• Journal of Information Processing Systems
• /
• 제8권2호
• /
• pp.241-262
• /
• 2012

An understanding of quality attributes is relevant for the software organization to deliver high software reliability. An empirical assessment of metrics to predict the quality attributes is essential in order to gain insight about the quality of software in the early phases of software development and to ensure corrective actions. In this paper, we predict a model to estimate fault proneness using Object Oriented CK metrics and QMOOD metrics. We apply one statistical method and six machine learning methods to predict the models. The proposed models are validated using dataset collected from Open Source software. The results are analyzed using Area Under the Curve (AUC) obtained from Receiver Operating Characteristics (ROC) analysis. The results show that the model predicted using the random forest and bagging methods outperformed all the other models. Hence, based on these results it is reasonable to claim that quality models have a significant relevance with Object Oriented metrics and that machine learning methods have a comparable performance with statistical methods.

• East Asian mathematical journal
• /
• 제16권2호
• /
• pp.251-266
• /
• 2000

The problem of solution uniqueness to the task of determining unknown parameters of mathematical models from input-output observations is studied. This problem is known as structural identifiability problem. We offer a new approach for testing structural identifiability of linear state space models. The approach compares favorably with numerous methods proposed by other authors for two main reasons. First, it is formulated in obvious mathematical form. Secondly, the method does not involve unfeasible symbolic computations and thus allows to test identifiability of large-scale models. In case of non-identifiability, when there is a set of solutions to the task, we offer a method of computing functions of the unknown parameters which can be determined uniquely from input-output observations and later used as new parameters of the model. Such functions are called parametric functions capable of estimation. To develop the method of computation of these functions we use Lie group transformation theory. Illustrative example is given to demonstrate applicability of presented methods.

• 품질경영학회지
• /
• 제37권4호
• /
• pp.52-60
• /
• 2009

The North and South Koreas have different standards for computer keyboard designs, though both are two sets type keyboards. We expect that the unified design will be three sets type as the third standard. Korean keyboards of three sets type are classified as 3 modes in this paper, such as limited models, medium models, and extended models. Based on design principles of three sets type, 27, 29, 30, 36, and 38 keys models for a unified standard are suggested.

• 로봇학회논문지
• /
• 제14권2호
• /
• pp.114-121
• /
• 2019

This paper presents a new benchmark system for visual odometry (VO) and monocular depth estimation (MDE). As deep learning has become a key technology in computer vision, many researchers are trying to apply deep learning to VO and MDE. Just a couple of years ago, they were independently studied in a supervised way, but now they are coupled and trained together in an unsupervised way. However, before designing fancy models and losses, we have to customize datasets to use them for training and testing. After training, the model has to be compared with the existing models, which is also a huge burden. The benchmark provides input dataset ready-to-use for VO and MDE research in 'tfrecords' format and output dataset that includes model checkpoints and inference results of the existing models. It also provides various tools for data formatting, training, and evaluation. In the experiments, the exsiting models were evaluated to verify their performances presented in the corresponding papers and we found that the evaluation result is inferior to the presented performances.

• Journal of Information Processing Systems
• /
• 제18권2호
• /
• pp.268-281
• /
• 2022

The development of information technology is bringing many changes to everyday life, and machine learning can be used as a technique to solve a wide range of real-world problems. Analysis and utilization of data are essential processes in applying machine learning to real-world problems. As a method of processing data in machine learning, we propose an approach based on applying multiple linear regression models by interlacing data to the task of classifying similar software. Linear regression is widely used in estimation problems to model the relationship between input and output data. In our approach, multiple linear regression models are generated by training on interlaced feature data. A combination of these multiple models is then used as the prediction model for classifying similar software. Experiments are performed to evaluate the proposed approach as compared to conventional linear regression, and the experimental results show that the proposed method classifies similar software more accurately than the conventional model. We anticipate the proposed approach to be applied to various kinds of classification problems to improve the accuracy of conventional linear regression.

• Journal of information and communication convergence engineering
• /
• 제21권1호
• /
• pp.17-23
• /
• 2023

Recently, several high-performance instance segmentation models have used the Mask R-CNN model as a baseline, which reached a historical peak in instance segmentation in 2017. There are numerous derived models using the Mask R-CNN model, and if the performance of Mask R-CNN is improved, the performance of the derived models is also anticipated to improve. The Mask R-CNN uses interpolation to adjust the image size, and the input differs depending on the interpolation method. Therefore, in this study, the performance change of Mask R-CNN was compared when various interpolation methods were applied to the transform layer to improve the performance of Mask R-CNN. To train and evaluate the models, this study utilized the PennFudan and Balloon datasets and the AP metric was used to evaluate model performance. As a result of the experiment, the derived Mask R-CNN model showed the best performance when bicubic interpolation was used in the transform layer.

• ETRI Journal
• /
• 제46권1호
• /
• pp.137-153
• /
• 2024

This study introduces a new approach in Korean morphological analysis combining dictionary-based techniques with Transformer-based deep learning models. The key innovation is the use of a BERT-based reranking system, significantly enhancing the accuracy of traditional morphological analysis. The method generates multiple suboptimal paths, then employs BERT models for reranking, leveraging their advanced language comprehension. Results show remarkable performance improvements, with the first-stage reranking achieving over 20% improvement in error reduction rate compared with existing models. The second stage, using another BERT variant, further increases this improvement to over 30%. This indicates a significant leap in accuracy, validating the effectiveness of merging dictionary-based analysis with contemporary deep learning. The study suggests future exploration in refined integrations of dictionary and deep learning methods as well as using probabilistic models for enhanced morphological analysis. This hybrid approach sets a new benchmark in the field and offers insights for similar challenges in language processing applications.

• Journal of Radiation Protection and Research
• /
• 제41권4호
• /
• pp.389-394
• /
• 2016

Background: Tetrahedral-mesh geometries can be used in the MCNP code, but the MCNP code accepts only the geometry in the Abaqus input file format; hence, the existing tetrahedral-mesh models first need to be converted to the Abacus input file format to be used in the MCNP code. In the present study, we developed a simple but useful computer program, TET2MCNP, for converting TetGen-generated tetrahedral-mesh models to the Abacus input file format. Materials and Methods: TET2MCNP is written in C++ and contains two components: one for converting a TetGen output file to the Abacus input file and the other for the reverse conversion process. The TET2MCP program also produces an MCNP input file. Further, the program provides some MCNP-specific functions: the maximum number of elements (i.e., tetrahedrons) per part can be limited, and the material density of each element can be transferred to the MCNP input file. Results and Discussion: To test the developed program, two tetrahedral-mesh models were generated using TetGen and converted to the Abaqus input file format using TET2MCNP. Subsequently, the converted files were used in the MCNP code to calculate the object- and organ-averaged absorbed dose in the sphere and phantom, respectively. The results show that the converted models provide, within statistical uncertainties, identical dose values to those obtained using the PHITS code, which uses the original tetrahedral-mesh models produced by the TetGen program. The results show that the developed program can successfully convert TetGen tetrahedral-mesh models to Abacus input files. Conclusion: In the present study, we have developed a computer program, TET2MCNP, which can be used to convert TetGen-generated tetrahedral-mesh models to the Abaqus input file format for use in the MCNP code. We believe this program will be used by many MCNP users for implementing complex tetrahedral-mesh models, including computational human phantoms, in the MCNP code.

• Journal of Electrical Engineering and information Science
• /
• 제1권2호
• /
• pp.15-26
• /
• 1996

MPAs(Massively Parallel Architectures) should address two fundamental issues for scalability: synchronization and communication latency. Dataflow architecture faces problems of excessive synchronization overhead and inefficient execution of sequential programs while they offer the ability to exploit massive parallelism inherent in programs. In contrast, MPAs based on von Neumann computational model may suffer from inefficient synchronization mechanism and communication latency. DAVRID (DAtaflow/Von Neumann RISC hybrID) is a massively parallel multithreaded architecture which takes advantages of von Neumann and dataflow models. It has good single thread performance as well as tolerates synchronization and communication latency. In this paper, we describe the DAVRID architecture in detail and evaluate its performance through simulation runs over several benchmarks.

• Computers and Concrete
• /
• 제18권1호
• /
• pp.1-16
• /
• 2016

In this study, an Artificial Neural Network (ANN) and Adaptive Network-based Fuzzy Inference Systems (ANFIS) prediction models for flexural strength of the cement mortars have been developed. For purpose of constructing this models, 12 different mixes with 144 specimens of the 2, 7, 28 and 90 days flexural strength experimental results of cement mortars containing pure Portland cement (PC), blast furnace slag (BFS), waste tire rubber powder (WTRP) and BFS+WTRP used in training and testing for ANN and ANFIS were gathered from the standard cement tests. The data used in the ANN and ANFIS models are arranged in a format of four input parameters that cover the Portland cement, BFS, WTRP and age of samples and an output parameter which is flexural strength of cement mortars. The ANN and ANFIS models have produced notable excellent outputs with higher coefficients of determination of $R^2$, RMS and MAPE. For the testing of dataset, the $R^2$, RMS and MAPE values for the ANN model were 0.9892, 0.1715 and 0.0212, respectively. Furthermore, the $R^2$, RMS and MAPE values for the ANFIS model were 0.9831, 0.1947 and 0.0270, respectively. As a result, in the models, the training and testing results indicated that experimental data can be estimated to a superior close extent by the ANN and ANFIS models.