• Journal of Broadcast Engineering
• /
• v.25 no.3
• /
• pp.428-438
• /
• 2020

The jitter in the digital videos lowers the visibility and degrades the efficiency of image processing and image compressing. In this paper, we propose a video stabilizer architecture based on triplet CNN and a method of synthesizing training datasets based on video synthesis. Compared with a conventional deep-learning video stabilization method, the proposed video stabilizer can reduce wobbling distortion.

• International journal of advanced smart convergence
• /
• v.5 no.3
• /
• pp.8-15
• /
• 2016

In recent years, one of deep learning models called Deep Belief Network (DBN) which formed by stacking restricted Boltzman machine in a greedy fashion has beed widely used for classification and recognition. With an ability to extracting features of high-level abstraction and deal with higher dimensional data structure, this model has ouperformed outstanding result on image and speech recognition. In this research, we assess the applicability of deep learning in dna classification level. Since the training phase of DBN is costly expensive, specially if deals with DNA sequence with thousand of variables, we introduce a new encoding method, using decimal-binary vector to represent the sequence as input to the model, thereafter compare with one-hot-vector encoding in two datasets. We evaluated our proposed model with different contrastive algorithms which achieved significant improvement for the training speed with comparable classification result. This result has shown a potential of using decimal-binary vector on DBN for DNA sequence to solve other sequence problem in bioinformatics.

• Journal of Multimedia Information System
• /
• v.7 no.1
• /
• pp.41-44
• /
• 2020

In autonomous driving, object recognition based on machine learning is one of the core software technologies. In particular, the object recognition using deep learning becomes an essential element for autonomous driving software to operate. In this paper, we introduce a drivable area detection method based on Region-based CNN model to support autonomous driving. To effectively detect the drivable area, we used the BDD dataset for model training and demonstrated its effectiveness. As a result, our R-CNN model using BDD datasets showed interesting results in training and testing for detection of drivable areas.

• Journal of Information Processing Systems
• /
• v.17 no.3
• /
• pp.441-452
• /
• 2021

Federated learning provides an efficient integrated model for distributed data, allowing the local training of different data. Meanwhile, the goal of multi-task learning is to simultaneously establish models for multiple related tasks, and to obtain the underlying main structure. However, traditional federated multi-task learning models not only have strict requirements for the data distribution, but also demand large amounts of calculation and have slow convergence, which hindered their promotion in many fields. In our work, we apply the rank constraint on weight vectors of the multi-task learning model to adaptively adjust the task's similarity learning, according to the distribution of federal node data. The proposed model has a general framework for solving optimal solutions, which can be used to deal with various data types. Experiments show that our model has achieved the best results in different dataset. Notably, our model can still obtain stable results in datasets with large distribution differences. In addition, compared with traditional federated multi-task learning models, our algorithm is able to converge on a local optimal solution within limited training iterations.

• Journal of Institute of Convergence Technology
• /
• v.11 no.1
• /
• pp.1-5
• /
• 2021

Deep learning based methods achieve state-of-the-art accuracy, however, they typically rely on supervised training with large labeled datasets. It is known in many medical applications that labeling medical images requires significant expertise and much time, and typical hand-tuned approaches for data augmentation fail to capture the complex variations in such images. This paper proposes a 3D image augmentation method to overcome these difficulties. It allows us to enrich diversity of training data samples that is essential in medical image segmentation tasks, thus reducing the data overfitting problem caused by the fact the scale of medical image dataset is typically smaller. Our numerical experiments demonstrate that the proposed approach provides significant improvements over state-of-the-art methods for 3D medical image segmentation.

• Proceedings of the Korea Information Processing Society Conference
• /
• 2023.05a
• /
• pp.230-232
• /
• 2023

Federated Learning (FL) is a technique that excels in training a global model using numerous clients while only sharing the parameters of their local models, which were trained on their private training datasets. As a result, clients can obtain a high-performing deep learning (DL) model without having to disclose their private data. This setup is based on the understanding that all clients share the common goal of developing a global model with high accuracy. However, recent studies indicate that the security of gradient sharing may not be as reliable as previously thought. This paper introduces the latest research on various attacks that threaten the privacy of federated learning.

• Korean Journal of Radiology
• /
• v.25 no.2
• /
• pp.126-133
• /
• 2024

Large language models (LLMs) have revolutionized the global landscape of technology beyond natural language processing. Owing to their extensive pre-training on vast datasets, contemporary LLMs can handle tasks ranging from general functionalities to domain-specific areas, such as radiology, without additional fine-tuning. General-purpose chatbots based on LLMs can optimize the efficiency of radiologists in terms of their professional work and research endeavors. Importantly, these LLMs are on a trajectory of rapid evolution, wherein challenges such as "hallucination," high training cost, and efficiency issues are addressed, along with the inclusion of multimodal inputs. In this review, we aim to offer conceptual knowledge and actionable guidance to radiologists interested in utilizing LLMs through a succinct overview of the topic and a summary of radiology-specific aspects, from the beginning to potential future directions.

• Asia-Pacific Journal of Business
• /
• v.13 no.1
• /
• pp.125-139
• /
• 2022

Purpose - This study examines how employees' perception of HR department expertise affect their satisfaction of education and training. Moreover, this study explores that the HR department's communication activities moderate the main effects between satisfaction of education and training opportunities. Design/methodology/approach - This study predicts the positive relationship between employees' perceptions of HR department expertise and their satisfaction of education and training. Furthermore, the HR department's communication activities will strengthen this positive relationship. To test these hypotheses, this study used the Human Capital Corporate Panel (HCCP) datasets, especially individual-level 2017 data. The final number of samples is 1,947 for the analyses. In addition, this study utilized a hierarchical regression model with SPSS program. Finding - The results analyzed with the hierarchical regression model showed that the perceptions of HR department expertise had a positive relationship with satisfaction of provided educational and training. In addition, the HR department's communication activities moderated the relationship between perception of HR department expertise and satisfaction of education and training opportunities. Research implications or Originality - This study suggests academic and practical implications for future research in the human resource development filed by clarifying the critical factors to increase employees' satisfaction and transferability of education and training.

• Journal of KIISE
• /
• v.44 no.12
• /
• pp.1275-1281
• /
• 2017

The supervised learning approach is suitable for classification of insulting sentences, but pre-decided training sentences are necessary. Since a Character-level Convolution Neural Network is robust for each character, so is appropriate for classifying abusive sentences, however, has a drawback that demanding a lot of training sentences. In this paper, we propose transfer learning method that reusing the trained filters in the real classification process after the filters get the characteristics of offensive words by generated abusive/normal pair of sentences. We got higher performances of the classifier by decreasing the effects of data shortage and class imbalance. We executed experiments and evaluations for three datasets and got higher F1-score of character-level CNN classifier when applying transfer learning in all datasets.

• Computers and Concrete
• /
• v.27 no.4
• /
• pp.333-341
• /
• 2021

Steel slag, an industrial reject from the steel rolling process, has been identified as one of the suitable, environmentally friendly materials for concrete production. Given that the coarse aggregate portion represents about 70% of concrete constituents, other economic approaches have been found in the use of alternative materials such as steel slag in concrete. Unfortunately, a standard framework for its application is still lacking. Therefore, this study proposed functional model equations for the determination of strength properties (compression and splitting tensile) of steel slag aggregate concrete (SSAC), using gene expression programming (GEP). The study, in the experimental phase, utilized steel slag as a partial replacement of crushed rock, in steps 20%, 40%, 60%, 80%, and 100%, respectively. The predictor variables included in the analysis were cement, sand, granite, steel slag, water/cement ratio, and curing regime (age). For the model development, 60-75% of the dataset was used as the training set, while the remaining data was used for testing the model. Empirical results illustrate that steel aggregate could be used up to 100% replacement of conventional aggregate, while also yielding comparable results as the latter. The GEP-based functional relations were tested statistically. The minimum absolute percentage error (MAPE), and root mean square error (RMSE) for compressive strength are 6.9 and 1.4, and 12.52 and 0.91 for the train and test datasets, respectively. With the consistency of both the training and testing datasets, the model has shown a strong capacity to predict the strength properties of SSAC. The results showed that the proposed model equations are reliably suitable for estimating SSAC strength properties. The GEP-based formula is relatively simple and useful for pre-design applications.