• Journal of Gastric Cancer
• /
• v.22 no.2
• /
• pp.120-134
• /
• 2022

Purpose: This study aimed to identify prognostic factors for patients with distant lymph node-involved gastric cancer (GC) using a machine learning algorithm, a method that offers considerable advantages and new prospects for high-dimensional biomedical data exploration. Materials and Methods: This study employed 79 features of clinical pathology, laboratory tests, and therapeutic details from 289 GC patients whose distant lymphadenopathy was presented as the first episode of recurrence or metastasis. Outcomes were measured as any-cause death events and survival months after distant lymph node metastasis. A prediction model was built based on possible outcome predictors using a random survival forest algorithm and confirmed by 5×5 nested cross-validation. The effects of single variables were interpreted using partial dependence plots. A contour plot was used to visually represent survival prediction based on 2 predictive features. Results: The median survival time of patients with GC with distant nodal metastasis was 9.2 months. The optimal model incorporated the prealbumin level and the prothrombin time (PT), and yielded a prediction error of 0.353. The inclusion of other variables resulted in poorer model performance. Patients with higher serum prealbumin levels or shorter PTs had a significantly better prognosis. The predicted one-year survival rate was stratified and illustrated as a contour plot based on the combined effect the prealbumin level and the PT. Conclusions: Machine learning is useful for identifying the important determinants of cancer survival using high-dimensional datasets. The prealbumin level and the PT on distant lymph node metastasis are the 2 most crucial factors in predicting the subsequent survival time of advanced GC.

• Journal of Intelligence and Information Systems
• /
• v.29 no.4
• /
• pp.165-184
• /
• 2023

Recently, Deep learning analysis of unstructured text data using language models, such as Google's BERT and OpenAI's GPT has shown remarkable results in various applications. Most language models are used to learn generalized linguistic information from pre-training data and then update their weights for downstream tasks through a fine-tuning process. However, some concerns have been raised that privacy may be violated in the process of using these language models, i.e., data privacy may be violated when data owner provides large amounts of data to the model owner to perform fine-tuning of the language model. Conversely, when the model owner discloses the entire model to the data owner, the structure and weights of the model are disclosed, which may violate the privacy of the model. The concept of offsite tuning has been recently proposed to perform fine-tuning of language models while protecting privacy in such situations. But the study has a limitation that it does not provide a concrete way to apply the proposed methodology to text classification models. In this study, we propose a concrete method to apply offsite tuning with an additional classifier to protect the privacy of the model and data when performing multi-classification fine-tuning on Korean documents. To evaluate the performance of the proposed methodology, we conducted experiments on about 200,000 Korean documents from five major fields, ICT, electrical, electronic, mechanical, and medical, provided by AIHub, and found that the proposed plug-in model outperforms the zero-shot model and the offsite model in terms of classification accuracy.

• Journal of Intelligence and Information Systems
• /
• v.17 no.1
• /
• pp.53-69
• /
• 2011

Vision and voice-based technologies are commonly utilized for human-robot interaction. But it is widely recognized that the performance of vision and voice-based interaction systems is deteriorated by a large margin in the real-world situations due to environmental and user variances. Human users need to be very cooperative to get reasonable performance, which significantly limits the usability of the vision and voice-based human-robot interaction technologies. As a result, touch screens are still the major medium of human-robot interaction for the real-world applications. To empower the usability of robots for various services, alternative interaction technologies should be developed to complement the problems of vision and voice-based technologies. In this paper, we propose the use of accelerometer-based gesture interface as one of the alternative technologies, because accelerometers are effective in detecting the movements of human body, while their performance is not limited by environmental contexts such as lighting conditions or camera's field-of-view. Moreover, accelerometers are widely available nowadays in many mobile devices. We tackle the problem of classifying acceleration signal patterns of 26 English alphabets, which is one of the essential repertoires for the realization of education services based on robots. Recognizing 26 English handwriting patterns based on accelerometers is a very difficult task to take over because of its large scale of pattern classes and the complexity of each pattern. The most difficult problem that has been undertaken which is similar to our problem was recognizing acceleration signal patterns of 10 handwritten digits. Most previous studies dealt with pattern sets of 8~10 simple and easily distinguishable gestures that are useful for controlling home appliances, computer applications, robots etc. Good features are essential for the success of pattern recognition. To promote the discriminative power upon complex English alphabet patterns, we extracted 'motion trajectories' out of input acceleration signal and used them as the main feature. Investigative experiments showed that classifiers based on trajectory performed 3%~5% better than those with raw features e.g. acceleration signal itself or statistical figures. To minimize the distortion of trajectories, we applied a simple but effective set of smoothing filters and band-pass filters. It is well known that acceleration patterns for the same gesture is very different among different performers. To tackle the problem, online incremental learning is applied for our system to make it adaptive to the users' distinctive motion properties. Our system is based on instance-based learning (IBL) where each training sample is memorized as a reference pattern. Brute-force incremental learning in IBL continuously accumulates reference patterns, which is a problem because it not only slows down the classification but also downgrades the recall performance. Regarding the latter phenomenon, we observed a tendency that as the number of reference patterns grows, some reference patterns contribute more to the false positive classification. Thus, we devised an algorithm for optimizing the reference pattern set based on the positive and negative contribution of each reference pattern. The algorithm is performed periodically to remove reference patterns that have a very low positive contribution or a high negative contribution. Experiments were performed on 6500 gesture patterns collected from 50 adults of 30~50 years old. Each alphabet was performed 5 times per participant using $Nintendo{(R)}$ $Wii^{TM}$ remote. Acceleration signal was sampled in 100hz on 3 axes. Mean recall rate for all the alphabets was 95.48%. Some alphabets recorded very low recall rate and exhibited very high pairwise confusion rate. Major confusion pairs are D(88%) and P(74%), I(81%) and U(75%), N(88%) and W(100%). Though W was recalled perfectly, it contributed much to the false positive classification of N. By comparison with major previous results from VTT (96% for 8 control gestures), CMU (97% for 10 control gestures) and Samsung Electronics(97% for 10 digits and a control gesture), we could find that the performance of our system is superior regarding the number of pattern classes and the complexity of patterns. Using our gesture interaction system, we conducted 2 case studies of robot-based edutainment services. The services were implemented on various robot platforms and mobile devices including $iPhone^{TM}$. The participating children exhibited improved concentration and active reaction on the service with our gesture interface. To prove the effectiveness of our gesture interface, a test was taken by the children after experiencing an English teaching service. The test result showed that those who played with the gesture interface-based robot content marked 10% better score than those with conventional teaching. We conclude that the accelerometer-based gesture interface is a promising technology for flourishing real-world robot-based services and content by complementing the limits of today's conventional interfaces e.g. touch screen, vision and voice.

• Korean Journal of Remote Sensing
• /
• v.39 no.5_3
• /
• pp.1009-1029
• /
• 2023

Urban trees play a vital role in urban ecosystems,significantly reducing impervious surfaces and impacting carbon cycling within the city. Although previous research has demonstrated the efficacy of employing artificial intelligence in conjunction with airborne light detection and ranging (LiDAR) data to generate urban tree information, the availability and cost constraints associated with LiDAR data pose limitations. Consequently, this study employed freely accessible, high-resolution multispectral satellite imagery (i.e., Sentinel-2 data) to estimate fractional tree canopy cover (FTC) within the urban confines of Suwon, South Korea, employing machine learning techniques. This study leveraged a median composite image derived from a time series of Sentinel-2 images. In order to account for the diverse land cover found in urban areas, the model incorporated three types of input variables: average (mean) and standard deviation (std) values within a 30-meter grid from 10 m resolution of optical indices from Sentinel-2, and fractional coverage for distinct land cover classes within 30 m grids from the existing level 3 land cover map. Four schemes with different combinations of input variables were compared. Notably, when all three factors (i.e., mean, std, and fractional cover) were used to consider the variation of landcover in urban areas(Scheme 4, S4), the machine learning model exhibited improved performance compared to using only the mean of optical indices (Scheme 1). Of the various models proposed, the random forest (RF) model with S4 demonstrated the most remarkable performance, achieving R² of 0.8196, and mean absolute error (MAE) of 0.0749, and a root mean squared error (RMSE) of 0.1022. The std variable exhibited the highest impact on model outputs within the heterogeneous land covers based on the variable importance analysis. This trained RF model with S4 was then applied to the entire Suwon region, consistently delivering robust results with an R² of 0.8702, MAE of 0.0873, and RMSE of 0.1335. The FTC estimation method developed in this study is expected to offer advantages for application in various regions, providing fundamental data for a better understanding of carbon dynamics in urban ecosystems in the future.

• The Journal of Society for e-Business Studies
• /
• v.24 no.4
• /
• pp.119-133
• /
• 2019

Recently, the importance prediction of photovoltaic power (PV) is considered as an essential function for scheduling adjustments, deciding on storage size, and overall planning for stable operation of PV facility systems. In particular, since most of PV power is generated in peak time, PV power prediction in a peak time is required for the PV system operators that enable to maximize revenue and sustainable electricity quantity. Moreover, Prediction of the PV power output in peak time without meteorological information such as solar radiation, cloudiness, the temperature is considered a challenging problem because it has limitations that the PV power was predicted by using predicted uncertain meteorological information in a wide range of areas in previous studies. Therefore, this paper proposes the LSTM (Long-Short Term Memory) based the PV power prediction model only using the meteorological, seasonal, and the before the obtained PV power before peak time. In this paper, the experiment results based on the proposed model using the real-world data shows the superior performance, which showed a positive impact on improving the PV power in a peak time forecast performance targeted in this study.

• Journal of The Korean Association For Science Education
• /
• v.19 no.1
• /
• pp.29-40
• /
• 1999

The purpose of the present study was to test hypothesis that, because it uses tri-dimensional sensory pathway which have been showed a higher rate of neural activities than uni- or bi-dimensional's, lab-activity-based instruction is more effective teaching strategy in learning science than verbal-based instruction. In the present study, manipulative teaching strategy that uses visual, somatosensory and auditory information pathway was regarded as a mode of tri-dimensional sensory inputs. In addition, verbal teaching strategy that uses mainly auditory and a little visual information pathway was used as a mode of bi-dimensional sensory inputs. Fifty-six students who failed to successfully solve two proportional reasoning tasks (i.e., pouring water tasks) were sampled for this research from a junior high school. The subjects were randomly divided into a manipulative or a verbal teaching group, and given manipulative or verbal tutoring on the use of proportional reasoning strategies and a test of proportional reasoning during instruction. The results showed that manipulative group's performance on the test of proportional reasoning during instruction showed significantly higher performance than verbal group's (t=2.45, p<0.02). The present study also discussed some educational implications of the results.

• Journal of Internet Computing and Services
• /
• v.24 no.4
• /
• pp.127-135
• /
• 2023

This research paper explores the application of FinBERT, a variational BERT-based model pre-trained on financial domain, for sentiment analysis in the financial domain while focusing on the process of identifying suitable training data and hyperparameters. Our goal is to offer a comprehensive guide on effectively utilizing the FinBERT model for accurate sentiment analysis by employing various datasets and fine-tuning hyperparameters. We outline the architecture and workflow of the proposed approach for fine-tuning the FinBERT model in this study, emphasizing the performance of various datasets and hyperparameters for sentiment analysis tasks. Additionally, we verify the reliability of GPT-3 as a suitable annotator by using it for sentiment labeling tasks. Our results show that the fine-tuned FinBERT model excels across a range of datasets and that the optimal combination is a learning rate of 5e-5 and a batch size of 64, which perform consistently well across all datasets. Furthermore, based on the significant performance improvement of the FinBERT model with our Twitter data in general domain compared to our news data in general domain, we also express uncertainty about the model being further pre-trained only on financial news data. We simplify the complex process of determining the optimal approach to the FinBERT model and provide guidelines for selecting additional training datasets and hyperparameters within the fine-tuning process of financial sentiment analysis models.

• The Journal of Society for e-Business Studies
• /
• v.27 no.2
• /
• pp.205-218
• /
• 2022

Recently, digital transformation in manufacturing has been accelerating. It results in that the data collection technologies from the shop-floor is becoming important. These approaches focus primarily on obtaining specific manufacturing data using various sensors and communication technologies. In order to expand the channel of field data collection, this study proposes a method to automatically collect manufacturing data based on vision-based artificial intelligence. This is to analyze real-time image information with the object detection and tracking technologies and to obtain manufacturing data. The research team collects object motion information for each frame by applying YOLO (You Only Look Once) and DeepSORT as object detection and tracking algorithms. Thereafter, the motion information is converted into two pieces of manufacturing data (production performance and time) through post-processing. A dynamically moving factory model is created to obtain training data for deep learning. In addition, operating scenarios are proposed to reproduce the shop-floor situation in the real world. The operating scenario assumes a flow-shop consisting of six facilities. As a result of collecting manufacturing data according to the operating scenarios, the accuracy was 96.3%.

• Communications of Mathematical Education
• /
• v.25 no.1
• /
• pp.99-118
• /
• 2011

For this case study of gifted education, two classrooms in two locations, show life in general of the gifted educational system. And for this case study the identity of teachers and the gifted, help to activate the mathematically gifted education for these research questions, which are as followed: Firstly, how is the gifted education classroom life? Secondly, what kind of identity do the teachers and gifted students bring to mathematics, mathematics teaching and mathematics learning? Being selected in the gifted children's education center solves the research problem of characteristic and approach. Backed by the condition and the permission possibility, 2 selected classes and 2 people, which are coming and going. Gifted education classroom life, the identity of teachers and gifted students in mathematics and mathematics teaching and mathematic learning. It will be for 3 months, with various recordings and vocal instruction between teacher and students. Collected observations and interviews will be analyzed over the course of instruction. The results analyzed include, social participation, structure, and the formation of the gifted education classroom life. The organization of classes were analyzed by the classes conscious levels to collect and retain data. The classes verification levels depended on the program's first class incentive, teaching and learning levels and understanding of gifted math. A performance assessment will be applied after the final lesson and a consultation with parents and students after the final class. The six kinds of social participation structure come out of the type of the most important roles in gifted education accounts, for these types of group discussions and interactions, students must have an interaction or individual activity that students can use, such as a work product through the real materials, which release teachers and other students for that type of questions to evaluate. In order for the development of meaningful mathematical concepts to formulate, mathematical principles require problem solving among all students, which will appear in the resolution or it will be impossible to map the meaning of the instruction from which it was formed. These results show the analysis of the mathematics, mathematics teaching, mathematics learning and about the identity of the teachers and gifted. Gifted education teachers are defined by gifted math, which is more difficult and requires more differentiated learning, suitable for gifted students. Gifted was defined when higher level math was created and challenged students to deeper thinking. Gifted students think that gifted math is creative learning and they are forward or passive to one-way according to the education atmosphere.

• Korean Journal of Remote Sensing
• /
• v.39 no.5_3
• /
• pp.979-995
• /
• 2023

As wildfires are difficult to predict, real-time monitoring is crucial for a timely response. Geostationary satellite images are very useful for active fire detection because they can monitor a vast area with high temporal resolution (e.g., 2 min). Existing satellite-based active fire detection algorithms detect thermal outliers using threshold values based on the statistical analysis of brightness temperature. However, the difficulty in establishing suitable thresholds for such threshold-based methods hinders their ability to detect fires with low intensity and achieve generalized performance. In light of these challenges, machine learning has emerged as a potential-solution. Until now, relatively simple techniques such as random forest, Vanilla convolutional neural network (CNN), and U-net have been applied for active fire detection. Therefore, this study proposed an active fire detection algorithm using state-of-the-art (SOTA) deep learning techniques using data from the Advanced Himawari Imager and evaluated it over East Asia and Australia. The SOTA model was developed by applying EfficientNet and lion optimizer, and the results were compared with the model using the Vanilla CNN structure. EfficientNet outperformed CNN with F1-scores of 0.88 and 0.83 in East Asia and Australia, respectively. The performance was better after using weighted loss, equal sampling, and image augmentation techniques to fix data imbalance issues compared to before the techniques were used, resulting in F1-scores of 0.92 in East Asia and 0.84 in Australia. It is anticipated that timely responses facilitated by the SOTA deep learning-based approach for active fire detection will effectively mitigate the damage caused by wildfires.