Rule Learning Method Research Articles

Research On Purchasing Behavior and Marketing Strategy Optimization of Commercial Medical Insurance Based on Unsupervised Learning Algorithms

Being a marketer in today's world is not an easy job, especially in the insurance industry, where many factors affect whether customers buy insurance and how much they buy. This field also has many research topics and areas worth exploring. In this article, the data used is real insurance sales data and used unsupervised learning methods in machine learning to clean, mine and analyze the data. First, using clustering methods to divide people who buy insurance into three categories based on weight and insurance expenses. Then using association rule learning methods to analyze the recurring features among people who buy insurance. Finally, based on these results, many policies and suggestions can be given from the perspective of insurance companies and sellers.

Symbolic metaprogram search improves learning efficiency and explains rule learning in humans

Throughout their lives, humans seem to learn a variety of rules for things like applying category labels, following procedures, and explaining causal relationships. These rules are often algorithmically rich but are nonetheless acquired with minimal data and computation. Symbolic models based on program learning successfully explain rule-learning in many domains, but performance degrades quickly as program complexity increases. It remains unclear how to scale symbolic rule-learning methods to model human performance in challenging domains. Here we show that symbolic search over the space of metaprograms—programs that revise programs—dramatically improves learning efficiency. On a behavioral benchmark of 100 algorithmically rich rules, this approach fits human learning more accurately than alternative models while also using orders of magnitude less search. The computation required to match median human performance is consistent with conservative estimates of human thinking time. Our results suggest that metaprogram-like representations may help human learners to efficiently acquire rules.

The Computational Method for Supporting Thai VerbNet Construction

VerbNet is a lexical resource for verbs that has many applications in natural language processing tasks, especially ones that require information about both the syntactic behavior and the semantics of verbs. This article presents an attempt to construct the first version of a Thai VerbNet corpus via data enrichment of the existing lexical resource. This corpus contains the annotation at both the syntactic and semantic levels, where verbs are tagged with frames within the verb class hierarchy and their arguments are labeled with the semantic role. We discuss the technical aspect of the construction process of Thai VerbNet and survey different semantic role labeling methods to make this process fully automatic. We also investigate the linguistic aspect of the computed verb classes and the results show the potential in assisting semantic classification and analysis. At the current stage, we have built the verb class hierarchy consisting of 28 verb classes from 112 unique concept frames over 490 unique verbs using our association rule learning method on Thai verbs.

A heuristic method for discovering multi-class classification rules from multi-source data in cloud–edge system

The integration of diverse devices has made the establishment of multi-class classification models for multi-source data a primary concern for data mining in cloud–edge system. Developing rule-based classifiers is essential because they present results that express the reasons for classification. However, existing rule learning methods are not compatible with multi-source tabular data containing mixed features, imbalanced labels and missing values, making it difficult to build cross-data source and cross-device data mining models. We developed a heuristic multi-class rule learning method that can handle complex tabular datasets without relying too much on cumbersome preprocessing techniques. We abstract the training process of the classifier into a multi-objective optimization problem and design a novel hybrid evolutionary algorithm to obtain Pareto-optimal solutions as rule-based classifier. Compared with the existing explainable classification methods, this method has obvious advantages in the classification performance of tabular data.

An Approach to Assessing the State of Organic Waste Generation in Community Households Based on Associative Learning

The purpose of this work is to substantiate the approach to assessing the state of organic waste generation by households of a given community, which is based on passive production observations and intellectual analysis of statistical data, which ensures consideration of the factors and features of organic waste generation, as well as the development of qualitative models for forecasting their receipt. To achieve the goal, the following tasks were solved: the analysis of the state of organic waste generation by households in the EU countries was performed; an approach to assessing the state of organic waste generation by households of a given community is proposed; based on the use of the proposed approach, and models for assessing the state of organic waste generation of households in a given community were substantiated. The hypothesis of the study is to substantiate and use an approach to assessing the generation of organic waste by households in individual communities, based on the method of association learning and search for association rules, which will identify factors that have a significant impact on the volume of organic waste generated by households, the consideration of which will improve the accuracy of forecasting models and improve the quality of management of the processes of collection and processing of this waste in communities. The research methodology used allows for the use of data mining, probability theory, mathematical statistics, machine learning technology, and the Associative Rule Learning (ARL) method. Based on the use of a reasonable algorithm, they identify key trends and relationships between the factors of organic waste generation in communities in different countries, which is the basis for creating accurate models for predicting the volume of collection and processing of this waste in communities. The study found that the largest number of households produced organic waste per capita in the range of 0.14–0.25 kg/person. At the same time, most households have from two to four residents and are located on the adjoining territory from 350 m2 to 680 m2. Based on the method of learning associative rules, it was found that there are no close correlations between individual factors that determine the daily volume of organic waste generation by households per capita. The highest correlation coefficient between the type of housing and the income level of household residents is 0.13. The number of residents and the occupied area of the adjacent territory have the greatest impact on the daily volume of organic waste generated by households per capita. The substantiated associative rules of relationships, as well as the diagrams of relationships between factors, have helped to identify those factors that have the greatest impact on the volume of organic waste generation. They are the basis for creating accurate models for predicting the volume of collection and planning the processing of this waste in communities. Based on the proposed approach, Python 3.9 software was developed. It makes it possible to quickly carry out calculations and perform a quantitative assessment of the state of organic waste generation by households of a given community according to the specified rules of association between the volumes of organic waste generation and their factors. The results of the study are the basis for the further development of models for accurate forecasting of the collection and planning of the processing of organic waste from households in communities.

Incorporating logic rules with textual representations for interpretable knowledge graph reasoning

Reasoning on knowledge graphs (KGs) is significant for downstream applications, such as question answering and information extraction. On the basis of using factual triples in KGs, learning logic rules has been explored to handle KG reasoning in an interpretable pattern. Some rule learning methods simply apply thresholds to reduce the exponential complexity caused by the vast candidate rules derived from KGs. Others are designed to prune candidates on the basis of embedding facts in the KG. However, these methods only consider the factual triples, which mainly contain structural information, and ignore sufficient semantic information in real-world KGs. This situation weakens the methods in searching for high-quality rules. To this end, we propose a joint method by incorporating logic rules with textual representations for interpretable KG reasoning, named LoTus. Firstly, LoTus obtains the structural embeddings with only factual triples, and the textual embeddings with descriptions of entities, which are later integrated into the representations of entities and relations. Secondly, top candidate rules are selected by a proposed joint pruning strategy incorporating structural and textual representations in KGs. Finally, LoTus captures high-quality rules and removes low-quality ones by a filtering process with metrics of rules. It reduces the time complexity on KG reasoning by learning rules and provides interpretability simultaneously. Extensive experiments on different datasets show that LoTus obtains competitive performance comparing to baselines on KG reasoning metrics Hits@1, Hits@10 and mean reciprocal rank (MRR), and achieves significant superiority among the recent rule learning methods.

Robust sample weighting to facilitate individualized treatment rule learning for a target population

Summary Learning individualized treatment rules is an important topic in precision medicine. Current literature mainly focuses on deriving individualized treatment rules from a single source population. We consider the observational data setting when the source population differs from a target population of interest. Compared with causal generalization for the average treatment effect that is a scalar quantity, individualized treatment rule generalization poses new challenges due to the need to model and generalize the rules based on a prespecified class of functions that may not contain the unrestricted true optimal individualized treatment rule. The aim of this paper is to develop a weighting framework to mitigate the impact of such misspecification, and thus facilitate the generalizability of optimal individualized treatment rules from a source population to a target population. Our method seeks covariate balance over a nonparametric function class characterized by a reproducing kernel Hilbert space and can improve many individualized treatment rule learning methods that rely on weights. We show that the proposed method encompasses importance weights and overlap weights as two extreme cases, allowing for a better bias-variance trade-off in between. Numerical examples demonstrate that the use of our weighting method can greatly improve individualized treatment rule estimation for the target population compared with other weighting methods.

Simple, efficient, and generalized ECG signal quality assessment method for telemedicine applications

Electrocardiogram (ECG) signals recorded by paramedics in an unsupervised environment are prone to errors and noise due to factors such as electrode misplacements and daily activities during prolonged monitoring. Signal Quality Assessment (SQA) systems can prevent false diagnoses by automated arrhythmia detection systems and reduce the workload of cardiologists. However, current SQA methods are quite complex and/or exhibit poor performances in presence of certain arrhythmias. This study proposed a novel SQA method consisting of three main steps: (1) feasibility conditions check, (2) average beat correlation algorithm, and (3) beat clustering algorithm. The system uses empirical probability functions to automatically adapt the quality score according to the presence of pathological beats and/or noises. We evaluated the proposed method on three datasets and considered both three-quality classes ('Bad', 'HR' and 'Diagnostic') and two-quality classes ('Acceptable' and 'Unacceptable'). For the three-class classification, we obtained an average accuracy of 81.42 % while for the two-class, our algorithm achieved an average sensitivity of 94.59 %, a specificity of 98.38 %, and an accuracy of 97.10 %. The results show that our method achieves high accuracy, specificity and generalizability, outperforming the simple heuristic rule, machine learning, and deep learning methods. Additionally, the algorithm requires less computational time, making it suitable for telemedicine applications.

Graph-based association rule learning for context-based health monitoring to enable user-centered assistance

In response to the demographic change and the accompanying challenges for effective healthcare, approaches to enable using advancements of digitalization and IoT infrastructures as well as AI methods to deliver results in the field of personalized health assistance are necessary. In our research, we aim at enabling user-centered assistance with the help of networked sensors and Health Assistance Systems as well as learning methods based on connected graph data that model the shared system, user, and environmental context. In particular, this paper demonstrates a graph-based dynamic context model for a medication assistance system and presents an association rule learning method using Apriori algorithm to learn correlations between user vitals, activities as well as medication intake behavior. An application scenario for context-based heart rate monitoring is consequently presented as proof of concept, where associated contextual elements from the modeled context relating surges in monitored heart rate to environmental and user activity are shown.

"Review of Literature for Machine and Deep Learning in Human Posture of Ergonomics"

In the realm of human-computer interaction, one of the hottest topics right now is the utilization of skeletal data for the purpose of human posture detection. Both the local relationship between the joints and the global spatial position of the joint are taken into consideration when determining the angle and distance attributes of a joint. After that, the method of rule learning is combined with the method of bagging and the method of random subspace to create unique samples and features for the purpose of improving the classification performance of sub-classifiers for a variety of samples used in human posture classification. In the final step of this evaluation, the performance of our proposed technique is assessed using four human posture datasets. The results of our experiments demonstrate that our algorithm is able to accurately recognize a diverse range of human postures, and this improvement also results in outputs that are easier to read. Deep learning is a subfield of machine learning and artificial intelligence that attempts to model the way in which humans pick up new information. In its most basic form, it is a neural network with three or more layers. Deep learning helps in the development of a wide variety of artificial intelligence applications, which in turn helps in the improvement of automation, the performance of analytical and physical activities without the need for human participation, and therefore develops applications that are disruptive to existing methodologies. Deep learning has several applications, and one of these is the detection of human poses, which traditionally would have been done using more conventional methods.

Missing-edge aware knowledge graph inductive inference through dual graph learning and traversing

Knowledge graph (KG) is a kind of structured human knowledge of modeling the relations between real-world entities. This paper studies the KG inductive inference problem, i.e., predicting the relations for out-of-KG entities. However, due to the incomplete nature of the KGs, the connections of some relations are missing. This makes existing differentiable rule learning methods unable to represent some possible rule candidates, which will further affect the inductive inference result. To solve this challenge, our research hypothesis is that the semantics of relation’s argument can be well used to reflect the possible connections between relations. We propose a KG inductive inference model, RuleNet, which consists of two parts. Firstly, a query-dependent dual graph construction method is proposed, which is able to learn the relation connections using the information of the relation’s argument. Secondly, a dual graph traversing method is proposed, which is able to traverse all possible rule candidates even if some rules cannot be formed due to the missing edges. Performance of the proposed methods is evaluated using the FB15K237 (10%–20%), WN18RR (10%–20%) and YAGO3-10 (10%–20%) benchmarks. Experimental results show that RuleNet achieves a superior performance compared with many strong baselines. Ablation studies have verified the effectiveness of the proposed network components. Qualitative analysis shows that RuleNet can learn meaningful dual graph and logic rules.

Multi-Target Regression Rules With Random Output Selections

In this article, we address the task of multi-target regression (MTR), where the goal is to predict multiple continuous variables. We approach MTR by learning global models that simultaneously predict all of the target variables, as opposed to learning a separate model for predicting each of the target variables. Specifically, we learn rule ensembles by generating many candidate rules and assigning them weights that are then optimized in order to select the best performing subset of rules. Candidate rules are generated by transforming ensembles of generalized decision trees, called predictive clustering trees (PCTs), into rules. We propose to extend an existing multi-target regression rule learning method named FIRE by learning tree ensembles that use random output selections (ROS). Such ensembles force individual PCTs to focus only on randomly selected subsets of target variables. The rules obtained from the tree ensemble also focus on various subsets of the target variables (FIRE-ROS). We use three different ensemble methods to generate candidate rules: bagging and random forests of PCTs, and ensembles of extremely randomized PCTs. An experimental evaluation on a range of benchmark datasets has been conducted, where FIRE-ROS is compared to three interpretable methods, namely predictive clustering rules, MTR trees and the original FIRE method, as well as state-of-the-art MTR methods, in particular ensembles of extremely randomized PCTs with ROS, random linear combinations and extremely randomized MTR trees with random projections of the target space. The results show that FIRE-ROS can improve the predictive performance of the FIRE method and that it performs on par with state-of-the-art (non-interpretable) MTR methods.

A greedy belief rule base generation and learning method for classification problem

Among many rule-based systems employed in classification problems, the belief rule-based (BRB) system has been significant for its ability to deal with both quantitative and qualitative information under uncertainty. However, it may face excessive information and low accuracy in some applications due to the limitations of the conventional BRB generation method. To this end, a greedy-based BRB learning method is proposed in this paper. Firstly, the BRB is generated by selecting a reduced number of belief rules from a set of candidate rules. Then, the BRB learning process is conducted by exploiting a selection and reduction strategy, which searches and selects the optimal rules from candidate rules as well as removes noise and redundant rules. Moreover, the original procedures of the inference process and class estimation are retained from conventional BRB systems. Thirty standard classification benchmarks are tested to validate the effectiveness and efficiency of the proposed method, and the classification results are compared with existing rule-based systems, novel belief rule-based systems, and conventional machine learning methods. The comparison results show that the proposed method could achieve relatively satisfactory accuracy while having a significantly smaller BRB. Furthermore, the results derived from benchmarks with two or three classes show the superior performance of the proposed method compared with some state-of-the-art classification methods.

Efficient Utilization of Dependency Pattern and Sequential Covering for Aspect Extraction Rule Learning

The use of dependency rules for aspect extraction tasks in aspect-based sentiment analysis is a promising approach. One problem with this approach is incomplete rules. This paper presents an aspect extraction rule learning method that combines dependency rules with the Sequential Covering algorithm. Sequential Covering is known for its characteristics in constructing rules that increase positive examples covered and decrease negative ones. This property is vital to make sure that the rule set used has high performance, but not inevitably high coverage, which is a characteristic of the aspect extraction task. To test the new method, four datasets were used from four product domains and three baselines: Double Propagation, Aspectator, and a previous work by the authors. The results show that the proposed approach performed better than the three baseline methods for the F-measure metric, with the highest F-measure value at 0.633.

Human posture recognition based on multiple features and rule learning

The use of skeleton data for human posture recognition is a key research topic in the human-computer interaction field. To improve the accuracy of human posture recognition, a new algorithm based on multiple features and rule learning is proposed in this paper. Firstly, a 219-dimensional vector that includes angle features and distance features is defined. Specifically, the angle and distance features are defined in terms of the local relationship between joints and the global spatial location of joints. Then, during human posture classification, the rule learning method is used together with the Bagging and random subspace methods to create different samples and features for improved classification performance of sub-classifiers for different samples. Finally, the performance of our proposed algorithm is evaluated on four human posture datasets. The experimental results show that our algorithm can recognize many kinds of human postures effectively, and the results obtained by the rule-based learning method are of higher interpretability than those by traditional machine learning methods and CNNs.

Anomaly detection in business processes using process mining and fuzzy association rule learning

Much corporate organization nowadays implement enterprise resource planning (ERP) to manage their business processes. Because the processes run continuously, ERP produces a massive log of processes. Manual observation will have difficulty monitoring the enormous log, especially detecting anomalies. It needs the method that can detect anomalies in the large log. This paper proposes the integration of process mining, fuzzy multi-attribute decision making and fuzzy association rule learning to detect anomalies. Process mining analyses the conformance between recorded event logs and standard operating procedures. The fuzzy multi-attribute decision making is applied to determine the anomaly rates. Finally, the fuzzy association rule learning develops association rules that will be employed to detect anomalies. The results of our experiment showed that the accuracy of the association rule learning method was 0.975 with a minimum confidence level of 0.9 and that the accuracy of the fuzzy association rule learning method was 0.925 with a minimum confidence level of 0.3. Therefore, the fuzzy association rule learning method can detect fraud at low confidence levels.

Genetic fuzzy rule-based system using MOGUL learning methodology for energy consumption forecasting

This paper presents the application of a Methodology to Obtain Genetic fuzzy rule-based systems Under the iterative rule Learning approach (MOGUL) to forecast energy consumption. Historical data referring to the energy consumption gathered from three groups, namely lights, HVAC and electrical socket, are used to train the proposed approach and achieve forecasting results for the future. The performance of the proposed method is compared to that of previous approaches, namely Hybrid Neural Fuzzy Interface System (HyFIS) and Wang and Mendel’s Fuzzy Rule Learning Method (WM). Results show that the proposed methodology achieved smaller forecasting errors for the following hours, with a smaller standard deviation. Thus, the proposed approach is able to achieve more reliable results than the other state of the art methodologies

On the influence of using fuzzy extensions in linguistic fuzzy rule-based regression systems

Fuzzy Rule-Based Systems, FRBSs, are powerful tools to address regression problems. They can model the relationship between inputs and outputs by linguistic concepts. However, those FRBSs which are based on the conventional Type-1 fuzzy sets may not be able to handle some difficulties of real-world applications. In such situations, using novel representations of fuzzy sets seems like a good idea. Different extensions of fuzzy sets usually help to provide more precise models in the real-world problems. In this study, the influence of using fuzzy extensions in improving the efficiency of linguistic fuzzy rule-based regression models is investigated. For this purpose, a conventional Type-1 Mamdani FRBS is adapted to the three extensions of fuzzy sets, namely Interval Type-2, Intuitionistic, and Interval Type-2 Intuitionistic fuzzy sets. A two-pass method is proposed to define membership (non-membership) functions of these fuzzy sets; this method is based on the 3-tuples representation of the standard Type-1 membership functions. Wang and Mendel’s rule learning method is adapted to extract fuzzy rules from regression data. In order to tune the membership functions up to different extents, three evolutionary extensions are also presented for each type of the proposed FRBSs. Individual, internal, and external comparisons of the proposed FRBSs were done using 22 real-world regression datasets and statistical tests. Experimental results confirm that all the three proposed FRBSs outperform the classical Type-1 framework; furthermore, the Interval Type-2 Intuitionistic FRBS is the superior system so that an appropriate tuning of its parameters makes it the most accurate model.

EEFR-R: extracting effective fuzzy rules for regression problems, through the cooperation of association rule mining concepts and evolutionary algorithms

Fuzzy rule-based systems, due to their simplicity and comprehensibility, are widely used to solve regression problems. Fuzzy rules can be generated by learning from data examples. However, this strategy may result in high numbers of rules that most of them are redundant and/or weak, and they affect the systems’ interpretability. Hence, in this paper, a new rule learning method, EEFR-R, is proposed to extract the effective fuzzy rules from regression data samples. This method is formed through the cooperation of association rule mining concepts and evolutionary algorithms in the three stages. Indeed, the components of a Mamdani fuzzy rule-based system are generated during the first two stages, and then, they will be refined through some modifications in the last stage. In EEFR-R, fuzzy rules are extracted from numerical data using the idea of Wang and Mendel’s method and utilizing the concepts of Support and Confidence; furthermore, a new rule pruning method is presented to refine these rules. By employing this method, non-effective rules can be pruned in three different modes as the preferences of a decision maker. The proposed model and its stages were validated using 19 real-world regression datasets. The experimental results and the conducted statistical tests confirmed the effectiveness of EEFR-R in terms of complexity and accuracy and in comparison with the three state-of-the-art regression solutions.

Seismic vulnerability assessment at urban scale using data mining and GIScience technology: application to Urumqi (China)

Seismic vulnerability assessments play a significant role in comprehensive risk mitigation efforts and seismic emergency planning, especially for urban areas with a high population density and a complex construction environment. Traditional approaches such as in situ fieldwork are accurate for conducting seismic vulnerability assessments of buildings; however, they are too much time and cost-consuming, especially in moderate to low seismic hazard regions. To address this issue, an integrated approach for a macroseismic vulnerability assessment composed of data mining methods and GIScience technology was presented and applied to Urumqi, China. First, vulnerability proxies were established via in situ data of buildings in the Tianshan District with an EMS-98 vulnerability classification scheme and two data mining methods, namely, support vector machine and association rule learning methods. Then, vulnerability proxies were applied to the Urumqi database, and the accuracy was validated. Finally, seismic risk maps were constructed through data consisting of direct damage to buildings and human casualties. The results indicated that the two data mining methods could achieve desirable accuracies and stabilities when estimating the seismic vulnerability. The seismic risk of Urumqi was estimated as Slight with a predicted number of 61,380 homeless people for a seismic intensity scenario of VIII.