Search

Acquiring knowledge for conversation modeling is an important task in the process of building a Conversational Agent (Chatbot). However, it is a quite difficult process that requires too much time and efforts. To overcome these difficulties, in this paper, we demonstrate a novel methodology for the automatic conversational knowledge extraction from an existing Chatbot. Extracted knowledge will be used as training dataset for building a Neural Network Conversational Agent. The experiments in the paper show that our proposed approach can be used for the automatic knowledge extraction from any type of Chatbot on the Internet. The experiment that is presented in this paper has two phases. In the first phase, we present a methodology for the conversational knowledge extraction. In the second phase of the experiment, we introduce a methodology for building a new Neural Conversational Agent using a deep learning Neural Network framework. The key novelty of our proposed approach is the automated machine-machine conversational knowledge sharing and reuse. This is an important step towards building the new conversational agents skipping the difficult and time-consuming procedure of knowledge acquisition.

Text summarization has become an important research area, especially in the biomedical domain, where information overload is a major problem. In this paper, we propose a novel biomedical text summarization system that combines two popular data mining techniques: clustering and frequent itemset mining. Biomedical paper is expressed as a set of biomedical concepts using the UMLS metathesaurus. The K-means algorithm is used to cluster similar sentences. Then, the Apriori algorithm is applied to discover the frequent itemsets among the clustered sentences. Finally, the salient sentences from each cluster are selected to build the summary using the discovered frequent itemsets. For the evaluation step, we selected randomly 100 biomedical papers from the BioMed Central database full-text, and we evaluated the performances of our system by comparing the resulting summaries with the abstracts of these papers using the ROUGE metrics in term of recall, precision, and F-measure. We also compared the obtained summaries with those achieved by five well-known summarizers: TextRank, TextTeaser, SweSum, ItemSet Based Summarizer, Microsoft AutoSummarize, and two baselines: summarization using only the frequent itemsets mining (FRQ-CL), and summarization using only the clustering (CL-FRQ). The results demonstrate that this combination can successfully enhance the summarization performances, and the proposed system outperforms other tested summarizers.

This paper proposes a method of recognizing the intention of human activity based on percept sequences that represent the activities of daily living (ADL) in a residential space. Based on the activity intention ontology representing actions, poses, and nearby objects related to human activity intentions, the proposed method identifies a human activity intention by using the event calculus when a percept sequence is entered. It is very difficult to recognize ADL occurring in various places in a residence by using regular percept sequences without error. Furthermore, a human activity can have complex intentions. To solve these problems, this paper proposes an activity intention recognition process consisting of three steps. First, the activity intention inference step recognizes the intention of a given percept sequence based on the activity intention ontology. Second, the complex intention-identifying step determines whether to terminate the previously maintained activity intention or continue maintaining it in a complex manner based on the newly inferred activity intention. Third, the uncertainty handling step corrects an inaccurate activity intention caused by an error in the percept sequence. To evaluate the proposed method presented in this paper, activity intention recognition experiments were conducted based on the ADL data that were collected for six households of elderly persons older than 70 years. The results showed that the proposed method has a precision of 99.62% and recall of 92.83%.

Software clone detection is an emerging research area in the field of software engineering. Software systems are subjected to continuous modifications in source code to improve the performance of the software, which may lead to code redundancy. Duplicate code/code clone is a piece of code reworked several times in software programs due to copy paste activity or reusability of existing software. Code clone is a prime subject in software evolution. Detection of software clones at the time of software evolution may improve the performance of software and reduce the maintenance cost and effort. This paper proposes metric based methods to detect code clones, as software clone is a universal problem in large scale programming environment. This paper introduces two metric based approaches to detect code clones by comparing (i) Control Statement Features (ii) Program Features like different types of statements, operators and operands. In order to demonstrate the effectiveness of the proposed approaches, extensive experiments are conducted on two datasets, C projects of Bellon's benchmark dataset and student lab programs (SLP).The methods efficiently identify similar functional clones. Proposed models only find similarity of whole programs but intelligent enough to highlight similar code segments across program files.

Low quality of business process event logs, as determined by anomalous and missing values, is often unavoidable in practical contexts. The output of process analysis that uses event logs with missing and anomalous values is also likely to be of low quality, thus decreasing the quality of any decisions based on it. While previous work has focused on reconstructing missing events in an event log or removing anomalous traces, in this paper we focus on detecting anomalous values and reconstructing missing values at the level of attributes in event logs. We propose methods based on autoencoders, which are a class of neural networks that can reconstruct their own input and are particularly suitable to learn a model of the complex relationships among attribute values in an event log. These methods do not rely on any a-priori knowledge about the business process that generated an event log and are evaluated using real world and artificially-generated event logs. The paper also discusses a qualitative analysis of the impact of event log cleaning and reconstruction on the output of process discovery. The proposed approach shows remarkable performance regarding activity labels and timestamps in artificial event logs. The performance in the case of real world event logs, in particular timestamp anomaly detection, is lower, which may be due to high variability of attribute values in the chosen event logs. Process models discovered from reconstructed event logs are characterised by lower variability of allowed behaviour and, therefore, are more usable in practice.

Classification of text based on its substance is an essential part of analysis to organize enormously large text data and to mine the salient information contained in it. It is gaining greater attention with the surge in the volume of on-line data available. Classical algorithms like k-NN (k-nearest neighbor), SVM (Support Vector Machine) and their variations have been observed to yield only reasonable results in addressing the problem, leaving enough room for further improvement. A class of algorithms commonly referred to as Sparse Methods has been emerged recently from compressive sensing and found numerous effective applications in many areas of data analysis and image processing. Sparse Methods as a tool for text analysis is an alley that is largely unexplored rigorously. This paper presents exploration of sparse representation-based methods for text classification. Based on the success of sparse representation based methods in different areas of data analysis, we intuitively hypothesized that it should work well on text classification problems as well. This paper empirically reinforces the hypothesis by testing the method on Reuters and WebKB data sets. The empirical results on Reuters and WebKB benchmark data show that it can outperform classical classification algorithms like SVM and k-NN. It has been observed that obtaining the basis of representation and sparse codes are computationally costly operations affecting the performance of the system. We also propose a class-wise dictionary refinement algorithm and dynamic dictionary selection algorithm to make sparse coding faster. The addition of dictionary refinement to the classification system not only reduces the time taken for sparse coding but also gives improved classification accuracy. The outcomes of the study are empirical verification of sparse representation classifier as a text classification tool and a computationally efficient solution for the bottleneck operation of sparse coding.

Predictive analytics is an essential capability in business process management to forecast future status and performance of business processes. In this paper, we focus on one particular predictive monitoring task that is solved using classification techniques, i.e. predicting the next event in a case. Several different classifiers have been recently employed in the literature in this task. However, a quantitative benchmark of different classifiers is currently lacking. In this paper, we build such a benchmark by taking into account 20 classifiers from five families, i.e. trees, Bayesian, rule-based, neural and meta classifiers. We employ six real-world process event logs and consider two different sampling approaches, i.e. case and event-based sampling, and three different validation methods in order to acquire a comprehensive evaluation about the classifiers’ performance. According to our benchmark, the classifier most likely to be the overall superior performer is the credal decision tree (C-DT), followed by the other top-4 performers, i.e. random forest, decision tree, dagging ensemble, and nested dichotomies ensemble. We also provide a qualitative discussion of how features of an event log can affect the choice of best classifier.

Machine learning has become a pillar of today’s expert and intelligent systems where a special attention has been drawn to unsupervised methods. Support vector data description is one of the most interesting methods proposed in this context that addresses two main challenges: the absence of ground truth labels and the presence of outliers in data. It is well known that the ultimate problem has a great impact on a system performance since it typically prevents and limits its generalization and hence its failure. Even though the strengths of the Support vector data description method, its application into multi class and large scale context, as computer vision systems, raises bottleneck issues due to its runtime complexity and mono-class context application. We propose, in this paper, an approach inspired from the movement behavior of particle swarm and relies on the use of core set paradigm to overcome the aforementioned drawbacks of Support vector data description method. Indeed, since the proposed method uses some parameters corresponding to kernel function and swarm’s movement formulation, we present and study in this paper some strategies to ensure a dynamic adaptation of these parameters. The main qualities of the proposed method is that it provides a fast scheme for clustering a set of points without neither a prior knowledge on the number of naturally occurring cluster in the data nor an assumption on clusters shapes. A quantitative and comparative performance assessment is carried out over the Berkeley and BuffaloXiph datasets. The results witness the efficiency of the method in terms of both classification accuracy and computational performance. These quantitative assessments reinforce the significance as well as the importance of embedding the proposed method in other intelligent systems application areas.

The present paper aims to show the necessity of information augmentation to cope with the natural decrease in information content in multi-layered neural networks. It is natural to make an effort to collect as much information as possible, because it is impossible to know which information is necessary or useful before learning. Thus, the present paper tries to force neural networks to form new network configurations to have as much information as possible, contrary to the conventional approach of information reduction, such as many types of regularization. For information augmentation, we use the self-organizing map (SOM), which can over-represent inputs and produce as many similar weights as possible. The method was applied to two data sets: the banknote authentication data set and the character recognition data set. In both experimental results, it was confirmed that redundant and excessive information generation in terms of the excessive number of connection weights was connected with improved generalization.

Clustering of multi-view data has received increasing attention since it explores multiple views of data sets aiming at improving clustering accuracy. Particle Swarm Optimization (PSO) is a well-known population-based meta-heuristic successfully used in cluster analysis. This paper introduces two hybrid clustering methods for multi-view relational data. These hybrid methods combine PSO and hard clustering algorithms based on multiple dissimilarity matrices. These methods take advantage of the global convergence ability of PSO and the local exploitation of hard clustering algorithms in the position update step, aiming to improve the balance between exploitation and exploration processes. Moreover, the paper provides adapted versions of 11 fitness functions suitable for vector data aiming at dealing with multi-view relational data. Two performance criteria were used to evaluate the clustering quality using the two proposed methods over eleven real-world data sets including image and document data sets. Among new findings, it was observed that the top three fitness functions are Silhouette index, Xu index and Intra-cluster homogeneity. The performance of the proposed algorithms was compared with previous single and multi-view relational clustering algorithms. The results show that the proposed methods significantly outperformed the other algorithms in the majority of cases. The results reinforce the importance of the application of techniques such as PSO-based clustering algorithms in the field of expert systems and machine learning. Such application enhances classification accuracy and cluster compactness. Besides, the proposed algorithms can be useful tools in content-based image retrieval systems, providing good categorizations and automatically learning relevance weights for each cluster of images and sets of views.

Summarization, is to reduce the size of the document while preserving the meaning, is one of the most researched areas among the Natural Language Processing (NLP) community. Summarization techniques, on the basis of whether the exact sentences are considered as they appear in the original text or new sentences are generated using natural language processing techniques, are categorized into extractive and abstractive techniques. Extractive summarization has been a very extensively researched topic and has reached to its maturity stage. Now the research has shifted towards the abstractive summarization. The complexities underlying with the natural language text makes abstractive summarization a difficult and a challenging task. This paper presents a comprehensive review of the various works performed in abstractive summarization field. For this purpose, we have selected the recent papers on this topic from Elsevier, ACM, IEEE, Springer, ACL Anthology, Cornell University Library and Google Scholar. The papers are categorized according to the type of abstractive technique used. The paper lists down the various challenges and discusses the future direction for research in this field. Along with these, we have identified the advantages and disadvantages of various methods used for abstractive summarization. We have also listed down the various tools which have been used or developed by researchers for abstractive summarization. The paper also discusses the evaluation techniques being used for assessing the abstractive summaries.

People with type 1 diabetes must control their blood glucose level through insulin infusion either with several daily injections or with an insulin pump. However, estimating the required insulin dose is not easy. Recommender systems, mainly based on Case-Based Reasoning (CBR), are being developed to provide recommendations to users. These systems are designed to keep the experiences or cases of the user in a case-base, which requires maintenance to keep system's response accurate and efficient. This paper proposes a case-base maintenance methodology that combines case-base redundancy reduction and attribute weight learning. Contrary to previous approaches designed for classification problems, the maintenance methodology presented in this paper deals with numerical recommendations. It can manage a potentially huge case-base due to the combinatorial derived from the number of attributes used to represent a case. The proposed approach has been tested using the UVA/PADOVA type 1 diabetes simulator and the results demonstrate that it can accomplish better levels of accuracy than other insulin recommender systems mentioned in the literature, when a large number of attributes is considered This project has received funding from the European Union Horizon 2020 research and innovation programme under grant agreement No. 689810, www.pepper.eu.com/, PEPPER, and the grant of the University of Girona 20162018 (MPCUdG2016). The work has been developed with the support of the research group SITES awarded with distinction by the Generalitat de Catalunya (SGR 20142016)

This paper presents an application of a recently well performing evolutionary algorithm for continuous numerical optimization, Success-History Based Adaptive Differential Evolution Algorithm (SHADE) including Linear population size reduction (L-SHADE), to an expert system for underwater glider path planning (UGPP). The proposed algorithm is compared to other similar algorithms and also to results from literature. The motivation of this work is to provide an alternative to the current glider mission control systems, that are based mostly on multidisciplinary human-expert teams from robotic and oceanographic areas. Initially configured as a decision-support expert system, the natural evolution of the tool is targeting higher autonomy levels. To assess the performance of the applied optimizers, the test functions for UGPP are utilized as defined in literature, which simulate real-life oceanic mission scenarios. Based on these test functions, in this paper, the performance of the proposed application of L-SHADE to UGPP is aggregated using statistical analyis. The depicted fitness convergence graphs, final obtained fitness plots, trajectories drawn, and per-scenario analysis show that the new proposed algorithm yields stable and competitive output trajectories. Over the set of benchmark missions, the newly obtained results with a configured L-SHADE outperforms existing literature results in UGPP and ranks best over the compared algorithms. Moreover, some additional previously applied algorithms have been reconfigured to yield improved performance. Thereby, this new application of evolutionary algorithms to UGPP contributes significantly to the capacity of the decision-makers, when they use the improved UGPP expert system yielding better trajectories.

Cross purchase modeling (cross model) is a method to simultaneously analyze multiple product categories for their interrelated influences on consumers’ decisions to purchase each category. Traditional econometric modeling methods are hard to compute and complex in model specifications. This paper examines an artificial neural network (ANN) approach in this context. To understand ANN models’ performance in increasing data scale, this paper uses four datasets of 4, 8, 16 and 32 categories to train ANN models and make predictions. Results show that an ANN model of 12 hidden nodes can be trained in about 3 h and make 88% prediction hit rate for the 16-category dataset. In contrast, a traditional econometric model requires more than 300 h to finish computing for the 16-category dataset. A typical critique of ANN technique is that it hides learned knowledge inside its model structure. This is a major obstacle preventing ANN from being more frequently applied in business research and practices. The current study takes two approaches to address this issue, i.e., (1) customizing ANN components to allow prior knowledge plugging in, and (2) using the generalized weight technique to reveal knowledge that is embedded in ANN training results. Findings of this paper help managers understand applicability of ANN technique in cross purchase model. By demonstrating plugging in prior knowledge to ANN model training and extracting knowledge from ANN training results, this paper shows approaches for understanding ANN embedded knowledge.

In knowledge discovery studies, association rules mining has been extensively studied to discover hidden knowledge and relationships among set of items in a transactional dataset. Most research on association rule mining focuses on discovering frequent patterns based on the most frequent items occurring in the dataset. However, the process of extracting rare rules has received less attention. In medical dataset studies, the discovery of rare association rules (RARs) is more challenging, because it could likely be used to obtain more potentially rare and unusual knowledge for physicians, beside frequent association rules. Hence, the aim of this paper is to discover non-frequent or rare-unusual association rules (RUARs) from a stroke medical dataset to provide potential meaningful knowledge to the user domain. A discretization method needs to be performed as the data preprocessing step before generating rules. To the best of our knowledge, fewer studies have focused on the role of discretization results to support the extraction of a better amount and quality of RUARs, particularly for medical datasets. In addition, the extracted RUARs is expected to provide potential new unusual insights on stroke risk patterns. This paper applies mutual information measure to discretize a stroke examination dataset collected from a medical center in Taiwan. The interval merging method was proposed to simplify the discrete form and enrich the quality of generated rules. Towards the end, rare association rules, with relatively low support, were generated by employing the Apriori-Rare method accordingly. In addition, a filtering process was applied to the content of the rule itemsets to discover the expected set of RUARs for physicians. Furthermore, the extracted RUARs was analyzed based on the relative risk values toward the occurrence of stroke. Results indicated that the mutual information discretization outperformed the traditional discretization methods in terms of how the discretization scheme can support the extraction of RUARs with a better quantity and quality measurements for further analysis purpose in medical point of view. Moreover, the proposed method had a relatively higher number of RUARs. The knowledge of unusual rule patterns from rare association rules might provide potential new and unusual insights for medical pratitioners and increase the awareness of stroke examination results.

The learning algorithm for multivariable data analysis (LAMDA) is a learning method to group or classify quantitative and qualitative historical data. LAMDA can be applied for self-learning and directed learning. Considering self-learning, LAMDA groups historical data through the global membership degrees of a sample with respect to a class, considering the contribution of all features for each sample. Such contribution is called marginal adequacy degree (MAD). To mix all MADs, several aggregation functions such as linear and geometrical connectives are used. The aggregation function calculates the global adequacy degree (GAD) of a sample with respect to a class. This procedure avoids assigning the class number as initial parameter due to that LAMDA applies an automatic threshold called non-informative class (NIC) which finds the best assignment of a portion of data to one class. One improvement for the LAMDA algorithm is the use of a full reinforcement operator called Yager–Rybalov Triple Pi, which can reduce the number of classes. However, this kind of solution ignores the degree to which one sample belongs to one class relative to all samples. Another limitation is that the classes reduction is not controlled by a parameter or an algorithm, which removes some classes required by a human expert. Such removed classes can be useful for the human expert in the data analysis of complex processes or for intelligent systems when the searching of additional information is required in decision-making. In this paper, the typicality degree and the intuitionistic fuzzy sets (IFSs) are used to propose one improvement of the LAMDA method in order to develop a strategy that avoids the loss of significant classes. This paper shows two functions: global typicality degree (GTD) and intuitionistic global adequacy degree (IGAD). By integrating both functions, a new function called typicality and intuitionistic global adequacy degree (TIGAD) is proposed. The proposal is applied in three study cases. The results show an improvement of clustering which were validated by the human expert and the information of data. Furthermore, the impact of the research is focused on data analysis, where the expert should make the best decision.

In this paper we apply a series of Machine Learning models to a recently published unique dataset on the mortality of COVID19 patients. We use a dataset consisting of blood samples of 375 patients admitted to a hospital in the region of Wuhan, China. There are 201 patients who survived hospitalisation and 174 patients who died whilst in hospital. The focus of the paper is not only on seeing which Machine Learning model is able to obtain the absolute highest accuracy but more on the interpretation of what the Machine Learning models provides. We find that age, days in hospital, Lymphocyte and Neutrophils are important and robust predictors when predicting a patients mortality. Furthermore, the algorithms we use allows us to observe the marginal impact of each variable on a case-by-case patient level, which might help practicioneers to easily detect anomalous patterns. This paper analyses the global and local interpretation of the Machine Learning models on patients with COVID19.

This paper introduces a novel proposal to discover the best associative classification rules through studying the influence of the attributes used in robust catalogues. Notice that a catalogue is defined as a dataset free of duplicate records. Moreover, a robust catalogue is obtained when incomplete records and those with uncertainty are eliminated from a catalogue. Therefore, a robust catalogue is a collection of association rules with 100% confidence and unitary support. In this paper we demonstrate that robust catalogues contain the same association rules as the datasets from which they were obtained, but can be managed in memory without eliminating any data from the analysis. In fact, the experiments performed show that all robust catalogues contained in a classification dataset are easily obtained, providing millions of associative classification rules with 100% confidence to the expert researcher in data mining.

This paper presents a complete review of computer vision algorithms and vision-based intelligent applications, that are developed in the field of the Unmanned Aerial Vehicles (UAVs) in the latest decade. During this time, the evolution of relevant technologies for UAVs; such as component miniaturization, the increase of computational capabilities, and the evolution of computer vision techniques have allowed an important advance in the development of UAVs technologies and applications. Particularly, computer vision technologies integrated in UAVs allow to develop cutting-edge technologies to cope with aerial perception difficulties; such as visual navigation algorithms, obstacle detection and avoidance and aerial decision-making. All these expert technologies have developed a wide spectrum of application for UAVs, beyond the classic military and defense purposes. Unmanned Aerial Vehicles and Computer Vision are common topics in expert systems, so thanks to the recent advances in perception technologies, modern intelligent applications are developed to enhance autonomous UAV positioning, or automatic algorithms to avoid aerial collisions, among others. Then, the presented survey is based on artificial perception applications that represent important advances in the latest years in the expert system field related to the Unmanned Aerial Vehicles. In this paper, the most significant advances in this field are presented, able to solve fundamental technical limitations; such as visual odometry, obstacle detection, mapping and localization, et cetera. Besides, they have been analyzed based on their capabilities and potential utility. Moreover, the applications and UAVs are divided and categorized according to different criteria. This research is supported by the Spanish Government through the CICYT projects (TRA2015-63708-R and TRA2013-48314-C3-1-R).

Digital pre-assembly process can evaluate flaws in design prior to manufacturing.As these reports are written in free text form, they are not fully utilized.This paper proposes a method of applying text mining techniques to extract insights.The proposed method successfully extracts useful information from the text database. After digitally designing components of vehicles, a design team creates a virtual manufacturing environment that resembles actual manufacturing facilities. During this digital pre-assembly process, a review team examines each component, and records its problems and requirements in part verification reports. Once these reports are delivered to specific design team responsible for each part, the design team can make appropriate adjustments to their designs. This digital pre-assembly process can evaluate and prevent flaws in design prior to actual manufacturing, improving production quality and reducing manufacturing cost. As these reports are written in free text form, they, however, are not fully utilized for understanding problems arising from the design process. This paper proposes a method of applying text mining techniques on verification reports to extract insights for quality improvement. In this paper, following three text mining approaches are proposed: (1) Extracting n-grams for text preprocessing and constructing domain ontology; (2) Extracting meaningful insights from text preprocessing; (3) Creating intuitive visual tools to understand the extracted insights. The proposed method is applied on approximately 140,000 reports, and is validated through the quality of the answers obtained for the questions posed by the domain experts. The proposed method successfully extracts useful information from the text database, and provides intuitive graphical interface, thereby satisfying the need of the domain experts. This paper proposes a systematic framework of transforming huge amount of raw text data into intuitive visualization. Through this framework, meaningful knowledge can be extracted, analyzed and shared to improve the quality of the products. Main contribution of our paper is that it proposes a framework for knowledge extraction from pre-assembly process. Not only does it systematically arrange the data, but it also combines various data sources and creates a knowledge system to improve efficiency of the design process.

This short communication provides a discourse emerged after reading “Stacking ensemble based deep neural networks modeling for effective epileptic seizure detection, Expert Systems with Applications, 148, 113239, 2020.” The discussed paper proposes a novel application of stacking-based ensemble for seizure detection, where several deep neural networks were used as base classifiers. The ensemble design and experimental results presented by the author show some weaknesses, which is indicated by, one of which, an inability of the proposed model to outperform previous studies. In this note, controversy of the discussed paper is explained and an improved version of stacking-based deep neural network is also further introduced and detailed to prevent it in the future.

Music radio data is currently underutilised in radio program management. Software tools that listen to and analyse music airplay are in many markets nonexistent, limited, or unaffordable. In this paper we present a novel knowledge discovery and visualisation framework for broadcast radio, ZeitMetric. The ZeitMetric framework uses machine learning and music information retrieval techniques to label radio audio automatically for knowledge discovery. The framework incorporates a novel music dataset collection technique (MusiGrab) to leverage online music services for ground-truth data, as well as a novel knowledge visualisation and presentation technique based on self-organizing maps (ZeitViz). The framework is compared to what little literature relating to this topic exists, and a set of requirements for a high-quality broadcast radio knowledge discovery is developed. MusiGrab specifically is compared to an existing static music information retrieval dataset and shown to offer superior results in this context. Future research directions using and extending the framework are also discussed. On acceptance of the paper, code for a use-case of the MusiGrab dataset collection technique will be released on GitHub.

The variety and plethora of research papers available on the Web motivated researchers to propose models that could assist users with personalized citation recommendations. In recent years, citation recommendation models using Network Representation Learning (NRL) methods have shown promising results. Nevertheless, existing NRL-based models are limited in terms of exploiting semantic relations and contextual information between the objects of bibliographic papers’ networks. Additionally, these models cannot adequately explore the structure of heterogeneous information networks, topical relevance, and relevant semantics. Consequently, they suffer from network sparsity and inadequate personalization problems. To overcome these shortcomings, we present a network embedding model termed as Global Citation Recommendation employing Generative Adversarial Network (GCR-GAN). The proposed model exploits the Heterogeneous Bibliographic Network (HBN) to generate personalized citation recommendations. In particular, the proposed model utilizes semantic relations corresponding to the objects of the heterogeneous bibliographic network and captures network structure proximity employing the Scientific Paper Embeddings using Citation-informed Transformers (SPECTER) and Denoising Auto-encoder networks to learn semantic-preserving graph representations. Compared to baseline models, the recommendations generated by our model over the DBLP and ACM datasets prove that it outperforms baseline methods by gaining almost 11% and 12% improvement in terms of Mean Average Precision (MAP) and Normalized Discounted Cumulative Gain (nDCG) metrics, respectively. Furthermore, we analyzed the effectiveness of the proposed model considering network sparsity issue, where our model gains almost 7% better recall@100 score against the second-best counterpart.

In the Authorship Attribution (AA) task, the most likely author of textual documents, such as books, papers, news, and text messages and posts are identified using statistical and computational methods. In this paper, a new computational approach is presented for identifying the most likely author of text documents. The proposed solution emphasizes lazy profile-based classification and, by using the Term Frequency-Inverse Document Frequency (TF_IDF) scheme, introduces a new measure for identifying important terms of documents. The importance of the terms is then used to calculate the similarity between an anonymous document and known documents. The proposed solution works with raw text documents and does not require any NLP tools for preprocessing, which makes it language-independent. The efficiency of the proposed solution has been evaluated by conducting several experiments on two English and Persian datasets, each of which contains six corpora with different number of authors. The obtained results demonstrate that the proposed solution outperforms state-of-the-art stylometric features, employed by seven well-known classifiers, by obtaining 0.902 accuracy for the English dataset and 0.931 accuracy for the Persian dataset. In addition, supplementary experiments have been conducted to evaluate the effects of documents’ length on the accuracy of the proposed solution, to examine the computation time of the proposed solution and competitive classifiers, and to identify the most effective stylometric features and classifiers.

Crowdsourcing as an important computing resource for the Crowd-based Cooperative, crowdsourcing workers, due to the complexity of their personnel composition and the ambiguity of group characteristics, make it difficult to accurately recommend suitable worker groups for the task. Recommending crowdsourcing tasks to suitable workers can greatly improve the efficiency and quality of the execution of crowdsourcing tasks. The current crowdsourcing recommendation algorithm is based on the single characteristics of the workers, ignoring the multi-dimensional characteristics of the workers and the community characteristics of the crowd. This paper proposes a worker group recommendation method based on multi-community collaboration (GRMBC) by utilizing the worker characteristics information extracted from crowdsourcing platform. Based on the characteristics of workers'reputation, preference and activity, the method divides the worker group into several characteristics communities with similar behavior to discover the potential multi-community structure in the crowdsourcing worker group, and then selects Top-N worker group for recommendation through the interaction among multi-communities. At the same time, we also proposed two mitigation strategies for the cold start problem of data in crowdsourcing. This paper uses the public data collected by AMT to do the experiments, and compares the aggregation results of the recommendations generated by different algorithms. The results show the recommendations generated by the GRBMC algorithm proposed in this paper performs the best comprehensively. When the GRBMC algorithm that considers both worker attributes and communi-ty characteristics has an accuracy improvement of 0.03 ~ 0.04 compared with not using the recommendation algorithm on each index, the accuracy of 0.01 ~ 0.02 is improved com-pared with the method considering single characteristic. Moreover, compared with the individual recommendation method of the workers, the recommendation method of the workers' group is more in line with the platform requirements, and can well reflect the wisdom of crowd wisdom.

In this paper, we propose a Deep Belief Network (DBN) based approach for the classification of audio signals to improve work activity identification and remote surveillance of construction projects. The aim of the work is to obtain an accurate and flexible tool for consistently executing and managing the unmanned monitoring of construction sites by using distributed acoustic sensors. In this paper, ten classes of multiple construction equipment and tools, frequently and broadly used in construction sites, have been collected and examined to conduct and validate the proposed approach. The input provided to the DBN consists in the concatenation of several statistics evaluated by a set of spectral features, like MFCCs and mel-scaled spectrogram. The proposed architecture, along with the preprocessing and the feature extraction steps, has been described in details while the effectiveness of the proposed idea has been demonstrated by some numerical results, evaluated by using real-world recordings. The final overall accuracy on the test set is up to 98% and is a significantly improved performance compared to other state-of-the-are approaches. A practical and real-time application of the presented method has been also proposed in order to apply the classification scheme to sound data recorded in different environmental scenarios.

This paper investigates the link prediction in multiplex networks. Multiplex networks that represent multiple types of interaction between the same group of individuals are a special case of complex networks. Each type of interaction is modeled as a layer in a multiplex network. Usually, the topological structures between different layers of a multiplex network have a certain extent of correlation. As a result, the accuracy of link prediction in multiplex networks can be enhanced by combining the information of different layers. In this paper, link prediction in multiplex networks is regarded as a multiple-attribute decision-making problem, in which the potential links in the target layer are considered as alternatives, layers are viewed as attributes, and the similarity score of a potential link in each layer is an attribute value. In implementation, the TOPSIS method is employed to rank alternatives, and interlayer relevance is used to weight the attributes. The experimental results show that the proposed method is not sensitive to the parameter and the interlayer relevance measure, and achieves superior prediction accuracy.

Semantic web technology seems to be in the infant stage as only little efforts have been taken on ontology construction with cross-domain application. This paper intends to take an effort on a new workspace, in which the ontology construction model under cross-domain application is performed. The core concern of this work is on two decision-making process namely data filtering and data annotation. Certain process is followed in this work: (i) Preprocessing (ii) Proposed Jaccard Similarity Evaluation (iii) Data filtering and Outlier Detection (iv) Semantic annotation and clustering. More particularly, data filtering is performed based on the evaluated similarity function. The outliers are identified and grouped separately. The data annotation is performed based on the semantics and thereby the clustering process takes place to form the ontology precisely. This clustering process obviously relies to the optimization crisis as the optimal centroid selection becomes the greatest issue. In order to solve this, this paper extends with the introduction of a hybrid algorithm named Circling Insisted-Rider Optimization Algorithm (CI-ROA), which hybrids the concept of Whale Optimization Algorithm (WOA) and Rider Optimization Algorithm (ROA), respectively. Finally, the performance of proposed work is compared and proved over other state-of-the-art models.

In recent years, the advancement in technology and the availability of a large number of sensors has led to a huge amount of real-time streaming data. The real-time data measured by Phasor Measurement Units (PMU), are of streaming data type and their analysis can identify the status of the power system failure which leads to system instability. Hence online monitoring of PMU data is extremely important which includes preprocessing, filtering, and analysis before taking any corrective actions in the power system. One of the crucial analyses that can be performed over PMU data is the identification of anomalies that arise due to several factors. Early detection of anomalies in the PMU streaming data is highly useful though in real-time it is difficult to detect anomalies consistently. This paper proposes a novel framework for detecting anomalies in multivariate streaming PMU data based on the Density Estimation technique. The main novelty of this paper is that the proposed model is suitable for any kind of anomaly detection in multivariate streaming PMU data which is not addressed in literatures. The proposed system implements Principal Component Analysis in streaming real-time data to select the relevant features. The K-Means algorithm is performed over the selected feature vectors to find the number of clusters in the streaming PMU data. The Gaussian Mixture Model (GMM) is developed to represent the normal PMU data of the system. The minimum and maximum value of the probability of the training samples being generated by the model is selected as a minimum and maximum threshold. The samples with probability less than the minimum threshold or greater than the maximum threshold are considered as anomalies. The main advantage of this proposed model is that it can perform anomaly detection in multivariate streaming PMU data even online within the specified window. The proposed system can identify real time anomalies of multivariate dataset even by offline training model of GMM. The proposed model is developed in MATLAB 2018b version and is tested in streaming PMU data both for offline and online datasets. The performance of the proposed model is validated using performance metrics and the results show that F1 score is 0.95 for 2 features, 0.91 for 3 features, and 0.99 for 8, 12 and 16 features and 1 for 10 features in offline mode. The performance evaluation of testing data in online mode for 2, 3, 8, 10, 12 and 16 features are obtained which proves that the system is adapting to the streaming PMU data with reduced false positive rate and F1 score of 1 for 8, 10, 12 and 16 features. The experimental results prove that the proposed framework for anomaly detection in streaming data is more effective and accurate in identifying the anomalies in real time PMU streamed dataset.

Customer purchase prediction aims to predict customers' future purchases, and the prediction results are of great importance for conducting future commercial activities. To obtain accurate predictions of customer purchases, this paper develops a machine learning framework based on historical behavioural data. First, considering the sparsity of behavioural data, this paper proposes a feature combination method based on the improved factorization machine algorithm. Second, due to the imbalance of customer purchase data, this paper proposes an imbalanced prediction method based on the maximized marginal category and cost-sensitive ensemble learning. Finally, a real-word travel service purchase dataset is adopted to test the feasibility of the proposed prediction framework. The experimental results and comparative analysis verify the validity of the proposed model.

Current software systems are inherently complex and this fact strongly complicates, and makes more expensive, to validate them. Therefore, it is a must to provide methodologies, supported by tools, that can direct validation activities so that they focus on specific aspects of the system (e.g. its critical parts, common errors produced by developers, components that are expensive to fix after deployment, etc). Among the different validation techniques, testing is the most widely used. In this paper we focus on one of the main problems when testing systems with many components: the likelihood of Failed Error Propagation (FEP). FEP appears when we have faulty components such that their wrong behaviour is not revealed when isolatedly testing them but that might produce an error when they are combined with other components. Given a component, it is not possible to automatically assess the likelihood of FEP. However, previous work has shown that there is a strong correlation between the likelihood of FEP and an Information Theory notion called Squeeziness . Recent work has shown that it is possible to compute different values of Squeeziness (essentially, Squeeziness depends on a positive real value) and some of them are more suitable to estimate FEP. In this paper we present our tool SqSelect . Our tool receives either a specific system or its more important characteristics (number of states, maximum and minimum number of outgoing transitions from a state, size of the input and output alphabets) and returns interesting data that can help the tester to estimate the presence of FEP. In particular, our tool provides the most promising value(s) of the parameter associated with Squeeziness so that the likelihood of FEP can be more accurately estimated. In order to compute these values, our tool relies on an artificial neural network that has been extensively trained (compressing the information from around 250 , 000 systems and around 1 , 500 , 000 executions).

Traditional social community discovery methods concentrate mainly on static social networks, but the analysis of dynamic networks is a prerequisite for real-time and personalized social services. Through the study of community changes, the community structure in a dynamic network can be tracked over time, which helps in the mining of dynamic network information. In this paper, we propose a method of tracking dynamic community evolution that is based on resistance distance. Specifically, we model the time-varying features of dynamic networks using the convergence of a resistance-based distance. In our model, the heterogeneity of neighboring nodes can be obtained in the local topology of nodes by analyzing the resistance distance between nodes. We design a community discovery algorithm that essentially discovers community structures on dynamic networks by identifying the so-called core node. During the process of community evolution analysis, both the dynamic contribution of ordinary nodes and core nodes in each community are considered. In addition, to avoid the inclusion of spurious communities in the community structure, we define the notion of noise community and account for it in our algorithm. Experimental results show that the method proposed in this paper can yield better accuracy than other existing methods.

The paper considers a problem of forecasting of news feeds content. Analysis of existing approaches to this problem solution reveals the need for development of methods with enhanced forecasting capabilities. A method is proposed with expanded accounting for space and time relationships of the processed data. The method is revealed through an example of a neural network forecasting system that implements it. Implementation includes data retrieval from news feeds, their special preprocessing, coding and forecasting of words sets and their interconnections, followed by highlighting news topics and describing the of news feeds content. Some variants of stream recurrent neural networks with spiral layer structures were investigated with due regard to their forecasting capabilities under direction and strength control of the associative call of signals from the network memory. The paper also presents and discusses experimental results, a description of the methodological contribution and recommendations on the method practical application.

This paper provides a multi-stage medical scheme selection process to obtain the suitable scheme in the referral system. Firstly, considering the uncertainty of the experts and physicians, the preference relations (PRs) under the probabilistic linguistic term environment are introduced to the system. Then, this paper checks the consistency of the PRs and selects out the inconsistent PRs. For the inconsistent one, this paper provides alternative evaluations by repairing the most inconsistent element. Then, let experts choose from the alternative evaluation set. Next, by introducing the Bonferroni mean (BM) operator and Choquet integral, this paper integrates the historical data and the evaluations with stage weights to obtain the comprehensive assessment. The weights are calculated by the provided discrete and continuous stage weight functions or the programming model. Finally, the medical scheme selection process for the lung cancer and its simulation experiment are investigated to demonstrate the effectiveness and application of provided methods. The simulation experiments on sensitivity analysis and comparative analysis with existing methods are also conducted to validate the provided method.

This paper presents a small-sample faulty line detection method for a small current grounded system (SCGS) using improved generative adversarial networks (GANs) to overcome the problems of low accuracy and poor economy associated with traditional methods. First, we collected real-time fault data of the dispatching system as the original sample and used the linear interpolation and normalization methods to preprocess the data. We extracted and labeled the features of the original sample to construct a feature vector that represents the single-phase grounded fault. Then, to solve the problem of insufficient faulty samples, this paper used the GAN to generate synthetic samples to expand the training sample size. To stabilize the GAN training, the Wasserstein distance was introduced to improve the GAN loss function (WGAN). Finally, for the synthetic samples, we used Batch Normalization (BN) and an Early Stopping algorithm to train the lightweight depthwise separable convolutional network (DSCN) classifier and obtain a faulty line detection model with fast convergence and high accuracy. With the data of an actual dispatching system in China as an example, the proposed method determined the faulty line within 100–200 ms with an accuracy of more than 96%, which is approximately 26% higher than the accuracy of traditional methods. Compared with mainstream machine learning algorithms, the proposed method realizes the self-extraction of faulty features and simplifies the algorithm. Moreover, the accuracy of the proposed method is much higher than that of other machine learning algorithms on small samples. It is worth noting that this method needs only the existing real-time dispatching data and does not need to install additional data acquisition and fault detection equipment, eliminating the complicated equipment installation and debugging process and greatly reducing equipment costs human operations and maintenance costs. The proposed method has been implemented in an actual power system in China for two years. The results show that it is highly accurate and cost-effective, performs well in real time, and has bright prospects for future applications.

The current paper investigates the problem of multimodal named entity recognition from Twitter data. Named entity recognition (NER) is an important task in natural language processing and has been carefully studied in recent decades. NER from tweets is particularly challenging because of 1) tweets are limited in length, 2) contains noisy text, and 3) contains hashtags. Moreover often tweets are associated with images and hyperlinks. Existing works on tweet-NER mostly concentrate on multimodal deep learning based models neglecting the use of hand-crafted features and usage of hyperlinks. The current paper investigates the incorporation of hand-crafted features extracted from different modalities like images, hyperlinks while extracting named entities from tweet-text. A large set of hand-crafted features are extracted from different modalities (images, hyperlinks) and those are added with the features extracted by a hybrid deep-neural model, bi-directional LSTM and CNN, followed by a conditional random field to perform this task. Several variants of these models in association with different hand-crafted feature sets are designed. Extensive experimentations on a multimodal Twitter data (containing text, images and urls) illustrate that character level hand-crafted features significantly improve the performance of the systems. In a part of the paper, results of the proposed models are also shown on a standard NER dataset, CoNLL 2003 dataset.

In some applications, a relation between a response variable and one or more explanatory variables (referred as a “profile”) characterizes the quality of a process. Profile monitoring is commonly performed through statistical methods, while machine learning schemes have not received much attention in this regard. In this paper, a control chart based on Artificial Neural Networks (ANN) is proposed to monitor linear profiles in phase II. In the proposed control chart, some novel run rules as the major contribution of this paper are also used to enhance the efficiency of the control chart and for faster detection of shifts. Simulation results revealed a good performance of the proposed control chart based on average run length (ARL) criterion. Further, a systematic ANN-based diagnostic procedure was proposed to identify which parameter has changed in the process. Finally, the implementation of the proposed scheme was illustrated through a real calibration example from the field of chemical engineering.

Priority rules are applied in many commercial software tools for scheduling projects under limited resources because of their known advantages such as the ease of implementation, their intuitive working, and their fast speed. Moreover, while numerous research papers present comparison studies between different priority rules, managers often do not know which rules should be used for their specific project, and therefore have no other choice than selecting a priority rule at random and hope for the best. This paper introduces a decision tree approach to classify and detect the best performing priority rule for the resource-constrained project scheduling problem (RCPSP). The research relies on two classification models to map project indicators onto the performance of the priority rule. Using such models, the performance of each priority rule can be predicted, and these predictions are then used to automatically select the best performing priority rule for a specific project with known network and resource indicator values. A set of computational experiment is set up to evaluate the performance of the newly proposed classification models using the most well-known priority rules from the literature. The experiments compare the performance of multi-label classification models with multi-class classification models, and show that these models can outperform the average performance of using any single priority rule. It will be argued that this approach can be easily extended to any extension of the RCPSP without changing the methodology used in this study. info:eu-repo/semantics/publishedVersion

This paper presents a systematic procedure to fairly compare experimental performance scores for machine learning methods for fault diagnosis based on vibration signals. In the vast majority of related scientific publications, the estimated accuracy and similar performance criteria are the sole quality parameter presented. However, the experimental design giving rise to these results is mostly biased, based on unacceptably simple validation methods and on recycling identical patterns in test data sets, previously used for training. Moreover, the methods in general overfit their hyperparameters, introducing additional overoptimistic results. In order to remedy this defect, we critically analyse the usual training-validation-test division and propose an algorithmic guideline in the form of a validation framework. This allows a well defined comparison of experimental results. In order to illustrate the ideas of the paper, the Case Western Reserve University Bearing Data benchmark is used as a case study. Four distinct classifiers are experimentally compared, under gradually more difficult generalization tasks using the proposed evaluation framework: K-Nearest-Neighbor, Support Vector Machine, Random Forest and One-Dimensional Convolutional Neural Network. An extensive literature review suggests that most vibration based research papers, particularly for the Case Western Reserve University Bearing Data, use similar patterns for training and testing, making their classification an easy task.

From unsupervised to supervised learning a fault detection model (for robots).Insights to why and when it becomes more accurate.Theoretical analysis and a prediction tool.Empirical results on 3 real-world domains that back these insights. The use of robots in our daily lives is increasing. As we rely more on robots, thus it becomes more important for us that the robots will continue on with their mission successfully. Unfortunately, these sophisticated, and sometimes very expensive, machines are susceptible to different kinds of faults. It becomes important to apply a Fault Detection (FD) mechanism which is suitable for the domain of robots. Two important requirements of such a mechanism are: high accuracy and low computational-load during operation (online). Supervised learning can potentially produce very accurate FD models, and if the learning takes place offline then the online computational-load can be reduced. Yet, the domain of robots is characterized with the absence of labeled data (e.g., faulty, normal) required by supervised approaches, and consequently, unsupervised approaches are being used. In this paper we propose a hybrid approach - an unsupervised approach can label a data set, with a low degree of inaccuracy, and then the labeled data set is used offline by a supervised approach to produce an online FD model. Now, we are faced with a choice should we use the unsupervised or the hybrid fault detector? Seemingly, there is no way to validate the choice due to the absence of (a priori) labeled data. In this paper we give an insight to why, and a tool to predict when, the hybrid approach is more accurate. In particular, the main impacts of our work are (1) we theoretically analyze the conditions under which the hybrid approach is expected to be more accurate. (2) Our theoretical findings are backed with empirical analysis. We use data sets of three different robotic domains: a high fidelity flight simulator, a laboratory robot, and a commercial Unmanned Arial Vehicle (UAV). (3) We analyze how different unsupervised FD approaches are improved by the hybrid technique and (4) how well this improvement fits our prediction tool. The significance of the hybrid approach and the prediction tool is the potential benefit to expert and intelligent systems in which labeled data is absent or expensive to create.

Novel fuzzy rule-based system (FRBS) for medical data classification tasks is proposed.FRBSs are designed using multi-objective evolutionary optimization algorithms.Set of FRBSs with various levels of accuracy-interpretability trade-off is generated.Benchmark medical data sets are used to evaluate the effectiveness of the system.Highly interpretable and accurate medical decision support is provided by our approach. In medical decision support systems, both the accuracy (i.e., the ability to adequately represent the decision making processes) as well as the transparency and interpretability (i.e., the ability to provide a domain user with compact and understandable explanation and justification of the proposed decisions) play essential roles. This paper presents an approach for automatic design of fuzzy rule-based classification systems (FRBCSs) from medical data using multi-objective evolutionary optimization algorithms (MOEOAs). Our approach generates, in a single run, a collection of solutions (medical FRBCSs) characterized by various levels of accuracy-interpretability trade-off. We propose a new complexity-related interpretability measure and we address the semantics-related interpretability issue by means of efficient implementation of the so-called strong fuzzy partitions of attribute domains. We also introduce a special-coding-free representation of the rule base and original genetic operators for its processing as well as we implement our ideas in the context of well-known and one of the presently most advanced MOEOAs, i.e., Non-dominated Sorting Genetic Algorithm II (NSGA-II). An important part of the paper is devoted to a broad comparative analysis of our approach and as many as 26 alternative techniques arranged in 32 experimental set-ups and applied to three well-known benchmark medical data sets (Breast Cancer Wisconsin (Original), Pima Indians Diabetes, and Heart Disease (Cleveland)) available from the UCI repository of machine learning databases (http://archive.ics.uci.edu/ml). A number of useful in medical applications performance measures including accuracy, sensitivity, specificity, and several interpretability measures are employed. The results of such a broad comparative analysis demonstrate that our approach significantly outperforms the alternative methods in terms of the interpretability of the obtained FRBCSs while remaining either competitive or superior in terms of their accuracy. It is worth stressing that the overwhelming majority of the existing medical classification methods concentrate almost exclusively on the accuracy issues.

We propose an approach for selecting appropriate defuzzification methods that fulfills the requirements of a group of experts.We propose a sensitivity analysis of defuzzification methods.We apply this approach to the assessment of dam performance.Methods were chosen in relation to the objectives (decision-support and communication) defined by a group of dam experts.The approach is applied to a real case. The performance assessment of dams is of major importance for optimizing maintenance management. A methodology is provided here to assess the risk of dam failure using data collected during in-situ inspections and from the design and follow-up files. This assessment takes into account uncertainty associated with the data and the assessment process. The result of this assessment is introduced in a fuzzy frame. This paper presents the approach taken to choose defuzzification methods that allow extracting the information necessary from this possibility distribution so that dam experts can both rank dam maintenance actions and communicate the results of this assessment.This paper first presents the process used to assess dam performance after which it presents the defuzzification methods available. A sensitivity analysis is then performed to select the methods most relevant for our case study. The last part presents the development of a tool useful for the application of defuzzification methods to our problematic and describes a real case study in which the selected defuzzification methods are used.

A novel method for the assessment of trademark similarity is proposed.The method blends together visual, semantic and phonetic similarity.It produces an aggregated score based on the individual assessments.Evaluation using information retrieval measures and human judgment. Trademarks are intellectual property assets with potentially high reputational value. Their infringement may lead to lost revenue, lower profits and damages to brand reputation. A test normally conducted to check whether a trademark is highly likely to infringe other existing, already registered, trademarks is called a likelihood of confusion test. One of the most influential factors in this test is establishing similarity in appearance, meaning or sound. However, even though the trademark registration process suggests a multi-faceted similarity assessment, relevant research in expert systems mainly focuses on computing individual aspects of similarity between trademarks. Therefore, this paper contributes to the knowledge in this field by proposing a method, which, similar to the way people perceive trademarks, blends together the three fundamental aspects of trademark similarity and produces an aggregated score based on the individual visual, semantic and phonetic assessments. In particular, semantic similarity is a new aspect, which has not been considered by other researchers in approaches aimed at providing decision support in trademark similarity assessment. Another specific scientific contribution of this paper is the innovative integration, using a fuzzy engine, of three independent assessments, which collectively provide a more balanced and human-centered view on potential infringement problems. In addition, the paper introduces the concept of degree of similarity since the line between similar and dissimilar trademarks is not always easy to define especially when dealing with blending three very different assessments. The work described in the paper is evaluated using a database comprising 1400 trademarks compiled from a collection of real legal cases of trademark disputes. The evaluation involved two experiments. The first experiment employed information retrieval measures to test the classification accuracy of the proposed method while the second used human collective opinion to examine correlations between the trademark scoring/rating and the ranking of the proposed method, and human judgment. In the first experiment, the proposed method improved the F-score, precision and accuracy of classification by 12.5%, 35% and 8.3%, respectively, against the best score computed using individual similarity. In the second experiment, the proposed method produced a perfect positive Spearman rank correlation score of 1.00 in the ranking task and a pairwise Pearson correlation score of 0.92 in the rating task. The test of significance conducted on both scores rejected the null hypotheses of the experiment and showed that both scores correlated well with collective human judgment. The combined overall assessment could add value to existing support systems and be beneficial for both trademark examiners and trademark applicants. The method could be further used in addressing recent cyberspace phenomena related to trademark infringement such as customer hijacking and cybersquatting.

Adaptive learning rate has been proposed for Deep Learning in MLP.Technique for updating learning rate is free of hyper-parameters.Learning rate is a function of a parameter called learning parameter.Learning parameter is updated based on error gradient.Learning rate is further updated based on the Laplacian score of activation values. An attempt has been made to improve the performance of Deep Learning with Multilayer Perceptron (MLP). Tuning the learning rate or finding an optimum learning rate in MLP is a major challenge. Depending on the value of the learning rate, classification accuracy can vary drastically. This issue has been taken as a challenge in this paper. In this paper, a new approach has been proposed to combine adaptive learning rate in conjunction with the concept of Laplacian score for varying the weights. Learning rate is taken as a function of parameter which itself is updated on the basis of error gradient by forming mini-batches. Laplacian score of the neuron isfurther used for updating the incoming weights. This removes the bottleneck involved in finding the optimum value for the learning rate in Deep Learning by using MLP. It is observed on benchmark datasets that this approach leads to increase in classification accuracy as compared to the existing benchmark levels achieved by the well known methods of deep learning.

Smartphone sensing capabilities have enabled human activity recognition (HAR) solutions to better understand human behavior through computational techniques. However, such solutions have suffered from scalability problems due to the high consumption of computational resources (e.g. memory and processing) and the difficulty of acting in real time due to not observing data evolution over time. These problems occur because the HAR solutions for smartphones have been solved through offline learning with models limited by a data history. The disadvantage of this approach is that human activities constantly change over time and are strongly influenced by the physical environment and user profile. To overcome such problem, this paper proposes a novel low-cost learning algorithm called NOHAR (NOvelty discrete data stream for Human Activity Recognition), focused on continuous flow of data analysis. NOHAR is an online classification algorithm based on symbolic data generated by a discretization process using algorithms as SAX and SFA. The advantages of these algorithms are their abilities to compress and reduce the dimensionality of data. In addition, this paper proposes a new framework called DISTAR (DIscrete STream learning for Activity Recognition) focused on the data streaming analysis. Its goals include standardizing the development of online algorithms for symbolic data. Experimental results using three databases show that NOHAR is on average 33 times faster than the state of the art and can reduce memory consumption by an average of 99.97%.

Several algorithms have been proposed towards discovering the graphical structure of Bayesian networks. Most of these algorithms are restricted to observational data and some enable us to incorporate knowledge as constraints in terms of what can and cannot be discovered by an algorithm. A common type of such knowledge involves the temporal order of the variables in the data. For example, knowledge that event B occurs after observing A and hence, the constraint that B cannot cause A . This paper investigates real-world case studies that incorporate interesting properties of objective temporal variable order, and the impact these temporal constraints have on the learnt graph. The results show that most of the learnt graphs are subject to major modifications after incorporating incomplete temporal objective information. Because temporal information is widely viewed as a form of knowledge that is subjective, rather than as a form of data that tends to be objective, it is generally disregarded and reduced to an optional piece of information that only few of the structure learning algorithms may consider. The paper argues that objective temporal information should form part of observational data, to reduce the risk of disregarding such information when available and to encourage its reusability across related studies.

Indoor localization systems have the capability to change the way of providing location-based services in a closed environment. Though there is no agreed-upon technology that works best in indoor, WiFi signal is an important alternative as most of such places are covered by WiFi Access Points (APs). In this paper, the problem of indoor localization is investigated from the perspective of expert systems through applying machine learning techniques. The significant variation of WiFi signal strength with ambient conditions as well as device configuration badly affects the localization accuracy. Thus, the fingerprinting effort required to train a localization system subject to context heterogeneity is huge. The uncertainty in localization performance due to varying contexts is hardly investigated in the literature. Consequently, the main contribution of this paper is to propose a weighted ensemble classifier based on Dempster–Shafer belief theory to efficiently handle context heterogeneity. Here, the context is defined in terms of different smartphone configurations used for training and testing the system as well as temporal variation of signals. The method presented here utilizes the Dempster–Shafer theory of belief functions to calculate the weights of the base learners in the decision of the ensemble. Belief theory is applied here to handle the inherent uncertainty in WiFi signal variations due to heterogeneous context. Real life experiments are conducted for two datasets, JUIndoorLoc and UJIIndoorLoc at different granularity levels. For JUIndoorLoc, with state-of-the-art classifiers, 86–97% accuracy can be achieved for 10-fold cross-validation. However, when the training context differs from the test conditions, accuracy drops to 62–87%. In such a scenario, the proposed weighted ensemble technique is found to achieve almost 98% localization accuracy when RSSIs, mean and variance of RSSIs are considered as features. The technique can lead to an effective expert system for indoor localization at varying granularity levels. Such systems would be beneficial for pervasive indoor positioning applications as no dedicated infrastructure is needed for positioning.

This paper examines the option pricing performance of the most popular Machine Learning algorithms. The classic parametrical models suffer from several limitations in term of computational power required for parametric calibration and unrealistic economical and statistical assumptions. Therefore, a data driven approach based on non-parametric models is are well justified. Most of the previous researchers focus especially on the neural networks method (NN), the other algorithms being unexplored. Beside NN, this paper also analyses the performance of the Support Vector Regressions and Genetic Algorithms and propose three other Decision Tree methods, respectively Random Forest, XGBoost and LightGMB. In order to emphasize the power of this algorithms, a comparison with classical methods like Black-Scholes and Corrado-Su with both historical and implied parameters have been conducted. The analyzes were performed on European call options who have as underlying asset the WTI crude oil future contracts. Machine Learning algorithms outperform by a great margin the classical approaches regardless of the moneyness and the maturity of the contracts.

In this paper, a case study on the role of fuzzy logic (FL) in data mining and machine learning is carried out. It is outlined that, in order to draw more attention of data-mining and machine-learning communities to FL, studies on FL could be more focused not on the activities that fuzzy methods can perform better but rather on the activities that fuzzy methods can perform and the non-fuzzy ones can’t. Such approach takes us away from discussing quantitative differences between fuzzy and non-fuzzy methods to discussing qualitative differences, which are possibly more favorable objects of scientific curiosity. Following the outlined suggestion, a novel speed-up technique is proposed in this paper to support association rule mining (ARM). The proposed technique is a clustering-based one and provides fusion of clustering and ARM. The catchy feature of this technique is that it works well if applied in fuzzy ARM and doesn’t work well if applied in non-fuzzy ARM. The proposed technique is put through experimental verification involving several real-world datasets, and the results substantiate its effectiveness.

This paper proposes a novel approach to the portfolio management using an AutoEncoder. In particular, features learned by an AutoEncoder with ReLU are directly exploited to portfolio constructions. Since the AutoEncoder extracts characteristics of data through a non-linear activation function ReLU, its realization is generally difficult due to the non-linear transformation procedure. In the current paper, we solve this problem by taking full advantage of the similarity of ReLU and an option payoff. Especially, this paper shows that the features are successfully replicated by applying so-called dynamic delta hedging strategy. An out of sample simulation with crypto currency dataset shows the effectiveness of our proposed strategy.