Research activities per year filter by: Publication Year
(Descending) Articles
Publication Date: 2025
World Wide Web (15731413) 28(3)
A session-based recommendation system (SBRS) focuses on the user’s interactions in the current session to provide recommendations. Recent works employ graph neural networks (GNN) to capture complex relationships between clicked items in a session. Some of these studies incorporate collaborative information from similar/neighbor sessions. To identify neighbor sessions, they calculate the similarity of sessions by counting the common items between two sessions. But this is too simplistic and the similarity depends on other features such as the order of common items. Furthermore, meta-data such as items’ categories have not been considered, while items can be categorized into a limited number of categories, which can be informative in finding similar sessions and predicting user intent. In this paper, we propose a novel method, named Collaborative Category and Time-aware Graph Neural Networks (CCT-GNN), which models users’ interactions in two levels: (i) Global-level, which identifies neighbor sessions effectively and explores collaborative information to improve model performance. (ii) Local-level, in which the current session interactions are transformed into an item category graph to model different types of relations between items and categories. Experimental results demonstrate CCT-GNN superiority over state-of-the-art methods. Source code is available at: https://github.com/Moosazadeh/CCT-GNN. © The Author(s), under exclusive licence to Springer Science+Business Media, LLC, part of Springer Nature 2025.
Publication Date: 2025
Molecular Informatics (18681751) 44(2)
The widespread use of polymer solutions in the chemical industry poses a significant challenge in determining optimal dissolution conditions. Traditionally, researchers have relied on experimental methods to estimate the processing parameters needed to dissolve polymers, often requiring numerous iterations of testing different temperatures and pressures. This approach is both costly and time-consuming. In this study, for the first time, we present a machine learning-based approach to predict the minimum temperature and pressure required for polymer dissolution, correlating molecular weight and chemical structure of both the polymer and solvent and its weight percent. Using a dataset compiled from existing literature, which includes key factors influencing polymer dissolution, we also extracted chemical bond information from the molecular structures of polymer-solvent systems. Six different machine learning algorithms, including linear regression, k-nearest neighbors, regression trees, random forests, multilayer perceptron neural networks, and support vector regression, were employed to develop predictive models. Among these, the Random Forest model achieved the highest accuracy, with R2 values of 0.931 and 0.942 for temperature and pressure predictions, respectively. This novel approach eliminates the need for repetitive experimental testing, offering a more efficient pathway to determining dissolution conditions. © 2025 Wiley-VCH GmbH.
Publication Date: 2025
Global Knowledge, Memory and Communication (25149350) 74(1-2)pp. 220-234
Purpose: This study aims to determine the most similar set of recommendation books to the user selections in LibraryThing. Design/methodology/approach: For this purpose, 30,000 tags related to History on the LibraryThing have been selected. Their tags and the tags of the related recommended books were extracted from three different recommendations sections on LibraryThing. Then, four similarity criteria of Jaccard coefficient, Cosine similarity, Dice coefficient and Pearson correlation coefficient were used to calculate the similarity between the tags. To determine the most similar recommended section, the best similarity criterion had to be determined first. So, a researcher-made questionnaire was provided to History experts. Findings: The results showed that the Jaccard coefficient, with a frequency of 32.81, is the best similarity criterion from the point of view of History experts. Besides, the degree of similarity in LibraryThing recommendations section according to this criterion is equal to 0.256, in the section of books with similar library subjects and classifications is 0.163 and in the Member recommendations section is 0.152. Based on the findings of this study, the LibraryThing recommendations section has succeeded in introducing the most similar books to the selected book compared to the other two sections. Originality/value: To the best of the authors’ knowledge, itis for the first time, three sections of LibraryThing recommendations are compared by four different similarity criteria to show which sections would be more beneficial for the user browsing. The results showed that machine recommendations work better than humans. © 2023, Emerald Publishing Limited.
Publication Date: 2024
pp. 18-23
Reducing the size of transistors in multi-core processors has caused reduced energy consumption, fixed power density, and exponential growth in performance. In new generations of integrated circuits, despite the smaller transistors, the energy consumption has not been scaled down anymore; therefore, with constant power consumption in each transistor, the increasing number of transistors leads to an exponential growth in total power consumption, thermal concerns and dark silicon problems. In addition, the effects of aging are essential for the design and construction of integrated circuits. Since aging reduces the service lifetime of a circuit, this deterioration can affect all aspects including performance and reliability. Dynamic voltage and frequency scaling are power management methods used to control the workload. This paper presents a supervised learning-based method that predicts the performance of the system through some available input features and then adjusts the appropriate frequency and voltage for each workload. Simulation results show 95% accuracy in a multi-core processor using the decision tree method. © 2024 IEEE.
Publication Date: 2024
pp. 167-171
A recommender system typically assumes that each row of the user-item rating matrix reflects the preferences of a single user. However, in many cases, an account is shared among multiple household members, resulting in mixed ratings data that do not accurately represent individual preferences. As a consequence, the recommendations will fail to align with the specific interests of each user. To address this issue, we introduce the concept of a "user character," which represents a common latent factor in both movie and account features. By establishing a movie feature matrix based on these character representations, we can identify the presence of different characters in shared accounts. This is achieved by factoring the account feature binary matrix from the rating matrix, a process that can be modeled as a binary quadratic optimization problem. For scalability, we relax the binary constraint using a penalty function and approximate the solutions through the gradient descent method. Additionally, we apply a thresholding function to obtain binary solutions that reveal the user characters within each account. Once we identify the characters associated with each account, we can learn users' distinct preferences through a demixing procedure. This allows us to reconstruct the rating matrix so that each row accurately represents a single user's preferences. To evaluate our method, we generated a shared account dataset from MovieLens ratings based on the CAMRa2011 dataset. Experiments conducted on this dataset demonstrate the effectiveness of our proposed approach. © 2024 IEEE.
Publication Date: 2024
BMC Medical Informatics and Decision Making (14726947) 24(1)
Background: DNA microarrays provide informative data for transcriptional profiling and identifying gene expression signatures to help prevent progression of latent tuberculosis infection (LTBI) to active disease. However, constructing a prognostic model for distinguishing LTBI from active tuberculosis (ATB) is very challenging due to the noisy nature of data and lack of a generally stable analysis approach. Methods: In the present study, we proposed an accurate predictive model with the help of data fusion at the decision level. In this regard, results of filter feature selection and wrapper feature selection techniques were combined with multiple-criteria decision-making (MCDM) methods to select 10 genes from six microarray datasets that can be the most discriminative genes for diagnosing tuberculosis cases. As the main contribution of this study, the final ranking function was constructed by combining protein-protein interaction (PPI) network with an MCDM method (called Decision-making Trial and Evaluation Laboratory or DEMATEL) to improve the feature ranking approach. Results: By applying data fusion at the decision level on the 10 introduced genes in terms of fusion of classifiers of random forests (RF) and k-nearest neighbors (KNN) regarding Yager’s theory, the proposed algorithm reached a sensitivity of 0.97, specificity of 0.90, and accuracy of 0.95. Finally, with the help of cumulative clustering, the genes involved in the diagnosis of latent and activated tuberculosis have been introduced. Conclusions: The combination of MCDM methods and PPI networks can significantly improve the diagnosis different states of tuberculosis. Clinical trial number: Not applicable. © The Author(s) 2024.
Publication Date: 2024
pp. 119-124
Developing a distinct brand personality enables companies to differentiate themselves from competitors and effectively engage with customers. However, evaluating the customers' brand personality perceptions is a challenge, as traditional methods are costly and not objective. In this study, we focus on the brand personality dimension of "competence"and leverage the social networks to analyze the perceptions of Persian customers. For this purpose, the comments written in the Persian language regarding the brands or users' experiences with the brands are extracted from social networks. Then, the natural language processing techniques such as TF-IDF and word2vec are employed to prepare the data for developing machine learning models. These models categorize users' comments into three classes: aligned, non-aligned, and neutral. The classification depends on whether the comment is in line with the brand's competence, against it, or neutra. The k-nearest neighbors, Naive Bayes, Artificial Neural Network and long short-term memory (LSTM) are trained on the dataset. The results demonstrate that the LSTM model surpasses the performance of other models by achieving the f1-score of 93 percent. Finally, the LSTM model used to evaluate the customers' perception of Snowa's brand personality, as a case study. © 2024 IEEE.
Publication Date: 2024
pp. 190-194
In recent years, aspect-based sentiment analysis has gained attention as a method for processing and summarizing customer reviews. This type of analysis aims to answer two main questions: First, which aspect of the product is the review referring to? Second, what is the customer’s sentiment towards that aspect? Given the importance of after-sales service quality in customer satisfaction, this study examines an aspect-based analysis of a set of reviews from Digikala regarding the after-sales service quality of three washing machine brands, including Snowa and its two competitors. Deep learning models are implemented and evaluated based on example-based and label-based metrics. At the end, an analysis is conducted on customer satisfaction with the quality of after-sales services for the three selected brands. © 2024 IEEE.
Publication Date: 2024
Information Sciences (0020-0255) 669
Recommender systems focusing solely on accuracy, defined as the similarity of items to users' interests, often encounter the long-tail problem. This issue arises because short-head items, having received more ratings, dominate users' recommendation lists, while long-tail items, which are less popular, are underrepresented to maintain recommendation list precision. Unlike other approaches that treat users' preferences as fixed, this work advocates for considering dynamic user preferences to address the long-tail problem effectively. Specifically, we demonstrate two key observations: 1) users register varying proportions of ratings for long-tail and short-head items over time, and 2) item popularity is dynamic and undergoes changes over time. Consequently, recommendation lists can be dynamically adjusted to include different proportions of popular and unpopular items. We propose adapting the recommendation lists based on users' tenure in the system and their accrued ratings, allowing for higher inclusion of long-tail items for users with longer membership and more registered ratings. Additionally, we maintain an updated list of popular items as their popularity can fluctuate over time. In this study, modifications are made to the memetic algorithm to leverage users' dynamic preferences, demonstrating notable improvements. The proposed method achieves a precision of 90%, surpassing related works by 7% in addressing the long-tail problem, leading to increased participation of unpopular items in recommendation lists. © 2024 Elsevier Inc.
Publication Date: 2024
Multimedia Tools and Applications (13807501) 83(27)pp. 69973-69987
These days, due to the increasing amount of information generated on the web, most web service providers try to personalize their services. Users also interact with web-based systems in multiple ways and state their interests and preferences by rating the provided items. In this paper, we propose a framework to predict users’ demographic based on ratings registered by users in a system. To the best of our knowledge, this is the first time that the item ratings are employed for users’ demographic prediction problem, which has extensively been studied in recommendation systems and service personalization. We apply the framework to Movielens dataset’s ratings and predict users’ age and gender. The experimental results show that using all ratings registered by users improves the prediction accuracy by at least 16% compared with previously studied models. Moreover, by classifying the items as popular and unpopular, we eliminate ratings belong to 95% of items and still reach an acceptable level of accuracy. This significantly reduces update cost in a time-varying environment. Besides this classification, we propose other methods to reduce data volume while keeping the predictions accurate. © The Author(s), under exclusive licence to Springer Science+Business Media, LLC, part of Springer Nature 2024.
Publication Date: 2023
International Journal of Engineering, Transactions B: Applications (1728144X) 36(2)pp. 335-347
Gender is an important aspect of a person's identity. In many applications, gender identification is useful for personalizing services and recommendations. On the other hand, many people today spend a lot of time on their mobile phones. Studies have shown that the way users interact with mobile phones is influenced by their gender. But the existing methods for identify the gender of mobile phone users are either not accurate enough or require sensors and specific user activities. In this paper, for the first time, the internet usage patterns are used to identify the gender of mobile phone users. To this end, the interaction data, and specially the internet usage patterns of a random sample of people are automatically recorded by an application installed on their mobile phones. Then, the gender identification is modeled using different machine learning classification methods. The evaluations showed that the internet features play an important role in recognizing the users gender. The linear support vector machine was the superior classifier with the accuracy of 85% and F-measure of 85%. © 2023 Materials and Energy Research Center. All rights reserved.
Publication Date: 2022
Journal of Systems and Information Technology (17588847) 24(2)pp. 112-130
Purpose: Online businesses require a deep understanding of their customers’ interests to innovate and develop new products and services. Users, on the other hand, rarely express their interests explicitly. The purpose of this study is to predict users’ implicit interest in products of an online store based on their mouse behavior through various product page elements. Design/methodology/approach: First, user mouse behavior data is collected throughout an online store website. Next, several mouse behavioral features on the product pages elements are extracted and finally, several models are extracted using machine learning techniques to predict a user’s interest in a product. Findings: The results indicate that focusing on mouse behavior on various page elements improves user preference prediction accuracy compared to other available methods. Research limitations/implications: User mouse behavior was used to predict consumer preferences in this study, therefore gathering additional data on user demography, personality dimensions and emotions may significantly aid in accurate prediction. Originality/value: Mouse behavior is the most repeated behavior during Web page browsing through personal computers and laptops. It has been referred to as implicit feedback in some studies and an effective way to ascertain user preference. In these studies, mouse behavior is only assessed throughout the entire Web page, lacking a focus on different page elements. It is assumed that in online stores, user interaction with key elements of a product page, such as an image gallery, user reviews, a description and features and specifications, can be highly informative and aid in determining the user’s interest in that product. © 2022, Emerald Publishing Limited.
Publication Date: 2022
Applied Soft Computing (1568-4946) 128
In cluster-based sensor networks, at each cluster, sensor nodes send the collected data to a cluster head which aggregates and forwards them to a sink node. Data transmission from a cluster head to the sink node can be done in a multi-hop fashion through other cluster heads. Hence, two problems need to be addressed in this regard: Selection of cluster heads, and optimal multi-hop routing. In previous studies, these two problems have been solved separately in two independent phases. This paper proposes a novel approach to solve them simultaneously in order to increase the network lifetime. In the proposed scheme, the cluster head's role in transmitting the inter-cluster traffic is considered during the cluster head selection process. In other words, cluster heads are selected in a way which reduces the energy consumption for transmitting data from a cluster head to the sink node. To achieve this goal, the genetic algorithm is used in two levels. The first-level genetic algorithm selects the cluster heads while the second-level one considers multi-hop routing among them. Simulation of the proposed method and comparison of its results with three previously proposed schemes which solve the problems separately indicate the superiority of the proposed optimization scheme in improving the lifetime of the network. © 2022 Elsevier B.V.
Publication Date: 2021
Multimedia Tools and Applications (13807501) 80(9)pp. 13559-13574
In emotion-aware music recommender systems, the user’s current emotion is identified and considered in recommending music to him. We have two motivations to extend the existing systems: (1) to the best of our knowledge, the current systems first estimate the user’s emotions and then suggest music based on it. Therefore, the emotion estimation error affects the recommendation accuracy. (2) Studies show that the pattern of users’ interactions with input devices can reflect their emotions. However, these patterns have not been used yet in emotion-aware music recommender systems. In this study, a music recommender system is proposed to suggest music based on users’ keystrokes and mouse clicks patterns. Unlike the previous ones, the proposed system maps these patterns directly to the user’s favorite music, without labeling its current emotion. The results show that even though this system does not use any additional device, it is highly accurate compared to previous methods. © 2021, The Author(s), under exclusive licence to Springer Science+Business Media, LLC part of Springer Nature.
Publication Date: 2021
Pervasive and Mobile Computing (15741192) 73
The daily and routine interactions of people with their smartphones in various situations make these devices valuable data sources to understand user behaviors. Passive users’ emotion recognition is one of the most essential user modeling areas and has been studied for various purposes so far. Psychological studies, on the other hand, show that the personality of users can influence their behavior when they experience different emotions. Individuals with varying types of personality exhibit different reactions in the same emotional situation. It is concluded that if we consider the user's personality in passive recognition of his/her emotion, the emotion can be identified more accurately. However, researchers have not paid enough attention to the users’ personality traits when identifying the users’ emotions based on their interaction with cell phones. In the present study, we strive to address this research gap. Among the various emotions, our focus is on happiness recognition. In our proposed method, the user's personality traits are first estimated based on his/her interactions with the smartphone. Then the estimated personality of the user, along with the data of his/her interactions with the smartphone, is taken into account to recognize his/her happiness. Evaluations showed that taking into account the users’ personality traits reduces the happiness recognition error. © 2021 Elsevier B.V.
Publication Date: 2020
Education and Information Technologies (13602357) 25(2)pp. 985-996
The use of online intervention in providing career counseling and guidance is one of the practical methods to help people improve their understanding of their conditions and existing career conditions. This method helps people to take fundamental steps in the decision-making process. Accordingly, the objective of the present study was to investigate the effect of online counseling and face-to-face counseling with a guidance paradigm on career decision-making self-efficacy of students of the University of Isfahan. For this purpose, three groups of students including the face-to-face group, the online group, and the control group were created. Different analyses pre and post tests for these groups showed that students in the online group were similar to students in the face-to-face group in terms of career decision-making self-efficacy and both interventions promoted career decision-making self-efficacy of students relative to the control group. © 2019, Springer Science+Business Media, LLC, part of Springer Nature.
Publication Date: 2020
International Journal of Artificial Intelligence (09740635) 18(2)pp. 113-131
Interactive evolutionary algorithms are a class of evolutionary algorithms adopted for customer centric product design. During the run of such algorithms, the customer (user) acts as a fitness function to evaluate the candidate designs based on his/her interests and preferences. These algorithms are usually iterated frequently to find the desirable design of customer; hence, the user fatigue problem during interaction with these algorithms is a major challenge. The present study develops a method to tackle this problem. In this method, the desired designs of former users are considered as valuable knowledge to support the algorithm execution in the future. This knowledge is applied to enrich the populations of interactive genetic algorithm to speed up finding the desired designs of users. The proposed method has been used for customer centric design of book covers. The results show that the proposed method improves the speed of algorithm and increase the user satisfaction. © 2020 [International Journal of Artificial Intelligence].
Publication Date: 2019
Journal of Theoretical and Applied Electronic Commerce Research (07181876) 14(1)pp. 16-29
The decisions made by the customers in online environments are influenced by their personality characteristics. Each customer in an online environment relies more heavily on certain features of a store to make decisions while ignoring others. Thus, personalizing these features may streamline the decision-making process and increase satisfaction. In this paper, an intelligent method for personalizing the features of an online store according to the users’ personality is presented. In the proposed method, using genetic programming several equations are developed to estimate how users with different personality characteristics prefer various features of an online store. These equations are then used for personalization of the store features to increase customers’ satisfaction and persuade them to make larger purchases. The evaluation on a sample of 194 individuals indicates that the obtained equations are able to estimate the user’s preferences with over 80% accuracy in most cases. In addition, empirical assessment of the obtained equations shows that the proposed personalization method improves the user satisfaction. © 2019 Universidad de Talca-Chile.
Publication Date: 2019
Knowledge-Based Systems (0950-7051) 164pp. 348-357
Recommender systems which focus only on the improvement of recommendations’ accuracy are named “accuracy-centric”. These systems encounter some problems the major of which is their failure in recommending long tail items. Long tail items are the ones rated by a few users, thus, their rare participation in recommendations. To overcome this problem, it is necessary to provide recommendations by considering other aspects in addition to accuracy. One of these aspects is diversity in recommendations. As to different users who may prefer different levels of diversity in recommendations, here diversification of recommendations in a personalized manner is suggested in order to increase the participation of long tail items. The recommendation list is optimized based on three objectives of increasing the accuracy, personalizing the diversity, and reducing the popularity of the recommended items to meet the purpose. The defined multi-objective optimization problem is solved through the archived multi-objective simulated annealing algorithm. The evaluation of this proposed method on the Netflix datasets reveals that this method overcomes the long tail recommendation problem and diversifies the recommendations according to user needs while maintaining an acceptable level of accuracy. © 2018 Elsevier B.V.
Publication Date: 2018
International Journal of Applied Metaheuristic Computing (19478283) 9(1)pp. 40-48
This paper develops a nature-inspired metaheuristic algorithm named sun and leaf optimization (SLO) which is inspired by the effect of sunlight on the leaves germination. In SLO, candidate solutions in the state space are considered as leaves grown on a tree, and high-quality solutions are considered as greener leaves germinated in the direction of sunlight. On a tree, usually greener leaves are found closed to each other, because such area is probably exposed more to the sun and hence it is suitable for hosting other greener leaves. Inspired by this phenomenon, in SLO, during the search, the authors take the existence of high quality solutions as a sign of promising areas for finding optimum; thus, they generate more candidate solutions near the higher quality solutions to search those areas more painstakingly. Wind effect is imitated to escape the local optima. The evaluation results demonstrate the high performance of proposed algorithm. Copyright © 2018, IGI Global.
Publication Date: 2018
Education and Information Technologies (13602357) 23(6)pp. 2655-2672
Counseling through the internet is one of the provided facilities by modern technologies that paves the way for the career development of students. This study aims to investigate and describe the role and effect of online career counseling interventions on the career development of students. In the current study, 45 university students were randomly assigned into three groups of online counseling (15 students), face-to-face counseling (15 students), and control (15 students). Participants completed short form of career development inventory (Creed and Patton 2004). The collected data in pretest, posttest, follow-up 1, and follow-up 2 were analyzed using SPSS package at descriptive and inferential levels as well as analysis of variance with repetitive measurements. The results showed that both interventions increased the students’ level of career development as compared to that of the students in the control group. © 2018, Springer Science+Business Media, LLC, part of Springer Nature.
Publication Date: 2018
Electronic Commerce Research (13895753) 18(4)pp. 813-836
In collaborative filtering recommender systems, items recommended to an active user are selected based on the interests of users similar to him/her. Collaborative filtering systems suffer from the ‘sparsity’ and ‘new user’ problems. The former refers to the insufficiency of data about users’ preferences and the latter addresses the lack of enough information about the new-coming user. Clustering users is an effective way to improve the performance of collaborative filtering systems in facing the aforementioned problems. In previous studies, users were clustered based on characteristics such as ratings given by them as well as their age, gender, occupation, and geographical location. On the other hand, studies show that there is a significant relationship between users’ personality traits and their interests. To alleviate the sparsity and new user problems, this paper presents a new collaborative filtering system in which users are clustered based on their ‘personality traits’. In the proposed method, the personality of each user is described according to the big-5 personality model and users with similar personality are placed in the same cluster using K-means algorithm. The unknown ratings of the sparse user-item matrix are then estimated based on the clustered users, and recommendations are found for a new user according to a user-based approach which relays on the interests of the users with similar personality to him/her. In addition, for an existing user in the system, recommendations are offered in an item-based approach in which the similarity of items is estimated based on the ratings of users similar to him/her in personality. The proposed method is compared to some former collaborative filtering systems. The results demonstrate that in facing the data sparsity and new user problems, this method reduces the mean absolute error and improves the precision of the recommendations. © 2018, Springer Science+Business Media, LLC, part of Springer Nature.
Publication Date: 2017
Journal Of Medical Signals And Sensors (22287477) 7(3)pp. 163-169
Wireless body area networks consist of several devices placed on the human body, sensing vital signs and providing remote recognition of health disorders. Low power consumption is crucial in these networks. A new energy-efficient topology is provided in this paper, considering relay and sensor nodes' energy consumption and network maintenance costs. In this topology design, relay nodes, placed on the cloth, are used to help the sensor nodes forwarding data to the sink. Relay nodes' situation is determined such that the relay nodes' energy consumption merges the uniform distribution. Simulation results show that the proposed method increases the lifetime of the network with nearly uniform distribution of the relay nodes' energy consumption. Furthermore, this technique simultaneously reduces network maintenance costs and continuous replacements of the designer clothing. The proposed method also determines the way by which the network traffic is split and multipath routed to the sink.
Publication Date: 2017
International Journal of Artificial Intelligence (09740635) 15(1)pp. 76-92
In this paper a population-based metaheuristic algorithm named fractal-based algorithm is developed to solve continuous optimization problems. In this algorithm, the density of high quality and promising points in an area is considered as a heuristic which estimates the degree of promise of that area for finding the optimal solution. Afterward, the promising areas of state space are iteratively detected and partitioned into self-similar and fractalshaped subspaces for being searched more precisely and more extensively. The proposed algorithm is compared with some metaheuristic algorithms. The results demonstrate that the algorithm is able to find high quality solutions within appropriate time. © 2017 [International Journal of Artificial Intelligence].
Publication Date: 2017
Human-centric Computing and Information Sciences (21921962) 7(1)
One of the effective factors in increasing sales is the consistency of products with the preference of the customers. Designing the products consistent with customer needs requires the engagement of customers in the product design process. One way to achieve this goal is the use of interactive evolutionary algorithms. During the running of such algorithms, the customer acts as a fitness function and imparts his/her opinion directly to the design process. Since these algorithms are usually iterated frequently, the user fatigue problem during interaction with them is a major challenge. The present study develops a method to tackle the user fatigue problem in the interactive genetic algorithm using the candidate elimination algorithm. In this method, customer preferences are gradually learned by applying the candidate elimination algorithm on the designs evaluated by the user in the early stages of algorithm. Using the learned preferences, designs which may not meet the customer preferences are discovered and automatically receive a predefined low score from the algorithm. The proposed method has been evaluated on the customer-centric design of book covers and its results have been compared with those of the two simple interactive genetic algorithm and multi-stage interactive genetic algorithm. The results are indicative of a considerable reduction of the number of algorithm generations, the number of chromosomes being evaluated by user, and the evaluating time in comparison with the two aforementioned methods. Reduction of these criteria leads to decrease of user fatigue. In addition, the proposed method has increased the user satisfaction. © 2017, The Author(s).
Publication Date: 2017
Applied Soft Computing (1568-4946) 61pp. 1125-1138
Probabilistic robustness evaluation is a promising approach to evolutionary robust optimization; however, high computational time arises. In this paper, we apply this approach to the Bayesian optimization algorithm (BOA) with a view to improving its computational time. To this end, we analyze the Bayesian networks constructed in BOA in order to extract the patterns of non-robust solutions. In each generation, the solutions that match the extracted patterns are detected and then discarded from the process of evaluation; therefore, the computational time in discovering the robust solutions decreases. The experimental results demonstrate that our proposed method reduces computational time, while increasing the robustness of solutions. © 2017
Publication Date: 2016
Information Sciences (0020-0255) 334pp. 44-64
When memory-based evolutionary algorithms are applied in dynamic environments, the certainly use of uncertain prior knowledge for future environments may mislead the evolutionary algorithms. To address this problem, this paper presents a new, memory-based evolutionary approach for applying the Bayesian optimization algorithm (BOA) in dynamic environments. Our proposed method, unlike existing memory-based methods, uses the knowledge of former environments probabilistically in future environments. For this purpose, the run of BOA is modeled as the movements in a Markov chain, in which the states become the Bayesian networks that are learned in every generation. When the environment changes, a stationary distribution of the Markov chain is defined on the basis of the retrieved prior knowledge. Then, the transition probabilities of BOA in the Markov chain are modified (biased) to comply with the defined stationary distribution. To this end, we employ the Metropolis algorithm and modify the K2 algorithm for learning the Bayesian network in BOA in order to reflect the obtained transition probabilities. Experimental results show that the proposed method achieves improved performance compared to conventional methods, especially in random environments. © 2015 Elsevier Inc. All rights reserved.
Publication Date: 2016
International Journal of Business Intelligence and Data Mining (17438195) 11(3)pp. 229-241
In impulsive shopping, proper packaging of a product attracts the consumer and makes that particular product outstanding among its counter parts on the same shelf. Colour is one of the major elements in packaging design. In food products packages, colour convey subconscious information regarding quality, taste, pleasure and even the price to the consumer. The objective here is to determine which one of the colours or their combination, on food products like as cakes, biscuits, and milk would lead to an increase in sales of the products' categories. Here for each of the food products, some brands with similar quality, price and commercialisation are selected. Then, the proper colours for packaging of each of the products are extracted using data mining. The results indicate that the extracted colours can contribute to distinction between slow-seller and best-seller brands by 21%; therefore, the use of such colours would increase the products' sales. © 2016 Inderscience Enterprises Ltd.
Publication Date: 2016
Applied Soft Computing (1568-4946) 45pp. 187-196
In the new decade due to rich and dense water resources, it is vital to have an accurate and reliable sediment prediction and incorrect estimation of sediment rate has a huge negative effect on supplying drinking and agricultural water. For this reason, many studies have been conducted in order to improve the accuracy of prediction. In a wide range of these studies, various soft computing techniques have been used to predict the sediment. It is expected that combining the predictions obtained by these soft computing techniques can improve the prediction accuracy. Stacking method is a powerful machine learning technique to combine the prediction results of other methods intelligently through a meta-model based on cross validation. However, to the best of our knowledge, the stacking method has not been used to predict sediment or other hydrological parameters, so far. This study introduces stacking method to predict the suspended sediment. For this purpose, linear genetic programming and neuro-fuzzy methods are applied as two successful soft computing methods to predict the suspended sediment. Then, the accuracy of prediction is increased by combining their results with the meta-model of neural network based on cross validation. To evaluate the proposed method, two stations including Rio Valenciano and Quebrada Blanca, in the USA were selected as case studies and streamflow and suspended sediment concentration were defined as inputs to predict the daily suspended sediment. The obtained results demonstrated that the stacking method greatly improved RMSE and R2 statistics for both stations compared to use of linear genetic programming or neuro-fuzzy solitarily. © 2016 Elsevier B.V. All rights reserved.
Publication Date: 2015
Cybernetics and Systems (10876553) 46(5)pp. 355-378
Sending promotional messages to a few numbers of users in a social network can propagate a product through word of mouth. However, choosing users that receive promotional messages, in order to maximize propagation, is a considerable issue. These recipients are named "influential nodes." To recognize influential nodes, according to the literature, criteria such as the relationships of network members or information shared by each member on a social network have been used. One of the effective factors in diffusion of messages is the personality characteristics of members. As far as we know, although this issue is considerable, so far it has not been applied in the previous studies. In this article, using the graph structure of social networks, two personality characteristics, openness and extroversion, are estimated for network members. Next, these two estimated characteristics together with other characteristics of social networks, are considered as the criteria of choosing influential nodes. To implement this process, the real coded genetic algorithm is used. The proposed method has been evaluated on a dataset including 1000 members of Twitter. Our results indicate that using the proposed method, compared with simple heuristic methods, can improve performance up to 37% © Taylor & Francis Group, LLC.
Publication Date: 2013
International Journal of Innovative Computing, Information and Control (13494198) 9(6)pp. 2485-2503
In this paper, a new evolutionary algorithm termed DBN-MBOA (Memory-based BOA with Dynamic Bayesian Networks) is proposed for the dynamic optimization. In DBN-MBOA, the knowledge obtained from previously solved problems is encoded in some structures called network translators. The network translators defined on non-stationary Dynamic Bayesian Networks (nsDBNs) describe the correlation between conditional dependencies of candidate solution variables before and after environmental changes. The network translators constructed for the changes are stored in memory. When any change occurs in the environment, a relevant network translator is retrieved from the memory and is used for modifying the dependencies of the current Bayesian network. In the retrieve stage, unlike existing memory-based methods, the relevant network translator is selected based on the characteristic of the change itself, not that of the new environmental state. Experimental results show that DBN-MBOA achieves better performance in random environments as well as cyclic environments. © 2013 ICIC International.
Publication Date: 2013
Science China Information Sciences (1674733X) 56(9)pp. 1-17
This paper presents a new evolutionary dynamic optimization algorithm, holographic memory-based Bayesian optimization algorithm (HM-BOA), whose objective is to address the weaknesses of sequential memorybased dynamic optimization approaches. To this end, holographic associative neural memory is applied to one of the recent successful memory-based evolutionary methods, DBN-MBOA (memory-based BOA with dynamic Bayesian networks). Holographic memory is appropriate for encoding environmental changes since its stimulus and response data are represented by a vector of complex numbers such that the phase and the magnitude denote the information and its confidence level, respectively. In the learning process in HM-BOA, holographic memory is trained by probabilistic models at every environmental change. Its weight matrix contains abstract information obtained from previous changes and is used for constructing a new probabilistic model when the environment changes. The unique features of HM-BOA are: 1) the stored information can be generalized, and 2) a small amount of memory is required for storing the probabilistic models. Experimental results adduce grounds for its effectiveness especially in random environments. © 2013 Science China Press and Springer-Verlag Berlin Heidelberg.
Publication Date: 2012
Intelligent Data Analysis (1088467X) 16(2)pp. 199-210
The Case-Based Reasoning (CBR) solves problems by using the past problem solving experiences. How to apply these experiences depends on the type of the problem. The method presented in this paper tries to overcome this difficulty in CBR for optimization problems, using Bayesian Optimization Algorithm (BOA). BOA evolves a population of candidate solutions through constructing Bayesian networks and sampling them. After solving the problems through BOA, Bayesian networks describing solutions features are obtained. In our method, these Bayesian networks are stored in a case-base. For solving a new problem, the Bayesian networks of those problems which are similar to the new problem, are retrieved and combined. This compound Bayesian network is used for generating the initial population and constructing the probabilistic models of BOA in solving the new problem. Our method improves CBR in two ways: first, in our method, how to use the knowledge stored in the case-base is disregarding the problem itself and is universally; second, this method stores the probabilistic descriptions of the previous solutions in order to make the stored knowledge more flexible. Experimental results showed that in addition to the mentioned advantages, our method improved the solutions quality. © 2012 - IOS Press and the authors. All rights reserved.
Publication Date: 2011
Knowledge-Based Systems (0950-7051) 24(8)pp. 1245-1253
Studies show that application of the prior knowledge in biasing the Estimation of Distribution Algorithms (EDAs), such as Bayesian Optimization Algorithm (BOA), increases the efficiency of these algorithms significantly. One of the main advantages of the EDAs over other optimization algorithms is that the former provides a trail of probabilistic models of candidate solutions with increasing quality. Some recent studies have applied these probabilistic models, obtained from previously solved problems in biasing the BOA algorithm, to solve the future problems. In this paper, in order to improve the previous works and reduce their disadvantages, a method based on Case Based Reasoning (CBR) is proposed for biasing the BOA algorithm. Herein, after running BOA for solving optimization problems, each problem, the corresponding solution, as well as the last Bayesian network obtained from the BOA algorithm, will be stored as an entry in the case-base. Upon introducing a new problem, similar problems from the case-base are retrieved and the last Bayesian networks of these solved problems are combined according to the degree of their similarity with the new problem; hence, a compound Bayesian network is constructed. The compound Bayesian network is sampled and the initial population for the BOA algorithm is generated. This network will be applied efficiently for biasing future probabilistic models during the runs of BOA for the new problem. The proposed method is tested on three well-known combinatorial benchmark problems. Experimental results show significant improvements in algorithm execution time and quality of solutions, compared to previous methods. © 2011 Elsevier B.V. All rights reserved.
Publication Date: 2009
Mathematical and Computer Modelling of Dynamical Systems (17445051) 15(6)pp. 535-553
Personality and emotions are effective factors in human decision-making processes. Thus, when an agent has to emulate human behaviour, not only should this agent think and reason but also should have emotions and personality. In this article a fuzzy agent with dynamic personality is modelled based on a five-factor personality model and implemented in a Java environment. Then it is extended with a fuzzy emotion component. This emotion component uses calculated personality factors and some related parameters and then determines the degree of anger. The proposed personality and emotion model provides a proper framework for human-like agent decision-making tasks. © 2009 Taylor & Francis.
Publication Date: 2008
pp. 139-143
The major drawback of fuzzy data mining is that after applying fuzzy data mining on the quantitative data, the number of extracted fuzzy association rules is very huge. When many association rules are obtained, the usefulness of them will be reduced. In this paper, we introduce an approach to reduce and summarize the extracted fuzzy association rules after fuzzy data mining. In our approach, in first, we encode each obtained fuzzy association rule to a string of numbers. Then we use self-organizing map (SOM) neural network iteratively in a tree structure for clustering these encoded rules and summarizing them to a smaller collection of fuzzy association rules. This approach has been applied on a data base containing information about 5000 employees and has shown good results. © 2008 IADIS.
Publication Date: 2008
Neural Computing And Applications (09410643) 17(2)pp. 193-200
Optimizing the traffic signal control has an essential impact on intersections efficiency in urban transportation. This paper presents a two-stage method for intersection signal timing control. First, the traffic volume is predicted using a neuro-fuzzy network called Adaptive neuro-fuzzy inference system (ANFIS). The inputs of this network include two-dimensional, hourly and daily, traffic volume correlations. In the second stage, appropriate signal cycle and optimized timing of each phase of the signal are estimated using a combination of Self Organizing and Hopfield neural networks. The energy function of the Hopfield network is based on a traffic model derived by queuing analysis. The performance of the proposed method has been evaluated for real data. The two-dimensional correlation presents superior performance compared to hourly traffic correlation. The evaluation of proposed overall method shows considerable intersection throughput improvement comparing to the results taken form Synchro software. © 2007 Springer-Verlag London Limited.
Publication Date: 2007
Proceedings of the IASTED International Conference on Modelling and Simulation (10218181) pp. 44-49
In this paper we have developed a fuzzy intelligent agent with a fuzzy emotional component. This component evaluates the degree of anger emotion which the intelligent agent may feel when it encounters to a persecuting event. For this purpose we have implemented a fuzzy inference system using Java environment and we have considered three parameters related on human anger emotion as the inputs of this system. The system outputs can be used in agent decision making process by combination this anger intensity to other emotion intensities and applies them for choosing a proper action from some available actions.
Address Isfahan, Azadi Square, University of Isfahan
