• KSII Transactions on Internet and Information Systems (TIIS)
• /
• v.16 no.6
• /
• pp.1833-1848
• /
• 2022

Named entity recognition (NER) is an important basic task in the field of Natural Language Processing (NLP). Recently deep learning approaches by extracting word segmentation or character features have been proved to be effective for Chinese Named Entity Recognition (CNER). However, since this method of extracting features only focuses on extracting some of the features, it lacks textual information mining from multiple perspectives and dimensions, resulting in the model not being able to fully capture semantic features. To tackle this problem, we propose a novel Multi-view Semantic Feature Fusion Model (MSFM). The proposed model mainly consists of two core components, that is, Multi-view Semantic Feature Fusion Embedding Module (MFEM) and Multi-head Self-Attention Mechanism Module (MSAM). Specifically, the MFEM extracts character features, word boundary features, radical features, and pinyin features of Chinese characters. The acquired font shape, font sound, and font meaning features are fused to enhance the semantic information of Chinese characters with different granularities. Moreover, the MSAM is used to capture the dependencies between characters in a multi-dimensional subspace to better understand the semantic features of the context. Extensive experimental results on four benchmark datasets show that our method improves the overall performance of the CNER model.

• Proceedings of the Korean Society of Broadcast Engineers Conference
• /
• 2021.06a
• /
• pp.223-225
• /
• 2021

실감 미디어에 대한 수요가 증가함에 따라 고품질의 실감 미디어에 대한 중요성이 증가하고 있다. 이러한 실감미디어를 제작하기 위해 사용되는 일반적인 기법 중 하나인 Multi View Stereo 는 깊이 영상 추정 및 해당 깊이 영상을 이용하여 3 차원에 point cloud 를 생성하는 fusion 과정을 거치게 된다. 본 논문에서는 다중 시점 영상의 깊이 영상을 정합하는 fusion 과정을 개선하기 위한 방법을 제안한다. 제안하는 방법에서는 깊이 영상, 색상정보를 이용하여 기준 시점의 depth map 을 이용한 fusion 과정을 거친다. 실험을 통하여 제안한 알고리즘을 이용한 결과가 기존보다 개선됨을 보인다.

• KSII Transactions on Internet and Information Systems (TIIS)
• /
• v.13 no.11
• /
• pp.5546-5559
• /
• 2019

Facial action units (AUs) have recently drawn increased attention because they can be used to recognize facial expressions. A variety of methods have been designed for frontal-view AU detection, but few have been able to handle multi-view face images. In this paper we propose a method for multi-view facial AU detection using a fused multilayer, multi-task, and multi-label deep learning network. The network can complete two tasks: AU detection and facial view detection. AU detection is a multi-label problem and facial view detection is a single-label problem. A residual network and multilayer fusion are applied to obtain more representative features. Our method is effective and performs well. The F1 score on FERA 2017 is 13.1% higher than the baseline. The facial view recognition accuracy is 0.991. This shows that our multi-task, multi-label model could achieve good performance on the two tasks.

• Journal of Information Processing Systems
• /
• v.15 no.5
• /
• pp.1179-1191
• /
• 2019

We design an ingenious view-pooling method named learning-based multiple pooling fusion (LMPF), and apply it to multi-view convolutional neural network (MVCNN) for 3D model classification or retrieval. By this means, multi-view feature maps projected from a 3D model can be compiled as a simple and effective feature descriptor. The LMPF method fuses the max pooling method and the mean pooling method by learning a set of optimal weights. Compared with the hand-crafted approaches such as max pooling and mean pooling, the LMPF method can decrease the information loss effectively because of its "learning" ability. Experiments on ModelNet40 dataset and McGill dataset are presented and the results verify that LMPF can outperform those previous methods to a great extent.

• KSII Transactions on Internet and Information Systems (TIIS)
• /
• v.16 no.3
• /
• pp.813-829
• /
• 2022

In multi-view subspace clustering, how to integrate the complementary information between perspectives to construct a unified representation is a critical problem. In the existing works, the unified representation is usually constructed in the original data space. However, when the data representation in each view is very diverse, the unified representation derived directly in the original data domain may lead to a huge information loss. To address this issue, different to the existing works, inspired by the latest revelation that the data across all perspectives have a very similar or close spectral block structure, we try to construct the unified representation in the spectral embedding domain. In this way, the complementary information across all perspectives can be fused into a unified representation with little information loss, since the spectral block structure from all views shares high consistency. In addition, to capture the global structure of data on each view with high accuracy and robustness both, we propose a novel low-rank approximation via the tight lower bound on the rank function. Finally, experimental results prove that, the proposed method has the effectiveness and robustness at the same time, compared with the state-of-art approaches.

• Journal of Korean Neurosurgical Society
• /
• v.30 no.sup1
• /
• pp.55-60
• /
• 2001

Objective : The purpose of the present study was to examine neurologic changes, fusion rate and degree of kyphosis from the surgical results of those patients who underwent multi-level anterior interbody fusion and internal fixation. Methods : Among 63 cases of the patients who received multi-level anterior interbody fusion and internal fixation in 5 years between 1995 to 1999 at the neurosurgery department, we performed a retrospective study in 52 cases that could be followed up with dynamic view imaging ; the results were compared and analyzed. The analysis was based on the results of history taking, physical findings and radiologic findings, and Odom criteria were used to classify those cases with neurologic changes. Results : Among those 52 cases in whom the follow-up was possible for at least a year and dynamic view imaging was possible, bone fusion was seen in 93% of the trauma cases and 95% in the non-trauma cases and overall bone fusion was observed in 94% of the cases. Bone fusion was seen in 93% of the autobone cases, 95% of the allobone cases, and 94% of the Mesh cases. Radiologic changes were observed by comparing the lateral view after surgery ; kyphosis was seen in 53% of the autobone cases, in 70% of the allobone cases, and in 35% of Mesh cases ; in 45% and 60% of the non-trauma cases and trauma cases, respectively ; and in 55% of the 2 level fusion cases and 46% of the 3 level fusion cases. Neurologic changes classified according to Odom criteria showed excellent result in 48% of all the cases, good in 23%, fair in 4%, and poor in 25%. Conclusion : Even those cases with multi-level fusion, a high fusion rate could be obtained by performing anterior interbody fusion and internal fixation ; those cases with kyphosis were related more with the presence or absence of posterior compartment injury rather than the fusion level ; and those trauma cases showed not much difference in the fusion rate compared with non-trauma cases but had a high possibility of kyphosis.

• KSII Transactions on Internet and Information Systems (TIIS)
• /
• v.12 no.5
• /
• pp.2366-2395
• /
• 2018

Multi-view super-resolution (MVSR) refers to the process of reconstructing a high-resolution (HR) image from a set of low-resolution (LR) images captured from different viewpoints typically by different cameras. These multi-view images are usually obtained by a camera array. In our previous work [1], we super-resolved multi-view LR images via image fusion (IF) and blind deblurring (BD). In this paper, we present a new MVSR method that jointly realizes IF and BD based on an integrated energy function optimization. First, we reformulate the MVSR problem into a multi-channel blind deblurring (MCBD) problem which is easier to be solved than the former. Then the depth map of the desired HR image is calculated. Finally, we solve the MCBD problem, in which the optimization problems with respect to the desired HR image and with respect to the unknown blur are efficiently addressed by the alternating direction method of multipliers (ADMM). Experiments on the Multi-view Image Database of the University of Tsukuba and images captured by our own camera array system demonstrate the effectiveness of the proposed method.

• Journal of Broadcast Engineering
• /
• v.9 no.3
• /
• pp.185-195
• /
• 2004

This paper presents a multi-depth map fusion method for the 3D scene reconstruction. It fuses depth maps obtained from the stereo matching technique and the depth camera. Traditional stereo matching techniques that estimate disparities between two images often produce inaccurate depth map because of occlusion and homogeneous area. Depth map obtained from the depth camera is globally accurate but noisy and provide a limited depth range. In order to get better depth estimates than these two conventional techniques, we propose a depth map fusion method that fuses the multi-depth maps from stereo matching and the depth camera. We first obtain two depth maps generated from the stereo matching of 3-view images. Moreover, a depth map is obtained from the depth camera for the center-view image. After preprocessing each depth map, we select a depth value for each pixel among them. Simulation results showed a few improvements in some background legions by proposed fusion technique.

• The Transactions of The Korean Institute of Electrical Engineers
• /
• v.63 no.11
• /
• pp.1551-1558
• /
• 2014

We propose a multi-rate sensor fusion of vision and radar using Kalman filter to solve problems of asynchronized and multi-rate sampling periods in object vehicle tracking. A model based prediction of object vehicles is performed with a decentralized multi-rate Kalman filter for each sensor (vision and radar sensors.) To obtain the improvement in the performance of position prediction, different weighting is applied to each sensor's predicted object position from the multi-rate Kalman filter. The proposed method can provide estimated position of the object vehicles at every sampling time of ECU. The Mahalanobis distance is used to make correspondence among the measured and predicted objects. Through the experimental results, we validate that the post-processed fusion data give us improved tracking performance. The proposed method obtained two times improvement in the object tracking performance compared to single sensor method (camera or radar sensor) in the view point of roots mean square error.

• Journal of the Korea Society of Computer and Information
• /
• v.19 no.12
• /
• pp.11-19
• /
• 2014

In this paper, we design an intelligent agent system with the data mining module and information fusion module as the core components of the system and investigate the possibility for the medical expert system. In the data mining module, fuzzy neural network, OFUN-NET analyzes multi-view data and produces fuzzy cluster knowledge base. In the information fusion module and application module, they serve the diagnosis result with possibility degree and useful information for diagnosis, such as uncertainty decision status or detection of asymmetry. We also present the experiment results on the BI-RADS-based feature data set selected form DDSM benchmark database. They show higher classification accuracy than conventional methods and the feasibility of the system as a computer aided diagnosis system.