Search results for: crow search algorithm

Authors: Ahmed E. Hodaib, Mohamed A. Hashem

Abstract:

In engineering applications, a design has to be as fully perfect as possible in some defined case. The designer has to overcome many challenges in order to reach the optimal solution to a specific problem. This process is called optimization. Generally, there is always a function called “objective function” that is required to be maximized or minimized by choosing input parameters called “degrees of freedom” within an allowed domain called “search space” and computing the values of the objective function for these input values. It becomes more complex when we have more than one objective for our design. As an example for Multi-Objective Optimization Problem (MOP): A structural design that aims to minimize weight and maximize strength. In such case, the Pareto Optimal Frontier (POF) is used, which is a curve plotting two objective functions for the best cases. At this point, a designer should make a decision to choose the point on the curve. Engineers use algorithms or iterative methods for optimization. In this paper, we will discuss the Evolutionary Algorithms (EA) which are widely used with Multi-objective Optimization Problems due to their robustness, simplicity, suitability to be coupled and to be parallelized. Evolutionary algorithms are developed to guarantee the convergence to an optimal solution. An EA uses mechanisms inspired by Darwinian evolution principles. Technically, they belong to the family of trial and error problem solvers and can be considered global optimization methods with a stochastic optimization character. The optimization is initialized by picking random solutions from the search space and then the solution progresses towards the optimal point by using operators such as Selection, Combination, Cross-over and/or Mutation. These operators are applied to the old solutions “parents” so that new sets of design variables called “children” appear. The process is repeated until the optimal solution to the problem is reached. Reliable and robust computational fluid dynamics solvers are nowadays commonly utilized in the design and analyses of various engineering systems, such as aircraft, turbo-machinery, and auto-motives. Coupling of Computational Fluid Dynamics “CFD” and Multi-Objective Evolutionary Algorithms “MOEA” has become substantial in aerospace engineering applications, such as in aerodynamic shape optimization and advanced turbo-machinery design.

Keywords: mathematical optimization, multi-objective evolutionary algorithms "MOEA", computational fluid dynamics "CFD", aerodynamic shape optimization

Procedia PDF Downloads 256

Authors: Soyoon Bak, Sunyoung Bu, Philsu Kim

Abstract:

In this paper, a backward semi-Lagrangian scheme combined with the second-order backward difference formula is designed to calculate the numerical solutions of nonlinear advection-diffusion equations. The primary aims of this paper are to remove any iteration process and to get an efficient algorithm with the convergence order of accuracy 2 in time. In order to achieve these objects, we use the second-order central finite difference and the B-spline approximations of degree 2 and 3 in order to approximate the diffusion term and the spatial discretization, respectively. For the temporal discretization, the second order backward difference formula is applied. To calculate the numerical solution of the starting point of the characteristic curves, we use the error correction methodology developed by the authors recently. The proposed algorithm turns out to be completely iteration-free, which resolves the main weakness of the conventional backward semi-Lagrangian method. Also, the adaptability of the proposed method is indicated by numerical simulations for Burgers’ equations. Throughout these numerical simulations, it is shown that the numerical results are in good agreement with the analytic solution and the present scheme offer better accuracy in comparison with other existing numerical schemes. Semi-Lagrangian method, iteration-free method, nonlinear advection-diffusion equation, second-order backward difference formula

Keywords: Semi-Lagrangian method, iteration free method, nonlinear advection-diffusion equation, second-order backward difference formula

Procedia PDF Downloads 322

Authors: Linyu Wang, Mingjun Zhu, Jianhong Xiang, Hanyu Jiang

Abstract:

Aiming at the problem that the spectral efficiency of hybrid precoding (HP) is too low in the current millimeter wave (mmWave) massive multiple input multiple output (MIMO) system, this paper proposes a digital joint equivalent channel hybrid precoding algorithm, which is based on the introduction of digital encoding matrix iteration. First, the objective function is expanded to obtain the relation equation, and the pseudo-inverse iterative function of the analog encoder is derived by using the pseudo-inverse method, which solves the problem of greatly increasing the amount of computation caused by the lack of rank of the digital encoding matrix and reduces the overall complexity of hybrid precoding. Secondly, the analog coding matrix and the millimeter-wave sparse channel matrix are combined into an equivalent channel, and then the equivalent channel is subjected to Singular Value Decomposition (SVD) to obtain a digital coding matrix, and then the derived pseudo-inverse iterative function is used to iteratively regenerate the simulated encoding matrix. The simulation results show that the proposed algorithm improves the system spectral efficiency by 10~20%compared with other algorithms and the stability is also improved.

Keywords: mmWave, massive MIMO, hybrid precoding, singular value decompositing, equivalent channel

Procedia PDF Downloads 96

Authors: Emmanuel Okechukwu Nwafor, Folake Olubunmi Akintayo, Denis Alcides Rezende

Abstract:

Decision tree models are versatile and interpretable machine learning algorithms widely used for both classification and regression tasks, which can be related to cities, whether physical or digital. The aim of this research is to assess how well decision tree algorithms can predict trip purposes in Makurdi, Nigeria, while also exploring their connection to the strategic digital city initiative. The research methodology involves formalizing household demographic and trips information datasets obtained from extensive survey process. Modelling and Prediction were achieved using Python Programming Language and the evaluation metrics like R-squared and mean absolute error were used to assess the decision tree algorithm's performance. The results indicate that the model performed well, with accuracies of 84% and 68%, and low MAE values of 0.188 and 0.314, on training and validation data, respectively. This suggests the model can be relied upon for future prediction. The conclusion reiterates that This model will assist decision-makers, including urban planners, transportation engineers, government officials, and commuters, in making informed decisions on transportation planning and management within the framework of a strategic digital city. Its application will enhance the efficiency, sustainability, and overall quality of transportation services in Makurdi, Nigeria.

Keywords: decision tree algorithm, trip purpose, intelligent transport, strategic digital city, travel pattern, sustainable transport

Procedia PDF Downloads 21

Authors: S. Niamkaeo, O. Robert, O. Chaowalit

Abstract:

In this investigation, we focus on discovering spatial relationship of drug smuggling along the northern border of Thailand. Thailand is no longer a drug production site, but Thailand is still one of the major drug trafficking hubs due to its topographic characteristics facilitating drug smuggling from neighboring countries. Our study areas cover three districts (Mae-jan, Mae-fahluang, and Mae-sai) in Chiangrai city and four districts (Chiangdao, Mae-eye, Chaiprakarn, and Wienghang) in Chiangmai city where drug smuggling of methamphetamine crystal and amphetamine occurs mostly. The data on drug smuggling incidents from 2011 to 2017 was collected from several national and local published news. Geo-spatial drug smuggling database was prepared. Decision tree algorithm was applied in order to discover the spatial relationship of factors related to drug smuggling, which was converted into rules using rule-based system. The factors including land use type, smuggling route, season and distance within 500 meters from check points were found that they were related to drug smuggling in terms of rules-based relationship. It was illustrated that drug smuggling was occurred mostly in forest area in winter. Drug smuggling exhibited was discovered mainly along topographic road where check points were not reachable. This spatial relationship of drug smuggling could support the Thai Office of Narcotics Control Board in surveillance drug smuggling.

Keywords: decision tree, drug smuggling, Geographic Information System, GIS knowledge discovery, rule-based system

Procedia PDF Downloads 169

Authors: Frederick W. Gooding, Jr., Mark Beeman

Abstract:

Nearly a century ago, the groundbreaking 1915 film ‘The Birth of a Nation’ popularized the way Hollywood made movies with its avant-garde, feature-length style. The movie's subjugating and demeaning depictions of African American women (and men) reflected popular racist beliefs held during the time of slavery and the early Jim Crow era. Although much has changed concerning race relations in the past century, American sociologist Patricia Hill Collins theorizes that the disparaging images of African American women originating in the era of plantation slavery are adaptable and endure as controlling images today. In this context, a comparative analysis of the successful contemporary film, ‘Bringing Down the House’ starring Queen Latifah is relevant as this 2004 film was designed to purposely defy and ridicule classic stereotypes of African American women. However, the film is still tied to the controlling images from the past, although in a modern hybrid form. Scholars of race and film have noted that the pervasive filmic imagery of the African American woman as the loyal mammy stereotype faded from the screen in the post-civil rights era in favor of more sexualized characters (i.e., the Jezebel trope). Analyzing scenes and dialogue through the lens of sociological and critical race theory, the troubling persistence of African American controlling images in film stubbornly emerge in a movie like ‘Bringing Down the House.’ Thus, these controlling images, like racism itself, can adapt to new social and economic conditions. Although the classic controlling images appeared in the first feature length film focusing on race relations a century ago, ‘The Birth of a Nation,’ this black and white rendition of the mammy figure was later updated in 1939 with the classic hit, ‘Gone with the Wind’ in living color. These popular controlling images have loomed quite large in the minds of international audiences, as ‘Gone with the Wind’ is still shown in American theaters currently, and experts at the British Film Institute in 2004 rated ‘Gone with the Wind’ as the number one movie of all time in UK movie history based upon the total number of actual viewings. Critical analysis of character patterns demonstrate that images that appear superficially benign contribute to a broader and quite persistent pattern of marginalization within the aggregate. This approach allows experts and viewers alike to detect more subtle and sophisticated strands of racial discrimination that are ‘hidden in plain sight’ despite numerous changes in the Hollywood industry that appear to be more voluminous and diverse than three or four decades ago. In contrast to white characters, non-white or minority characters are likely to be subtly compromised or marginalized relative to white characters if and when seen within mainstream movies, rather than be subjected to obvious and offensive racist tropes. The hybrid form of both the older Jezebel and Mammy stereotypes exhibited by lead actress Queen Latifah in ‘Bringing Down the House’ represents a more suave and sophisticated merging of past imagery ideas deemed problematic in the past as well as the present.

Keywords: African Americans, Hollywood film, hybrid, stereotypes

Procedia PDF Downloads 178

Authors: Elisa Boatti, Ulisse Stefanelli, Alessandro Reali, Ferdinando Auricchio

Abstract:

Thanks to their unusual properties, shape memory alloys (SMAs) are good candidates for advanced applications in a wide range of engineering fields, such as automotive, robotics, civil, biomedical, aerospace. In the last decades, the ever-growing interest for such materials has boosted several research studies aimed at modeling their complex nonlinear behavior in an effective and robust way. Since the constitutive response of SMAs is strongly thermomechanically coupled, the investigation of the non-isothermal evolution of the material must be taken into consideration. The present study considers an existing three-dimensional phenomenological model for SMAs, able to reproduce the main SMA properties while maintaining a simple user-friendly structure, and proposes a variational reformulation of the full non-isothermal version of the model. While the considered model has been thoroughly assessed in an isothermal setting, the proposed formulation allows to take into account the full nonisothermal problem. In particular, the reformulation is inspired to the GENERIC (General Equations for Non-Equilibrium Reversible-Irreversible Coupling) formalism, and is based on a generalized gradient flow of the total entropy, related to thermal and mechanical variables. Such phrasing of the model is new and allows for a discussion of the model from both a theoretical and a numerical point of view. Moreover, it directly implies the dissipativity of the flow. A semi-implicit time-discrete scheme is also presented for the fully coupled thermomechanical system, and is proven unconditionally stable and convergent. The correspondent algorithm is then implemented, under a space-homogeneous temperature field assumption, and tested under different conditions. The core of the algorithm is composed of a mechanical subproblem and a thermal subproblem. The iterative scheme is solved by a generalized Newton method. Numerous uniaxial and biaxial tests are reported to assess the performance of the model and algorithm, including variable imposed strain, strain rate, heat exchange properties, and external temperature. In particular, the heat exchange with the environment is the only source of rate-dependency in the model. The reported curves clearly display the interdependence between phase transformation strain and material temperature. The full thermomechanical coupling allows to reproduce the exothermic and endothermic effects during respectively forward and backward phase transformation. The numerical tests have thus demonstrated that the model can appropriately reproduce the coupled SMA behavior in different loading conditions and rates. Moreover, the algorithm has proved effective and robust. Further developments are being considered, such as the extension of the formulation to the finite-strain setting and the study of the boundary value problem.

Keywords: generalized gradient flow, GENERIC formalism, shape memory alloys, thermomechanical coupling

Procedia PDF Downloads 221

Authors: Jutharat Attawet, Alex Y. Wang, Cindy M. Farquhar, Elizabeth A. Sullivan

Abstract:

Background: Adverse maternal and perinatal outcomes of multiple pregnancies resulting from multiple embryo transfers (ET) has become significant concerns. This is particularly relevant for gestational carriers since they usually do not have infertility issues. Single embryo transfer (SET) therefore has been encouraged to assist reproductive technology (ART) practice in order to reduce multiple pregnancies. Objectives: This systematic review aims to investigate the pregnancy and birth outcomes of SET and multiple ET in surrogacy arrangements. Search methods: This study is a systematic review. Electronic databases were searched from CINAHL, Medline, Embase, Scopus and ProQuest for studies from 1980 to 2017. Cross-references and national ART reports were also manual searchings. Articles without restriction of English language and study types were accessed. Carrier cycles involving in SET and multiple ET were identified in database searching. The main outcome measures including clinical pregnancy, live delivery and multiple deliveries per gestational carrier cycle were compared between SET and multiple ET. Mantel-Haenzel risk ratios (RRs) with 95% confidence intervals (CIs), using the numbers of outcome events in SET and multiple ET of each study were calculated suing RevMan5.3. Outcomes: The search returned 97 articles of which 5 met the inclusion criteria. Approximately 50% of carrier cycles were transferred a single embryo and 50% were transferred more than one embryo. The clinical pregnancy rate (CPR) was 39% for SET and 53% for multiple ET, which was not significantly different with RR = 0.83 (95% CI: 0.67-1.03). The live delivery rate was 33% for SET and 57% for multiple ET which was not significantly different with RR = 0.78 (95% CI: 0.61-1.00). The multiple delivery rate per carrier was greater risks in the multiple ET carrier cycles (RR =0.4, 95% CI: 0.01-0.26). There were 104 sets of twins (including one set of twins selectively reduced from triplets to twins) and 1 set of triples in the multiple ET carrier cycle. In the SET carrier cycles, there were 2 sets of twins. Significance of the study: SET should be advocated among surrogate carriers to prevent multiple pregnancies and subsequent adverse outcomes for both carrier and baby. Surrogacy practice should be reviewed and surrogate carriers should be fully informed of the risk of adverse maternal and birth outcome of multiple pregnancies due to multiple embryo transfers.

Keywords: assisted reproduction, birth outcomes, carrier, gestational surrogacy, multiple embryo transfer, multiple pregnancy, pregnancy outcomes, single embryo transfer, surrogate mother, systematic review

Procedia PDF Downloads 404

Authors: Aussara Panya, Nunghathai Sawasdee, Mutita Junking, Chatchawan Srisawat, Kiattawee Choowongkomon, Pa-Thai Yenchitsomanus

Abstract:

Dengue virus (DENV) infection is a global public health problem with approximately 100 million infected cases a year. Presently, there is no approved vaccine or effective drug available; therefore, the development of anti-DENV drug is urgently needed. The clinical reports revealing the positive association between the disease severity and viral titer has been reported previously suggesting that the anti-DENV drug therapy can possibly ameliorate the disease severity. Although several anti-DENV agents showed inhibitory activities against DENV infection, to date none of them accomplishes clinical use in the patients. The surface envelope (E) protein of DENV is critical for the viral entry step, which includes attachment and membrane fusion; thus, the blocking of envelope protein is an attractive strategy for anti-DENV drug development. To search the safe anti-DENV agent, this study aimed to search for novel peptide inhibitors to counter DENV infection through the targeting of E protein using a structure-based in silico design. Two selected strategies has been used including to identify the peptide inhibitor which interfere the membrane fusion process whereby the hydrophobic pocket on the E protein was the target, the destabilization of virion structure organization through the disruption of the interaction between the envelope and membrane proteins, respectively. The molecular docking technique has been used in the first strategy to search for the peptide inhibitors that specifically bind to the hydrophobic pocket. The second strategy, the peptide inhibitor has been designed to mimic the ectodomain portion of membrane protein to disrupt the protein-protein interaction. The designed peptides were tested for the effects on cell viability to measure the toxic to peptide to the cells and their inhibitory assay to inhibit the DENV infection in Vero cells. Furthermore, their antiviral effects on viral replication, intracellular protein level and viral production have been observed by using the qPCR, cell-based flavivirus immunodetection and immunofluorescence assay. None of tested peptides showed the significant effect on cell viability. The small peptide inhibitors achieved from molecular docking, Glu-Phe (EF), effectively inhibited DENV infection in cell culture system. Its most potential effect was observed for DENV2 with a half maximal inhibition concentration (IC50) of 96 μM, but it partially inhibited other serotypes. Treatment of EF at 200 µM on infected cells also significantly reduced the viral genome and protein to 83.47% and 84.15%, respectively, corresponding to the reduction of infected cell numbers. An additional approach was carried out by using peptide mimicking membrane (M) protein, namely MLH40. Treatment of MLH40 caused the reduction of foci formation in four individual DENV serotype (DENV1-4) with IC50 of 24-31 μM. Further characterization suggested that the MLH40 specifically blocked viral attachment to host membrane, and treatment with 100 μM could diminish 80% of viral attachment. In summary, targeting the hydrophobic pocket and M-binding site on the E protein by using the peptide inhibitors could inhibit DENV infection. The results provide proof of-concept for the development of antiviral therapeutic peptide inhibitors to counter DENV infection through the use of a structure-based design targeting conserved viral protein.

Keywords: dengue virus, dengue virus infection, drug design, peptide inhibitor

Procedia PDF Downloads 357

Authors: Hermann Dornhackl, Konstantin Kadletz, Robert Luh, Paul Tavolato

Abstract:

In this paper we describe the use of formal methods to model malware behaviour. The modelling of harmful behaviour rests upon syntactic structures that represent malicious procedures inside malware. The malicious activities are modelled by a formal grammar, where API calls’ components are the terminals and the set of API calls used in combination to achieve a goal are designated non-terminals. The combination of different non-terminals in various ways and tiers make up the attack vectors that are used by harmful software. Based on these syntactic structures a parser can be generated which takes execution traces as input for pattern recognition.

Keywords: malware behaviour, modelling, parsing, search, pattern matching

Procedia PDF Downloads 332

Authors: Caroline Smith, Professor Debi Bhattacharya, Sion Scott

Abstract:

Introduction: Oropharyngeal Dysphagia (OD) effects around 15% of older people, however it is often unrecognised and under diagnosed until they are hospitalised. There is a need for primary care healthcare practitioners (HCPs) to assume a proactive role in identifying and managing OD to prevent adverse outcomes such as aspiration pneumonia. Understanding the determinants of primary care HCPs undertaking this new behaviour provides the intervention targets for addressing. This realist review, underpinned by the Theoretical Domains Framework (TDF), aims to synthesise relevant literature and develop programme theories to understand what interventions work, how they work and under what circumstances to facilitate HCPs to prevent harm from OD. Combining realist methodology with behavioural science will permit conceptualisation of intervention components as theoretical behavioural constructs, thus informing the design of a future behaviour change intervention. Furthermore, through the TDF’s linkage to a taxonomy of behaviour change techniques, we will identify corresponding behaviour change techniques to include in this intervention. Methods & analysis: We are following the five steps for undertaking a realist review: 1) clarify the scope 2) Literature search 3) appraise and extract data 4) evidence synthesis 5) evaluation. We have searched Medline, Google scholar, PubMed, EMBASE, CINAHL, AMED, Scopus and PsycINFO databases. We are obtaining additional evidence through grey literature, snowball sampling, lateral searching and consulting the stakeholder group. Literature is being screened, evaluated and synthesised in Excel and Nvivo. We will appraise evidence in relation to its relevance and rigour. Data will be extracted and synthesised according to its relation to Initial programme theories (IPTs). IPTs were constructed after the preliminary literature search, informed by the TDF and with input from a stakeholder group of patient and public involvement advisors, general practitioners, speech and language therapists, geriatricians and pharmacists. We will follow the Realist and Meta-narrative Evidence Syntheses: Evolving Standards (RAMESES) quality and publication standards to report study results. Results: In this ongoing review our search has identified 1417 manuscripts with approximately 20% progressing to full text screening. We inductively generated 10 IPTs that hypothesise practitioners require: the knowledge to spot the signs and symptoms of OD; the skills to provide initial advice and support; and access to resources in their working environment to support them conducting these new behaviours. We mapped the 10 IPTs to 8 TDF domains and then generated a further 12 IPTs deductively using domain definitions to fulfil the remaining 6 TDF domains. Deductively generated IPTs broadened our thinking to consider domains such as ‘Emotion,’ ‘Optimism’ and ‘Social Influence’, e.g. If practitioners perceive that patients, carers and relatives expect initial advice and support, then they will be more likely to provide this, because they will feel obligated to do so. After prioritisation with stakeholders using a modified nominal group technique approach, a maximum of 10 IPTs will progress to test against the literature.

Keywords: behaviour change, deglutition disorders, primary healthcare, realist review

Procedia PDF Downloads 85

Authors: Sundara Subramanian Karuppasamy, Che Hua Yang

Abstract:

In this work, the laser ultrasound technique has been used for analyzing and imaging the inner defects in metal blocks. To detect the defects in blocks, traditionally the researchers used piezoelectric transducers for the generation and reception of ultrasonic signals. These transducers can be configured into the sparse and phased array. But these two configurations have their drawbacks including the requirement of many transducers, time-consuming calculations, limited bandwidth, and provide confined image resolution. Here, we focus on the non-contact method for generating and receiving the ultrasound to examine the inner defects in aluminum blocks. A Q-switched pulsed laser has been used for the generation and the reception is done by using Laser Doppler Vibrometer (LDV). Based on the Doppler effect, LDV provides a rapid and high spatial resolution way for sensing ultrasonic waves. From the LDV, a series of scanning points are selected which serves as the phased array elements. The side-drilled hole of 10 mm diameter with a depth of 25 mm has been introduced and the defect is interrogated by the linear array of scanning points obtained from the LDV. With the aid of the Synthetic Aperture Focusing Technique (SAFT) algorithm, based on the time-shifting principle the inspected images are generated from the A-scan data acquired from the 1-D linear phased array elements. Thus the defect can be precisely detected with good resolution.

Keywords: laser ultrasonics, linear phased array, nondestructive testing, synthetic aperture focusing technique, ultrasonic imaging

Procedia PDF Downloads 134

Authors: Mohhamed Yassine Baroudi, Abdelkrim Benammar, Fethi Tarik Bendimerad

Abstract:

This paper proposes software architecture for dynamical service adaptation. The services are constituted by reusable software components. The adaptation’s goal is to optimize the service function of their execution context. For a first step, the context will take into account just the user needs but other elements will be added. A particular feature in our proposition is the profiles that are used not only to describe the context’s elements but also the components itself. An adapter analyzes the compatibility between all these profiles and detects the points where the profiles are not compatibles. The same Adapter search and apply the possible adaptation solutions: component customization, insertion, extraction or replacement.

Keywords: adaptative service, software component, service, dynamic adaptation

Procedia PDF Downloads 298

Authors: U. Borasi, Amit Kr. Jain, Rakesh, Piyush Jain

Abstract:

Data collection, data preparation, model training, model evaluation, and deployment are all processes in a typical machine learning workflow. Training data needs to be gathered and organised. This often entails collecting a sizable dataset and cleaning it to remove or correct any inaccurate or missing information. Preparing the data for use in the machine learning model requires pre-processing it after it has been acquired. This often entails actions like scaling or normalising the data, handling outliers, selecting appropriate features, reducing dimensionality, etc. This pre-processed data is then used to train a model on some machine learning algorithm. After the model has been trained, it needs to be assessed by determining metrics like accuracy, precision, and recall, utilising a test dataset. Every time a new model is built, both data pre-processing and model training—two crucial processes in the Machine learning (ML) workflow—must be carried out. Thus, there are various Machine Learning algorithms that can be employed for every single approach to data pre-processing, generating a large set of combinations to choose from. Example: for every method to handle missing values (dropping records, replacing with mean, etc.), for every scaling technique, and for every combination of features selected, a different algorithm can be used. As a result, in order to get the optimum outcomes, these tasks are frequently repeated in different combinations. This paper suggests a simple architecture for organizing this largely produced “combination set of pre-processing steps and algorithms” into an automated workflow which simplifies the task of carrying out all possibilities.

Keywords: machine learning, automation, AUTOML, architecture, operator pool, configuration, scheduler

Procedia PDF Downloads 58

Authors: Farwa Rizvi, Joanne Williams, Humaira Maheen, Elizabeth Hoban

Abstract:

There is an increase in risky sexual behavior and unsafe sex in adolescent girls and young women aged 15 to 24 years in Cambodia, which negatively affects their reproductive health by increasing the risk of contracting sexually transmitted infections and unintended pregnancies. Risky sexual behavior includes ‘having sex at an early age, having multiple sexual partners, having sex while under the influence of alcohol or drugs, and unprotected sexual behaviors’. A systematic review of quantitative research conducted in Cambodia was undertaken, using the theoretical framework of the Social Ecological Model to identify the personal, social and cultural factors associated with risky sexual behavior and unsafe sex in young Cambodian women. PRISMA guidelines were used to search databases including Medline Complete, PsycINFO, CINAHL Complete, Academic Search Complete, Global Health, and Social Work Abstracts. Additional searches were conducted in Science Direct, Google Scholar and in the grey literature sources. A risk-of-bias tool developed explicitly for the systematic review of cross-sectional studies was used. Summary item on the overall risk of study bias after the inter-rater response showed that the risk-of-bias was high in two studies, moderate in one study and low in one study. The search strategy included a combination of subject terms and free text terms. The medical subject headings (MeSH) terms included were; contracept* or ‘birth control’ or ‘family planning’ or pregnan* or ‘safe sex’ or ‘protected intercourse’ or ‘unprotected intercourse’ or ‘protected sex’ or ‘unprotected sex’ or ‘risky sexual behaviour*’ or ‘abort*’ or ‘planned parenthood’ or ‘unplanned pregnancy’ AND ( barrier* or obstacle* or challenge* or knowledge or attitude* or factor* or determinant* or choic* or uptake or discontinu* or acceptance or satisfaction or ‘needs assessment’ or ‘non-use’ or ‘unmet need’ or ‘decision making’ ) AND Cambodia*. Initially, 300 studies were identified by using key words and finally, four quantitative studies were selected based on the inclusion criteria. The four studies were published between 2010 and 2016. The study participants ranged in age from 10-24 years, single or married, with 3 to 10 completed years of education. The mean age at sexual debut was reported to be 18 years. Using the perspective of the Social Ecological Model, risky sexual behavior was associated with individual-level factors including young age at sexual debut, low education, unsafe sex under the influence of alcohol and substance abuse, multiple sexual partners or transactional sex. Family level factors included living away from parents, orphan status and low levels of family support. Peer and partner level factors included peer delinquency and lack of condom use. Low socioeconomic status at the society level was also associated with risky sexual behaviour. There is scant research on sexual and reproductive health of adolescent girls and young women in Cambodia. Individual, family and social factors were significantly associated with risky sexual behaviour. More research is required to inform potential preventive strategies and policies that address young women’s sexual and reproductive health.

Keywords: adolescents, high-risk sex, sexual activity, unplanned pregnancies

Procedia PDF Downloads 246

Authors: Chongya Song, Kang Yen, Alexander Pons, Jin Liu

Abstract:

In the field of machine learning, the ensemble has been employed as a common methodology to improve the performance upon multiple base classifiers. However, the true predictions are often canceled out by the false ones during consensus due to a phenomenon called “curse of correlation” which is represented as the strong interferences among the predictions produced by the base classifiers. In addition, the existing practices are still not able to effectively mitigate the problem of imbalanced classification. Based on the analysis on our experiment results, we conclude that the two problems are caused by some inherent deficiencies in the approach of consensus. Therefore, we create an enhanced ensemble algorithm which adopts a designed rank-based chain-mode consensus to overcome the two problems. In order to evaluate the proposed ensemble algorithm, we employ a well-known benchmark data set NSL-KDD (the improved version of dataset KDDCup99 produced by University of New Brunswick) to make comparisons between the proposed and 8 common ensemble algorithms. Particularly, each compared ensemble classifier uses the same 22 base classifiers, so that the differences in terms of the improvements toward the accuracy and reliability upon the base classifiers can be truly revealed. As a result, the proposed rank-based chain-mode consensus is proved to be a more effective ensemble solution than the traditional consensus approach, which outperforms the 8 ensemble algorithms by 20% on almost all compared metrices which include accuracy, precision, recall, F1-score and area under receiver operating characteristic curve.

Keywords: consensus, curse of correlation, imbalance classification, rank-based chain-mode ensemble

Procedia PDF Downloads 138

Authors: Monika Balach, Iwona Frydrych, Agnieszka Cichocka

Abstract:

The objective of this research is to develop an algorithm, taking into account material type and body type that will describe the fabric properties and quality of fit of a garment to the body. One of the objectives of this research is to develop a new algorithm to simulate cloth draping within CAD/CAM software. Existing virtual fitting does not accurately simulate fabric draping behaviour. Part of the research into virtual fitting will focus on the mechanical properties of fabrics. Material behaviour depends on many factors including fibre, yarn, manufacturing process, fabric weight, textile finish, etc. For this study, several different fabric types with very different mechanical properties will be selected and evaluated for all of the above fabric characteristics. These fabrics include woven thick cotton fabric which is stiff and non-bending, woven with elastic content, which is elastic and bends on the body. Within the virtual simulation, the following mechanical properties can be specified: shear, bending, weight, thickness, and friction. To help calculate these properties, the KES system (Kawabata) can be used. This system was originally developed to calculate the mechanical properties of fabric. In this research, the author will focus on three properties: bending, shear, and roughness. This study will consider current research using the KES system to understand and simulate fabric folding on the virtual body. Testing will help to determine which material properties have the largest impact on the fit of the garment. By developing an algorithm which factors in body type, material type, and clothing function, it will be possible to determine how a specific type of clothing made from a particular type of material will fit on a specific body shape and size. A fit indicator will display areas of stress on the garment such as shoulders, chest waist, hips. From this data, CAD/CAM software can be used to develop garments that fit with a very high degree of accuracy. This research, therefore, aims to provide an innovative solution for garment fitting which will aid in the manufacture of clothing. This research will help the clothing industry by cutting the cost of the clothing manufacturing process and also reduce the cost spent on fitting. The manufacturing process can be made more efficient by virtual fitting of the garment before the real clothing sample is made. Fitting software could be integrated into clothing retailer websites allowing customers to enter their biometric data and determine how the particular garment and material type would fit their body.

Keywords: 3D scanning, fabric mechanical properties, quality of fit, virtual fitting

Procedia PDF Downloads 178

Authors: Auday Al-Mayyahi, Phil Birch, William Wang

Abstract:

A vision system-based navigation for autonomous ground vehicle (AGV) equipped with a single camera in an indoor environment is presented. A proposed navigation algorithm has been utilized to detect obstacles represented by coloured mini- cones placed in different positions inside a corridor. For the recognition of the relative position and orientation of the AGV to the coloured mini cones, the features of the corridor structure are extracted using a single camera vision system. The relative position, the offset distance and steering angle of the AGV from the coloured mini-cones are derived from the simple corridor geometry to obtain a mapped environment in real world coordinates. The corridor is first captured as an image using the single camera. Hence, image processing functions are then performed to identify the existence of the cones within the environment. Using a bounding box surrounding each cone allows to identify the locations of cones in a pixel coordinate system. Thus, by matching the mapped and pixel coordinates using a projection transformation matrix, the real offset distances between the camera and obstacles are obtained. Real time experiments in an indoor environment are carried out with a wheeled AGV in order to demonstrate the validity and the effectiveness of the proposed algorithm.

Keywords: autonomous ground vehicle, navigation, obstacle avoidance, vision system, single camera, image processing, ultrasonic sensor

Procedia PDF Downloads 302

Authors: Rafael do Amaral Teodoro, Leandro Augusto da Silva

Abstract:

Originated from graphite, graphene is a two-dimensional (2D) material that promises to revolutionize technology in many different areas, such as energy, telecommunications, civil construction, aviation, textile, and medicine. This is possible because its structure, formed by carbon bonds, provides desirable optical, thermal, and mechanical characteristics that are interesting to multiple areas of the market. Thus, several research and development centers are studying different manufacturing methods and material applications of graphene, which are often compromised by the scarcity of more agile and accurate methodologies to characterize the material – that is to determine its composition, shape, size, and the number of layers and crystals. To engage in this search, this study proposes a computational methodology that applies deep learning to identify graphene oxide crystals in order to characterize samples by crystal sizes. To achieve this, a fully convolutional neural network called U-net has been trained to segment SEM graphene oxide images. The segmentation generated by the U-net is fine-tuned with a standard deviation technique by classes, which allows crystals to be distinguished with different labels through an object delimitation algorithm. As a next step, the characteristics of the position, area, perimeter, and lateral measures of each detected crystal are extracted from the images. This information generates a database with the dimensions of the crystals that compose the samples. Finally, graphs are automatically created showing the frequency distributions by area size and perimeter of the crystals. This methodological process resulted in a high capacity of segmentation of graphene oxide crystals, presenting accuracy and F-score equal to 95% and 94%, respectively, over the test set. Such performance demonstrates a high generalization capacity of the method in crystal segmentation, since its performance considers significant changes in image extraction quality. The measurement of non-overlapping crystals presented an average error of 6% for the different measurement metrics, thus suggesting that the model provides a high-performance measurement for non-overlapping segmentations. For overlapping crystals, however, a limitation of the model was identified. To overcome this limitation, it is important to ensure that the samples to be analyzed are properly prepared. This will minimize crystal overlap in the SEM image acquisition and guarantee a lower error in the measurements without greater efforts for data handling. All in all, the method developed is a time optimizer with a high measurement value, considering that it is capable of measuring hundreds of graphene oxide crystals in seconds, saving weeks of manual work.

Keywords: characterization, graphene oxide, nanomaterials, U-net, deep learning

Procedia PDF Downloads 160

Authors: Nkwa Agbor Etobi Arreneke

Abstract:

A well-designed earthing/grounding system will not only provide an effective path for direct dissipation of faulty currents into the earth/soil, but also ensure the safety of personnels withing and around its immediate surrounding perimeter is free from the possibility of fatal electric shock. In order to achieve the latter, it is of paramount importance to ensuring that both the step and touch potentials are kept within the allowable tolerance set by standards IEEE Std-80-2000. In this article, the step and touch potentials of an earthing system are simulated and conformity verified using the Finite Element Method (FEM), and has been found to be 242.4V and 194.80V respectively. The effect of injection current position is also analyzed to observe its effect on a person within or in contact with any active part of the earthing system of the substation. The values obtained closely matches those of other published works which made using different numerical methods and/or simulations Genetic Algorithm (GA). This current study is aimed at throwing more light to the dangers of step and touch potential of earthing systems of substation and electrical facilities as a whole, and the need for further in-dept analysis of these parameters. Observations made on this current paper shows that, the position of contact with an energize earthing system is of paramount important in determining its effect on living organisms in contact with any energized part of the earthing systems.

Keywords: earthing/grounding systems, finite element method (fem), ground/earth resistance, safety, touch and step potentials, generic algorithm

Procedia PDF Downloads 100

Authors: Bao Li, Aike Qiao, Gaoyang Li, Youjun Liu

Abstract:

Background: Lumped parameter model (LPM) is a common numerical model for hemodynamic calculation. LPM uses circuit elements to simulate the human blood circulatory system. Physiological indicators and characteristics can be acquired through the model. However, due to the different physiological indicators of each individual, parameters in LPM should be personalized in order for convincing calculated results, which can reflect the individual physiological information. This study aimed to develop an automatic and effective optimization method to personalize the parameters in LPM of the blood circulatory system, which is of great significance to the numerical simulation of individual hemodynamics. Methods: A closed-loop LPM of the human blood circulatory system that is applicable for most persons were established based on the anatomical structures and physiological parameters. The patient-specific physiological data of 5 volunteers were non-invasively collected as personalized objectives of individual LPM. In this study, the blood pressure and flow rate of heart, brain, and limbs were the main concerns. The collected systolic blood pressure, diastolic blood pressure, cardiac output, and heart rate were set as objective data, and the waveforms of carotid artery flow and ankle pressure were set as objective waveforms. Aiming at the collected data and waveforms, sensitivity analysis of each parameter in LPM was conducted to determine the sensitive parameters that have an obvious influence on the objectives. Simulated annealing was adopted to iteratively optimize the sensitive parameters, and the objective function during optimization was the root mean square error between the collected waveforms and data and simulated waveforms and data. Each parameter in LPM was optimized 500 times. Results: In this study, the sensitive parameters in LPM were optimized according to the collected data of 5 individuals. Results show a slight error between collected and simulated data. The average relative root mean square error of all optimization objectives of 5 samples were 2.21%, 3.59%, 4.75%, 4.24%, and 3.56%, respectively. Conclusions: Slight error demonstrated good effects of optimization. The individual modeling algorithm developed in this study can effectively achieve the individualization of LPM for the blood circulatory system. LPM with individual parameters can output the individual physiological indicators after optimization, which are applicable for the numerical simulation of patient-specific hemodynamics.

Keywords: blood circulatory system, individual physiological indicators, lumped parameter model, optimization algorithm

Procedia PDF Downloads 137

Authors: Getenet Dessie, Jinhu Li, Son Nghiem, Tinh Doan

Abstract:

Background: Concerns regarding malnutrition have evolved from focusing solely on single forms to addressing the simultaneous occurrence of multiple types, commonly referred to as the double or triple burden of malnutrition. Nevertheless, data concerning the concurrent occurrence of various types of malnutrition are scarce. Therefore, this systematic review and meta-analysis aims to assess the pooled prevalence of the double burden of malnutrition among children under five in Sub-Saharan Africa and other least-developed countries (LDCs). Methods: Electronic, web-based searches were conducted from January 15 to June 28, 2023, across several databases, including PubMed, Embase, Google Scholar, and the World Health Organization's Hinari portal, as well as other search engines, to identify primary studies published up to June 28, 2023. Laboratory-based cross-sectional studies on children under the age of five were included. Two independent authors assessed the risk of bias and the quality of the identified articles. The primary outcomes of this study were micronutrient deficiencies and the comorbidity of stunting and anemia, as well as wasting and anemia. The random-effects model was utilized for analysis. The association of identified variables with the various forms of malnutrition was also assessed using adjusted odds ratios (AOR) with a 95% confidence interval (CI). This review was registered in PROSPERO with the reference number CRD42023409483. Findings: The electronic search generated 6,087 articles, 93 of which matched the inclusion criteria for the final meta-analysis. Micronutrient deficiencies were prevalent among children under five in Sub-Saharan Africa and other LDCs, with rates ranging from 16.63% among 25,169 participants for vitamin A deficiency to 50.90% among 3,936 participants for iodine deficiency. Iron deficiency anemia affected 20.56% of the 63,121 participants. The combined prevalence of wasting anemia and stunting anemia was 5.41% among 64,709 participants and 19.98% among 66,016 participants, respectively. Both stunting and vitamin A supplementation were associated with vitamin A and iron deficiencies, with adjusted odds ratios (AOR) of 1.54 (95% CI: 1.01, 2.37) and 1.37 (95% CI: 1.21, 1.55), respectively. Interpretation: The prevalence of the double burden of malnutrition among children under the age of five was notably high in Sub-Saharan Africa and other LDCs. These findings indicate a need for increased attention and a focus on understanding the factors influencing this double burden of malnutrition.

Keywords: children, Sub-Saharan Africa, least developed countries, double burden of malnutrition, systematic review, meta-analysis

Procedia PDF Downloads 81

Authors: Deeman Yousif Mahmood, Mohammed Abdullah Hussein

Abstract:

The goal of a network-based intrusion detection system is to classify activities of network traffics into two major categories: normal and attack (intrusive) activities. Nowadays, data mining and machine learning plays an important role in many sciences; including intrusion detection system (IDS) using both supervised and unsupervised techniques. However, one of the essential steps of data mining is feature selection that helps in improving the efficiency, performance and prediction rate of proposed approach. This paper applies unsupervised K-means clustering algorithm with information gain (IG) for feature selection and reduction to build a network intrusion detection system. For our experimental analysis, we have used the new NSL-KDD dataset, which is a modified dataset for KDDCup 1999 intrusion detection benchmark dataset. With a split of 60.0% for the training set and the remainder for the testing set, a 2 class classifications have been implemented (Normal, Attack). Weka framework which is a java based open source software consists of a collection of machine learning algorithms for data mining tasks has been used in the testing process. The experimental results show that the proposed approach is very accurate with low false positive rate and high true positive rate and it takes less learning time in comparison with using the full features of the dataset with the same algorithm.

Keywords: information gain (IG), intrusion detection system (IDS), k-means clustering, Weka

Procedia PDF Downloads 296

Authors: Mace Bentley, Zhuojun Duan, Tobias Gerken, Dudley Bonsal, Henry Way, Endre Szakal, Mia Pham, Hunter Donaldson, Chelsea Lang, Hayden Abbott, Leah Wilcynzski

Abstract:

Air pollution has the potential to modify the lifespan and intensity of thunderstorms and the properties of lightning. Using data mining and geovisualization, we investigate how background climate and weather conditions shape variability in urban air pollution and how this, in turn, shapes thunderstorms as measured by the intensity, distribution, and frequency of cloud-to-ground lightning. A spatiotemporal analysis was conducted in order to identify thunderstorms using high-resolution lightning detection network data. Over seven million lightning flashes were used to identify more than 196,000 thunderstorms that occurred between 2006 - 2020 in the Washington, DC Metropolitan Region. Each lightning flash in the dataset was grouped into thunderstorm events by means of a temporal and spatial clustering algorithm. Once the thunderstorm event database was constructed, hourly wind direction, wind speed, and atmospheric thermodynamic data were added to the initiation and dissipation times and locations for the 196,000 identified thunderstorms. Hourly aerosol and air quality data for the thunderstorm initiation times and locations were also incorporated into the dataset. Developing thunderstorm climatologies using a lightning tracking algorithm and lightning detection network data was found to be useful for visualizing the spatial and temporal distribution of urban augmented thunderstorms in the region.

Keywords: lightning, urbanization, thunderstorms, climatology

Procedia PDF Downloads 76

Authors: Xiaoqing Wang, Junfeng Wang, Xiaolan Zhu

Abstract:

Android operating system has been recognized by most application developers because of its good open-source and compatibility, which enriches the categories of applications greatly. However, it has become the target of malware attackers due to the lack of strict security supervision mechanisms, which leads to the rapid growth of malware, thus bringing serious safety hazards to users. Therefore, it is critical to detect Android malware effectively. Generally, the permissions declared in the AndroidManifest.xml can reflect the function and behavior of the application to a large extent. Since current Android system has not any restrictions to the number of permissions that an application can request, developers tend to apply more than actually needed permissions in order to ensure the successful running of the application, which results in the abuse of permissions. However, some traditional detection methods only consider the requested permissions and ignore whether it is actually used, which leads to incorrect identification of some malwares. Therefore, a machine learning detection method based on the actually used permissions combination and API calls was put forward in this paper. Meanwhile, several experiments are conducted to evaluate our methodology. The result shows that it can detect unknown malware effectively with higher true positive rate and accuracy while maintaining a low false positive rate. Consequently, the AdaboostM1 (J48) classification algorithm based on information gain feature selection algorithm has the best detection result, which can achieve an accuracy of 99.8%, a true positive rate of 99.6% and a lowest false positive rate of 0.

Keywords: android, API Calls, machine learning, permissions combination

Procedia PDF Downloads 329

Authors: C. B. Le, V. N. Pham

Abstract:

In modern data analysis, multi-source data appears more and more in real applications. Multi-source data clustering has emerged as a important issue in the data mining and machine learning community. Different data sources provide information about different data. Therefore, multi-source data linking is essential to improve clustering performance. However, in practice multi-source data is often heterogeneous, uncertain, and large. This issue is considered a major challenge from multi-source data. Ensemble is a versatile machine learning model in which learning techniques can work in parallel, with big data. Clustering ensemble has been shown to outperform any standard clustering algorithm in terms of accuracy and robustness. However, most of the traditional clustering ensemble approaches are based on single-objective function and single-source data. This paper proposes a new clustering ensemble method for multi-source data analysis. The fuzzy optimized multi-objective clustering ensemble method is called FOMOCE. Firstly, a clustering ensemble mathematical model based on the structure of multi-objective clustering function, multi-source data, and dark knowledge is introduced. Then, rules for extracting dark knowledge from the input data, clustering algorithms, and base clusterings are designed and applied. Finally, a clustering ensemble algorithm is proposed for multi-source data analysis. The experiments were performed on the standard sample data set. The experimental results demonstrate the superior performance of the FOMOCE method compared to the existing clustering ensemble methods and multi-source clustering methods.

Keywords: clustering ensemble, multi-source, multi-objective, fuzzy clustering

Procedia PDF Downloads 189

Authors: Ali Bhayani

Abstract:

College is a high involvement decision making where students are expected to evaluate several college offerings before selecting a college or a course to study. However, even in high involvement product like college, students get influenced by opinion leaders and suffer from social contagion. This narrative style study, involving 98 first year students, was able to demonstrate that social contagion differs with regards to gender, ethnicity and personality. Recommendations from students with academically strong background would impact on the college choice of the undergraduate students and limit information search. Study was able to identify the incidence of anchoring heuristics amongst the students. Managerial implications with regards to design of marketing campaign follows at the end of the study.

Keywords: social contagion, opinion leaders, higher education, consumer behavior

Procedia PDF Downloads 365

Authors: Sardorbek Numonov, Hyohun Kim, Dongwha Shin, Yeonseok Kim, Ji-Su Ahn, Dongeun Choi, Byung-Woo Hong

Abstract:

The detection of curvilinear structures often plays an important role in the analysis of images. In particular, it is considered as a crucial step for the diagnosis of chronic respiratory diseases to localize the fissures in chest CT imagery where the lung is divided into five lobes by the fissures that are characterized by linear features in appearance. However, the characteristic linear features for the fissures are often shown to be subtle due to the high intensity variability, pathological deformation or image noise involved in the imaging procedure, which leads to the uncertainty in the quantification of anatomical or functional properties of the lung. Thus, it is desired to enhance the linear features present in the chest CT images so that the distinctiveness in the delineation of the lobe is improved. We propose a recursive diffusion process that prefers coherent features based on the analysis of structure tensor in an anisotropic manner. The local image features associated with certain scales and directions can be characterized by the eigenanalysis of the structure tensor that is often regularized via isotropic diffusion filters. However, the isotropic diffusion filters involved in the computation of the structure tensor generally blur geometrically significant structure of the features leading to the degradation of the characteristic power in the feature space. Thus, it is required to take into consideration of local structure of the feature in scale and direction when computing the structure tensor. We apply an anisotropic diffusion in consideration of scale and direction of the features in the computation of the structure tensor that subsequently provides the geometrical structure of the features by its eigenanalysis that determines the shape of the anisotropic diffusion kernel. The recursive application of the anisotropic diffusion with the kernel the shape of which is derived from the structure tensor leading to the anisotropic scale-space where the geometrical features are preserved via the eigenanalysis of the structure tensor computed from the diffused image. The recursive interaction between the anisotropic diffusion based on the geometry-driven kernels and the computation of the structure tensor that determines the shape of the diffusion kernels yields a scale-space where geometrical properties of the image structure are effectively characterized. We apply our recursive anisotropic diffusion algorithm to the detection of curvilinear structure in the chest CT imagery where the fissures present curvilinear features and define the boundary of lobes. It is shown that our algorithm yields precise detection of the fissures while overcoming the subtlety in defining the characteristic linear features. The quantitative evaluation demonstrates the robustness and effectiveness of the proposed algorithm for the detection of fissures in the chest CT in terms of the false positive and the true positive measures. The receiver operating characteristic curves indicate the potential of our algorithm as a segmentation tool in the clinical environment. This work was supported by the MISP(Ministry of Science and ICT), Korea, under the National Program for Excellence in SW (20170001000011001) supervised by the IITP(Institute for Information and Communications Technology Promotion).

Keywords: anisotropic diffusion, chest CT imagery, chronic respiratory disease, curvilinear structure, fissure detection, structure tensor

Procedia PDF Downloads 233

Authors: Pauline E. Ngo-Henha

Abstract:

Existing turnover intention theories are reviewed in this paper. This review was conducted with the help of the search keyword “turnover intention theories” in Google Scholar during the month of July 2017. These theories include: The Theory of Organizational Equilibrium (TOE), Social Exchange Theory, Job Embeddedness Theory, Herzberg’s Two-Factor Theory, the Resource-Based View, Equity Theory, Human Capital Theory, and the Expectancy Theory. One of the limitations of this review paper is that data were only collected from Google Scholar where many papers were sometimes not freely accessible. However, this paper attempts to contribute to the research in clarifying the distinction between theories and models in the context of turnover intention.

Keywords: Literature Review, Theory, Turnover, Turnover intention

Procedia PDF Downloads 457

Authors: Jack R. McKenzie, Peter A. Appleby, Thomas House, Neil Walton

Abstract:

Cold-start is a notoriously difficult problem which can occur in recommendation systems, and arises when there is insufficient information to draw inferences for users or items. To address this challenge, a contextual bandit algorithm – the Fast Approximate Bayesian Contextual Cold Start Learning algorithm (FAB-COST) – is proposed, which is designed to provide improved accuracy compared to the traditionally used Laplace approximation in the logistic contextual bandit, while controlling both algorithmic complexity and computational cost. To this end, FAB-COST uses a combination of two moment projection variational methods: Expectation Propagation (EP), which performs well at the cold start, but becomes slow as the amount of data increases; and Assumed Density Filtering (ADF), which has slower growth of computational cost with data size but requires more data to obtain an acceptable level of accuracy. By switching from EP to ADF when the dataset becomes large, it is able to exploit their complementary strengths. The empirical justification for FAB-COST is presented, and systematically compared to other approaches on simulated data. In a benchmark against the Laplace approximation on real data consisting of over 670, 000 impressions from autotrader.co.uk, FAB-COST demonstrates at one point increase of over 16% in user clicks. On the basis of these results, it is argued that FAB-COST is likely to be an attractive approach to cold-start recommendation systems in a variety of contexts.

Keywords: cold-start learning, expectation propagation, multi-armed bandits, Thompson Sampling, variational inference

Procedia PDF Downloads 108