Please use this identifier to cite or link to this item: https://ptsldigital.ukm.my/jspui/handle/123456789/476402
Full metadata record
DC FieldValueLanguage
dc.contributor.advisorSalwani Abdullah, Prof. Dr.
dc.contributor.authorZohreh Golbagtaklimi (P53659)
dc.date.accessioned2023-10-06T09:17:50Z-
dc.date.available2023-10-06T09:17:50Z-
dc.date.issued2012-06-25
dc.identifier.otherukmvital:85031
dc.identifier.urihttps://ptsldigital.ukm.my/jspui/handle/123456789/476402-
dc.descriptionDue to the wide availability of vast amount of data and vital need for turning these data into valuable information and knowledge, data mining techniques has attracted a great attention in recent years. Feature selection is considered as one of these techniques which choose a subset of input variables from large amount of data by eliminating features with little or no predictive information. According to the complexity and high dimension of real life data, feature selection is a time consuming process and assumed as a NP-hard problem. Microarray data is an example of such a data with high dimension. Biologists not just interested on accurate classification tools, but they also need to identify biomarkers of diseases, which are achieved by reduction of dimensionality, so that the relationship between the symptoms and their corresponding diagnosis can be inferred. Selection of useful genes reduces the large number of irrelevant and redundant genes while increasing the classification accuracy and decreasing the computational time. In this thesis, a hybrid Greedy Randomized Adaptive Search Procedure (GRASP) and Tabu Search (TS) has been applied to gene selection on microarray data with an aim to minimise the subset gene cardinality and maximise the classification accuracy. GRASP is a multi-start constructive method which constructs a solution in the first phase and improve that solution by applying a local search in second phase. However, using the seminal techniques in both phases of GRASP makes it prohibitive to use in high-dimensional datasets. A hybrid GRASP that uses a Tabu search (i.e. a metaheuristic with an adaptive memory) in the second phase allows the exploration around the constructed solution while looking for an improvement. It guides a local search procedure to explore the search space beyond local optimality. In this research, in order to evaluate the selected gene subset, the Naive Bayes classifier is chosen. The proposed approach is tested on five standard microarray datasets. The experimental results illustrate that the proposed method is able to obtain competitive results compared to other previous methods in the literature in terms of minimal cardinality and maximum accuracy.,Master / Sarjana
dc.language.isoeng
dc.publisherUKM, Bangi
dc.relationFaculty of Information Science and Technology / Fakulti Teknologi dan Sains Maklumat
dc.rightsUKM
dc.subjectDNA microarrays
dc.titleHybridisation of Greedy Randomized Adaptive Search Procedure with Tabu Search for gene selection on microarray data
dc.typetheses
dc.format.pages149
dc.identifier.callnoQ335.7.G585 2012 3
dc.identifier.barcode002068
Appears in Collections:Faculty of Information Science and Technology / Fakulti Teknologi dan Sains Maklumat

Files in This Item:
File Description SizeFormat 
ukmvital_85031+SOURCE1+SOURCE1.0.PDF
  Restricted Access
3.95 MBAdobe PDFThumbnail
View/Open


Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.