hospital admissions. Despite many studies identifying UTI-related risk factors for bacteremia or sepsis, a significant gap remains in developing predictive models for in-hospital mortality or the necessity for emergent intensive care unit admission in the emergency department. This study aimed to construct interpretable machine learning models capable of identifying patients at high risk for critical outcomes. (2) Methods: This was a retrospective study of adult patients with urinary tract infection (UTI), extracted from the Medical Information Mart for Intensive Care IV Emergency Department (MIMIC-IV-ED) database. The critical outcome is defined as either in-hospital mortality or transfer to an intensive care unit within 12 h. ED visits were randomly partitioned into a 70%/30% split for training and validation. The extreme gradient boosting (XGBoost), random forest (RF), and support vector machine (SVM) algorithms were constructed using variables selected from the stepwise logistic regression model. The XGBoost model was then compared to the traditional model and clinical decision rules (CDRs) on the validation data using the area under the curve (AUC). (3) Results: There were 3622 visits among 3235 unique patients diagnosed with UTI. Of the 2535 patients in the training group, 836 (33%) experienced critical outcomes, and of the 1087 patients in the validation group, 358 (32.9%) did. The AUCs for different machine learning models were as follows: XGBoost, 0.833; RF, 0.814; and SVM, 0.799. The XGBoost model performed better than others. (4) Conclusions: Machine learning models outperformed existing traditional CDRs for predicting critical outcomes of ED patients with UTI. Future research should prospectively evaluate the effectiveness of this approach and integrate it into clinical practice. Full article