Academia of Intelligence Computing

Selection And Identification Of A Prioritisation Factor For Test Case Optimisation

Nurul Shazani Sahidan — 2021-03-12

4FTC is a method to select and identify factor values that will be used to optimise fault detection by creating test case ordering to be executed in the testing phase. Many researchers, as mentioned in the literature review, have come out with techniques to prioritise and minimise the effort, time, and cost of software testing, namely test case prioritisation methods, regression selection techniques, and a test case reduction model. In this research, a new prioritisation technique for requirements based on system-level test cases to improve the rate of fault detection was proposed. The study outcomes indicate that the 4FTC model led to an improvement in the detection rate of faults in comparison to the random ordering of test cases.

A Comparison Between Stacked Denoising Autoencoder (SDAE) with Long Short-Term Memory (LSTM) for Cancer Classification

Nuraina Syaza Azman — 2020-12-18

Cancer is the serious type of disease as genetic modification can lead to cancer. Cancer disease can be controlled from spread if early diagnosed is make. The main problem to do cancer classification is because of the large data in dataset. This data will be meaningless if the dataset contains many unclean data. Unclean data meant that the data might be null data, redundant or else. Therefore, the result produce will be not much accurate, and the complexity of the algorithm can be increased. In this research, multi-omics dataset will be used. To get multi-omics dataset, the data need to undergo some pre-process phase to make sure that the dataset produce is clean. The dataset will be integrating by using python coding. This will help to combine the omics data from different file. The data also will undergo feature selection to select attribute or features from the complete set of data. The method that will be used for feature selection is SVM-RFE. SVM-RFE was proven as the best feature selection method to ranks the features or gene by training support vector machine classification model and choose lead genes feature with RFE strategy [1]. After the data have clean it will be implemented in stacked denoising autoencoder (SDAE) and recurrent neural network (RNN). Lastly, the performance of both methods will be compared to find out which method is the better in classifying cancer data using multi-omics dataset.

Named Entity Recognition in Biomedical Documents using Recurrent Neural Network

Diong Lee Juen — 2020-12-18

Named entity recognition is an information extraction task that detect and classify named entity mentions in free text into predefined categories. It will help to solve more complex text mining tasks such as information retrieval, question answering and text summarization. However, current research has been done on this task using available dataset or dataset annotated by medical experts. It is difficult to obtain the manually annotated dataset by medical experts especially for a new dataset. Application of available biomedical resources is very important to overcome this problem. Thus, this paper will specify in named entity recognition on PubMed dataset specifically for hypertension disease which is annotated through application of biomedical resources. GENIA Tagger is used to perform tokenization of the biomedical abstracts and MetaMapLite is used to perform semantic annotation. After that, the terms or phrases are annotated into the BIO format. Bidirectional LSTM-CRF, an example of Recurrent Neural Network that showed promising results for named entity recognition, will be applied to perform named entity recognition for the research dataset. The experiment setting of 500 abstracts, 32 batch sizes and 25 epochs presented the best results for precision, recall and F1-score which are 0.79, 0.77 and 0.78, respectively. The results showed that the research dataset achieved almost the same results for precision, recall and F1-score as shown in the previous study using dataset which was manually annotated by medical experts.

Test Predictive Model: Under/Over Prediction Of Test Case Size Estimation

Nurul Shazani Sahidan — 2021-03-12

The process of developing an estimation of adequate test set size utilising the test predictive model is presented and evaluated in this paper. The strategy to develop the model was previously presented in our different paper, whereas the details, design, and implementation idea for the model are described in this paper. The aims of the prediction model is to solve the problems related with the size of the test cases. For better fault detection in the retesting process, a good prediction of the size of test cases is required. Too few or too large a consumption of test cases may increase the budget and time in the tesing process. This model could potentially test the set constuction as an important factor that can provide an accuracy estimation of potential faults that will occur in the system based on a reliable guess (of a number) of the (good) adequacy of the test case in relation to the detected error. The essential elements for developing the model are explained in detail in the subsequent section. The evaluation conducted in this chapter includes a few steps, which are then applied in one real case study to reveal the effectiveness of the proposed prediction design model. The test predictive model is then evaluated for over-prediction or under-prediction compared to manual estimation by experts and then further evaluated using a seed fault analysis.

Multi-omics Cancer Classification Using Stacked Denoising Autoencoder (SDAE) and Deep Belief Network (DBN)

Nur Ainul Afieqah Nuzula — 2020-12-18

The multi-omics datasets from liver cancer are specifically being analysed and integrated which then will be classified according to cancerous and non-cancerous type. However, the dataset may have outlier, missing data and redundancy that may impact the accuracy of the result. Thus, pre-processing data is an important step that will remove the entire problem. Feature selection will help to select the full requirements data to that the overall research has great quality and give benefits to the community and future studies. Deep neural network algorithms like Stacked Denoising Autoencoder (SDAE) and Deep Belief Network (DBN) is reliable to perform classification for omics data. This research is mainly to find the best algorithm that will produce better result for this classification of multi-omics of liver cancer data