Fatimah Ahmad

Rough neural expert systems

The knowledge acquisition process is a crucial stage in the technology of expert systems. However, this process is not well defined. One of the promising structured sources of learning can be found in the recent work on neural network technology. A neural network can serve as a knowledge base of expert systems that does classification tasks. Another way of learning is by using the rough set as a new mathematical tool to deal with uncertain and imprecise data. Two methods based on rough set analysis were developed and merged with the integration of neural networks and expert systems, forming a new hybrid architecture of expert systems called a rough neural expert system. The first method works as a pre-processor for neural networks within the architecture, and it is called a pre-processing rough engine, while the second one was added to the architecture for building a new structure of inference engine called a rough neural inference engine. Consequently, a new architecture of knowledge base was designed. This new architecture was based on the connectionist of neural networks and the reduction of rough set analysis. The performance of the proposed system was evaluated by an application to the field of medical diagnosis using a real example of hepatitis diseases. The results indicate that the new methods have improved the inference procedures of the expert systems, and have showed that this new architecture has some properties over the conventional architectures of expert systems.

Parallel solution of high speed low order FDTD on 2D free space wave propagation

Finite Difference Time-Domain (FDTD) is one of the most widely used numerical method for solving electromagnetic problems. Solutions for these problems are computationally expensive in terms of processing time. Recently, we develop a method, High Speed Low Order FDTD (HSLO-FDTD) that is proven to solve one dimensional electro-magnetic problem with a reduction of 67% of processing time from the FDTD method. In this paper, we extend the method to solve two dimensional wave propagation problem. Since the problem is large, we develop the parallel version of HSLO-FDTD method on distributed memory multiprocessor machine using message-passing interface. We examine the parallelism efficiency of the algorithm by analyzing the execution time and speed-up.

Application of latent semantic indexing on Malay-English cross language information retrieval

This paper concerns an application of latent semantic indexing on Malay and English cross language information retrieval system. The retrieval effectiveness was tested on the actual Quranic collection using latent semantic indexing model. The results show that average precision on cross language is higher than monolingual retrieval.

Experiments in malay information retrieval

There have been very few studies on the use of conflation algorithms for indexing and retrieval of Malay documents. The two main classes of conflation algorithms are string-similarity algorithms and stemming algorithms. Stemming is used in information retrieval systems to reduce variant word forms to common roots in order to improve retrieval effectiveness. As in other languages, there is a need for an effective stemming algorithm for the indexing and retrieval of Malay documents. Again there are few research on n-gram string-similarity measures done on Malay. We have experimented on the application of stemming and string similarity matching on retrieving of verses from the Al-Quran in order to evaluated their effectiveness. Before retrieval effectiveness can be carried out an experimental data set need to be developed which comprises of a collection of documents, a set of queries, and their relevant judgements. In this paper we will describe the development of the experimental data set and the application of stemming and similarity matching algorithms in retrieving the verses from the Al-Quran. Inherent characteristics of n-grams and several variations of experiments performed on the queries and documents are discussed. The variations are: both non-stemmed queries and documents; stemmed queries and nonstemmed documents; and both stemmed queries and documents. Further experiment are then carried out by removing the most frequently occurring n-gram. The dice-coefficient is used as threshold and weight in ranking the retrieved documents. Beside using dice coefficients to rank documents, inverse document frequency weights are also used. Interpolation technique and standard recall-precision functions are used to evaluate the retrieval effectiveness.

Active warden as the main hindrance for steganography information retrieval

Our paper discusses how active warden operates and why it is important for steganographers to understand the impending threat in which they possess. It was a common belief that the main adversary for steganography is coming from steganalysis detection. However, we have found in some situation, the destruction of hidden information is more easily achievable compared to the task of detecting it. Active Wardens are attackers of steganography which aims to demolish possible hidden information within a carrier media. If the enemys objective is to disrupt the communication of hidden information, then the active approach is definitely a better choice compared to passive time consuming steganalysis.

Multi-resolution Joint Auto Correlograms: Determining the distance function

Distance function plays a role in content-based image retrieval where the ideal distance function will be able to close the gap between computerised image interpretation and similarity judgment by humans. In this paper, few distance functions in relation to the advancement of Colour Auto Correlogram are studied and compared in order to determine the most suitable distance function for the proposed Multi-resolution Joint Auto Correlograms descriptor. An experiment has been conducted on the SIMPLIcity image database consisting of 1000 images where the precision, recall, and rank of various distance functions are measured. Retrieval results have shown that the L1-norm has achieved higher precision rate of 78.52% and has able to rank similar images better (a rank of 199) compared to the Generalised Tversky Index distance function.

Invariant Generalised Ridgelet-Fourier for shape-based image retrieval

A new shape descriptor called the Invariant Generalised Ridgelet-Fourier is defined for the application of Content-based Image Retrieval (CBIR). The proposed spectral-based method is invariant to rotation, scaling, and translation (RST) as well as able to handle images of arbitrary size. The implementation of Ridgelet transform on the ellipse containing the shape and the normalisation of the Radon transform is introduced. The 1D Wavelet transform is then applied to the Radon slices. In order to extract the rotation invariant feature, Fourier transform is implemented in the Ridgelet domain. The performance of the proposed method is accessed on a standard MPEG-7 CE-1 B dataset in terms of few objective evaluation criteria. From the experiments, it is shown that the proposed method provides promising results compared to several previous methods.

Construction of Computational Lexicon for Malay Language

This paper focuses on construction of computational lexicon for Malay language that involves computational study and the use of electronic lexicons. To construct the lexicons, it includes a study on morphological arrangement of Malay affixation process which comprises of prefixes, suffixes, circumfixes and infixes with the intention of constructing a collection of new Malay lexicons or words that will be automatically constructed from a single root word. This research conducts experiments on 2101 unique Malay root words found in the Malay translated Quranic documents that are later experimented with Malay affixation rules using the affixed words analyser. Numerous new words are constructed from a single root word by adding 52 affix rules to the root word. Finally, each new word is compared with Malay dictionary to ensure whether it is truly a new generated Malay word. Results from this analysis open opportunity to construct new Malay word variant to enrich the Malay lexicon.

Multi-resolution Shape-Based Image Retrieval Using Ridgelet Transform

Complicated shapes can be effectively characterized using multi-resolution descriptors. One popular method is the Ridgelet transform which has enjoyed very little exposure in describing shapes for Content-based Image Retrieval (CBIR). Many of the existing Ridgelet transforms are only applied on images of size M×M. For M×N sized images, they need to be segmented into M×M sub-images prior to processing. A different number of orientations and cut-off points for the Radon transform parameters also need to be utilized according to the image size. This paper presents a new shape descriptor for CBIR based on Ridgelet transform which is able to handle images of various sizes. The utilization of the ellipse template for better image coverage and the normalization of the Ridgelet transform are introduced. For better retrieval, a template-option scheme is also introduced. Retrieval effectiveness obtained by the proposed method has shown to be higher compared to several previous descriptors.

Generalized Ridgelet-Fourier for M×N images: Determining the normalization criteria

Ridgelet transform (RT) has gained its popularity due to its capability in dealing with line singularities effectively. Many of the existing RT however is only applied to images of size M×M or the M×N images will need to be pre-segmented into M×M sub-images prior to processing. The research presented in this article is aimed at the development of a generalized RT for content-based image retrieval so that it can be applied easily to any images of various sizes. This article focuses on comparing and determining the normalization criteria for Radon transform, which will aid in achieving the aim. The Radon transform normalization criteria sets are compared and evaluated on an image database consisting of 216 images, where the precision and recall and Averaged Normalized Modified Retrieval Rank (ANMRR) are measured.