Azizah Suliman

Parallel execution of distributed SVM using MPI (CoDLib)

Support Vector Machine (SVM) is an efficient data mining approach for data classification. However, SVM algorithm requires very large memory requirement and computational time to deal with very large dataset. To reduce the computational time during the process of training the SVM, a combination of distributed and parallel computing method, CoDLib have been proposed. Instead of using a single machine for parallel computing, multiple machines in a cluster are used. Message Passing Interface (MPI) is used in the communication between machines in the cluster. The original dataset is split and distributed to the respective machines. Experiments results shows a great speed up on the training of the MNIST dataset where training time has been significantly reduced compared with standard LIBSVM without affecting the quality of the SVM.

Hybrid of HMM and Fuzzy Logic for handwritten character recognition

This paper presents a hybrid approach of HMM and Fuzzy Logic in the field of handwritten character recognition. Fuzzy Logic is used in the recognition phase while HMM is used in the process of extracting features for the preparation of linguistic variables of the fuzzy rules. Experimental results from a few sample images give a reasonable recognition rate on a more challenging database of lower-case handwritten characters. This proved the proposed hybrid of the two techniques are compatible and can be used to complement each other effectively.

Classification red blood cells using support vector machine

The shape of red blood cells (RBCs) contributes to clinical diagnoses of blood diseases. The field of medical imaging has become more important because of the increasing need for automated and efficient diagnoses within a short period of time. Imaging technique plays an important role in RBC research for hematology. Classification is an important component of the retrieval system which allows one to distinguish between normal RBCs and abnormal RBCs which indicate anemia. In this paper, image processing techniques that use the optimization segmentation and mean filter play an important role in obtaining the geometric, texture and color features related to RBC images by using a photo imaging microscope. The support vector machine, which is an advanced kernel-based technique, is used to classify RBC data as either normal or abnormal, the proposed classifier algorithm achieved very good accuracy rates with validation measure of sensitivity, specificity and Kappa to be 100%, 0.998% and 0.9944 respectively.

A new hybrid model of software engineering and systems engineering for embedded system development methodology

Despite it being an established area of research and development, embedded system does not own a dedicated development methodology that developers can adhere to. Being an inter-disciplinary activity, a cross of software and hardware, it brings challenge to developer to construct a development methodology for embedded system. This paper proposes a new development model for embedded system by the hybrid of selected development methodologies in software engineering and systems engineering, considering they are both essential in embedded system. The model is harmonized with embedded system design vital tasks and also non-functional properties following the ISO/IEC9126.

Extracting Features for the Linguistic Variables of Fuzzy Rules Using Hidden Markov Model

In classifying handwritten characters, the stages prior to the classification phase play a role as major as the classification itself. This research work will be classifying the characters using a syntactical classification method namely fuzzy logic but will use the statistical method of Hidden Markov Model as an approach in extracting features for the linguistic variables of the fuzzy rule‐based system. In this paper the feature extraction method will be highlighted and detailed. The HMM Model of a variable to be used in the classification system will be discussed. Experimental results from a few sample images show that the proposed technique is both effective and efficient to be used in extracting features for the linguistic variables of fuzzy rules.

Measuring GPU-accelerated parallel SVM performance using large datasets for multi-class machine learning problem

This paper presents performance evaluation of GPU-accelerated Support Vector Machines (SVMs) using large datasets. Although SVMs algorithm is popular among machine learning researchers and data mining practitioners, its computational time is too long and impractical for large datasets due to its complex Quadratic Programming (QP) solver. The result shows that using GPU-accelerated SVMs can significantly reduce computational time for training phase of SVMs and it can be a viable solution for any project that require real-time forecasting output.