HOME - 지능시스템

* 주요 연구 과제



- Re-identification 기반 Multi-Object Tracking (넥스트칩 2022)

- 딥러닝 지식증류 기법의 심화 및 물체검출 네트워크의 개선과 간소화 응용 (연구재단 2021-)

- 딥러닝 (LSTM) 기반 기상 예측

- 임베디드 적용을 위한 진화연산 기반 CNN 필터 축소 및 구조/파라미터 최적화 (연구재단2019-2022)

- 딥러닝 기반 동풍 예측 (한국기상산업기술원, 2018-2020)


* 관련 분야 : 인공지능, 계산지능, 기계학습, 컴퓨터비전, 지능로봇, 모델 예측



* 최근 주요 논문



- S. Kim, S. Kang, H. Choi, S. S. Kim, K. Seo, “Keypoint Aware Robust Representation for Transformer-based Re-identification of Occluded Person,” IEEE Signal Processing Letters, Accepted 2022. 12, IF 3.201


- K. Zhou, S-K. Oh, J. Quin, W. Pedrycz, K. Seo, “Reinforced Two-stream Fuzzy Neural Networks Architecture Realized with the Aid of 1D/2D Data Features,” IEEE Transactions on Fuzzy Systems, Early Access 2022, SCI, IF 12.029


- S. Lee, S. Kim, S. S. Kim, K. Seo, “Similarity-based adversarial knowledge distillation using graph convolutional neural network,“, Electronics Letters, Wiley, Volume , Issue , pp , To Be Appeared 2022, SCIE IF 1.314


- S. Kang, S. S. Kim, K. Seo, “Genetic Algorithm-Based Structure Reduction for Convolutional Neural Network,“, JEET(Journal of Electrical Engineering and Technology), Springer, Volume , Issue , pp , Online 2022, SCIE IF 1.069


- S-B. Roh, S-K. Oh, W. Pedrycz, Z. Wang, Z. Fu, K. Seo, “Design of Iterative Fuzzy Radial Basis Function Neural Networks Based on Iterative Weighted Fuzzy C-Means Clustering and Weighted LSE Estimation,” IEEE Transactions on Fuzzy Systems, Early Access 2022, SCI, IF 12.029


- J. Kim, K. Jeong, H. Choi, K. Seo, “GAN-based Anomaly Detection in Imbalance Problems”, ECCV-2020 Workshops / Top AI 컨퍼런스 워크샵 논문 발표


- E. Kim, J. Ko, S. Oh, K. Seo, “Design of Meteorological Pattern Classification System Based on FCM-based Radial Basis Function Neural Networks Using Meteorological Radar Data“, Soft Computing, Springer, March 2019, Volume 23, Issue 6,pp 1857–1872, SCIE IF 3.643 


- S-B J. Yoon, D. Kyeong, K. Seo, “A hybrid method based on F-transform for robust estimators”, International Journal of Approximate Reasoning, Elsevier, Volume 104, January 2019, pp. 75-83, SCI IF 3.816


- J. Kim, M. Lee, J. Choi, K. Seo, “GA-based Filter Selection for Representation in Convolutional Neural Networks”, ECCV-2018 Workshops / Top AI 컨퍼런스 워크샵 논문 발표




Keypoint Aware Robust Representation for Transformer-based Re-identification of Occluded Person


Occluded person Re-identification is a challenging task which aims to find or distinguish a specific person when the human body is occluded by obstacles, other persons or oneself. Some recent state-of-the-art works adopting a transformer and/or pose-guided methods have improved the feature representation and performances, but are still in trouble with both weak representation and heavy structure. In this paper, we suggest the novel methods of transformer-based Re-identification for the occluded person as follows. First, in data augmentation, instead of deleting an arbitrary area, only a part containing the keypoint features of a person is deleted for effective learning in occlusion. Second, we suggest a unique hierarchical patch and feature attention combining the reliability-enhanced heatmaps and the output of the transformer intermediate layer, which can more effectively pay attention to the non-occluded human region. Third, we propose an entire-partial loss function with non-similarity grouping for more robust feature representation. As a result, our suggested model not only outperforms the existing methods, but also has the smallest scale among state-of-the-art methods. We compare our approach and various existing methods for the mAP and Rank-1 performances on the Occluded-Duke, Occluded-ReID, Market-1501 and DukeMTMC datasets. Experimental results show that our proposed model outperforms the state-of-the-art methods.





Similarity-based adversarial knowledge distillation using graph convolutional neural network


This letter presents an adversarial knowledge distillation based on graph convolutional neural network. For knowledge distillation, many methods have been proposed in which the student model individually and independently imitates the output of the teacher model on the input data. Our method suggests the application of a similarity matrix to consider the relationship among output vectors, compared to the other existing approaches. The similarity matrix of the output vectors is calculated and converted into a graph structure, and a generative adversarial network using graph convolutional neural network is applied. We suggest similarity-based knowledge distillation in which a student model simultaneously imitates both of output vector and similarity matrix of the teacher model. We evaluate our method on ResNet, MobileNet and Wide ResNet using CIFAR-10 and CIFAR-100 datasets, and our results outperform results of the baseline model and other existing knowledge distillations like KLD, DML.





Boosted Knowledge Distillation Plug-in Using Controlled Knowledge Transfer


Knowledge distillation is a useful way to transfer knowledge from a teacher to a student network. The typical application is to transfer from a powerful large network to a small network, in order to meet the low-memory and/or fast execution requirements. Because conventional knowledge distillation depend only upon Kullback Leibler (KL) divergence, misled knowledge can be transferred, resulting in inaccuracy. In this paper, we suggest a controlling method to improve the efficiency of knowledge transfer. Specifically, new knowledge representation combining KL divergence and predictive accuracy of student is suggested and the knowledge are segmented to adjust the transfer rate based on the characteristics of the group. We reduce the reflection rate of losses for higher confidence of student, and increase the reflection rate of losses for lower confidence of student. By introducing sophisticated control technique, we can improve the efficiency of knowledge transfer in comparison with the existing knowledge distillation. The proposed method provides, 1) increasing the efficiency of knowledge transfer, thus improving the performance and accelerating the learning process, 2) preventing overfitting and obtaining effective regularization, 3) easily plugged in both conventional and collaborative knowledge distillation. To show the effectiveness of our approach for compression problem, we demonstrate the accuracy through performing experiments on CIFAR100, Cub-200-2011, Cars196, and Tiny ImageNet data for ResNet, MobileNet, and WRN models. We compare our results with those of existing knowledge distillation methods and obtain far better results. We also plug in ours to existing methods and to obtain much improved results than of original method. Code is available at https://github.com/ksj13/CKD.





Evolutionary Pruning of CNN Filters


Recent mobile and embedded systems re- quire e_cient visual recognition that builds on a deep learning model such as a convolutional neural network (CNN). Since features of a CNN are automatically generated for deep layers, there exists redundancy in features, which can be further compressed. Therefore, we propose evolutionary pruning algorithms to remove the irrelevant deep features in order to minimize the computational complexity and overfitting while maintaining a good quality of representation. Our proposed filter pruning via genetic algorithms (GA) helps to organize the totality of filters considering each filter's relationship and its niche among neighbors while the existing algorithmic pruning mainly focuses on weights of individual filters or activation reconstruction errors. With new GA operations included in our pruning process, we show that our GA method achieves the global optimization solution rather than a local optimization solution normally achieved by other algorithmic pruning approaches. We demonstrate a significant improvement of the filter representation by performing experiments on CIFAR-10 and smartphone metal exterior defect data. Our method is compared with results from the state-of-the-art algorithmic pruning methods in order to show its significant performance improvement by analyzing the classification performance as well as the compression rate and the oating point operations per second (FLOPS). We also show a way how to combine our computationally demanding GA method with existing fast algorithms, in order to implement our GA method to a larger network.





Genetic Programming based Gait Generation for Quadruped Robot


Planning gaits for legged robots is a challenging task that requires optimizing parameters in a highly irregular and multi-dimensional space. Several recent approaches have focused on using genetic algorithms (GAs) to generate gaits automatically and have shown significant improvement over previous gait optimization results. Most current GA-based approaches optimize only a small, pre-selected set of parameters, but it is difficult to decide which parameters should be included in the optimization to get the best results. Moreover, the number of pre-selected parameters is at least 10, so it can be relatively difficult to optimize them, given their high degree of interdependence. To overcome these problems of the typical GA-based approach, we have proposed a seemingly more efficient approach that optimizes joint trajectories instead of locus-related parameters in Cartesian space, using GP. Our GP-based method has obtained much-improved results over the GA-based approaches tested in experiments on the Sony AIBO ERS-7 in the Webots environment. The elite archive mechanism is introduced to combat the premature convergence problems in GP and has shown better results than a traditional multi-population approach.



CPG Based Gait Generation for Humanoid Robots


The research proposes generation methods for humanoid robot walking using a central pattern generator (CPG) based joint modification method to improve the adaptation ability on various slop terrains. Although CPG has basic adaptation ability for the different terrains, but it might be limited to small variations of those. In order to increase terrain adaptability we supplement the modification of joints using genetic algorithm on the output signals of the CPG oscillators. We have tested the two methods to investigate an adaptation capabilities of humanoid walking on various slope terrains with different slope angles using the humanoid robot Nao in the Webot simulation. The performances of humanoid walking on various slope terrains are analyzed.




Neural Networks Based on Iterative Weighted Fuzzy C-Means Clustering and Weighted LSE Estimation


In this study, a reinforced Iterative Fuzzy Radial Basis Function Neural Networks (IFRBFNN) is introduced as an augmented FRBFNN architecture generated through the iterative refinement process of weighted Fuzzy C-Means clustering and weighted LSE estimation. It is well known that the location of the clusters have an effect on the classification performance of the FRBFNN. The underlying idea behind this study is how to define the center points of clusters based on the data distribution analysis as well as the improvement of classification performance. In a nutshell, while the Fuzzy C-Means clustering to define the positions of Radial Basis Functions based in the unsupervised learning manner is usually used in a FRBFNN, the parameters (cluster centers and their ensuing polynomial coefficients) refinement of FRBFNN through the proposed iterative method lead to superb classification performance by relocating the positions of radial basis functions over the input space implied by supervised learning. The idea of the proposed approach is to relocate the centers (prototypes) of the fuzzy clusters by using weighted Fuzzy CMeans clustering algorithm and to re-estimate the coefficients by Weighted Least Square Estimation with the aid of the crossentropy loss values of data so that the classification performance is improved. The weight related to each data is determined by some auxiliary information (i.e., the modified version of the crossentropy loss value of each data). When it comes to the estimation of the coefficients of the consequent polynomial in IFRBFNN, they are estimated by the Weighted Least Square Error estimation technique, where the weights of the coefficients are defined based on the cross-entropy loss values. The auxiliary information such as the weights for re-location of the centers of RBFs and the weights for the re-estimation of the polynomial coefficients should be defined from the viewpoint of enhancement of the classification performance. The cross-entropy loss value of each data is able to meet the necessity of the required supervision signal. Several numerical experiments are provided to demonstrate the usefulness





Sketch Classification and Sketch Based Image Retrieval Using ViT with Self-Distillation for Few Samples


Sketch-based image retrieval (SBIR) with Zero-Shot are challenging tasks in computer vision, enabling to retrieve photo images relevant to sketch queries that have not been seen in the training phase. For sketch images without a sequence of information, we propose a modified Vision Transformer (ViT)-based approach that enhances or maintains the performance while reducing the number of sketch training data. In order to improve the structure of ViT network, we add a token for retrieval and integrate auxiliary classifiers of multiple branches in the network. Self-distillation is applied to enable fast transfer learning of sketch domains for our ViT network. For further improvement of combining self-distillation in ViT network, we adopt scheme of adding classifiers and embedding vectors to each intermediate layers in the network. Experiments on the TU-Berlin and Sketchy dataset demonstrate show that our method performs a significant improvement over other state-of-the-art methods on sketch classification and sketch-based image retrieval.





Surface Defect Image Classification of Metal Cases Using CNN


we propose a method of surface defect image classification for metal smartphone cases using a Convolutional Neural Network (CNN) deep learning model. Currently, the smartphone industry performs surface defect classification manually, which is subjective and highly expensive. Surface defect classification for metal smartphone cases is a highly difficult problem due to the presence of shallow and fine cracks as well as defects that are indistinguishable under different illumination conditions. This creates difficulties for automated surface inspection (ASI) methods using conventional machine vision techniques, such as support vector machine (SVM) models with feature engineering. We apply a modern machine vision technique based on CNN deep learning in order to improve the accuracy of defect classification over what can be achieved with conventional machine learning techniques. In contrast to other studies, we show the feasibility and effectiveness of our CNN models using real-word data on metal smartphone case images with and without defects under different surface and lighting conditions. In addition, we analyze different learning behaviors on three different data sets under different learning strategies. The results of our work in this study have the potential to have a significant impact on the smartphone manufacturing industry.





Corner Detection and Object Tracking


This research introduces evolutionary generation method of robust corner detectors for rotated images. Previous Harris, SUSAN and FAST corner detectors are highly efficient for well-defined corners, but frequently mis-detect as corners the corner–like edges which are often generated in rotated images. In this paper, we have focused on this challenging problem and proposed using Genetic Programming to do automated generation of corner detectors that work robustly on rotated images. Especially, a well-devised terminal set is proposed based on intensity-related information, several mask sizes, and amount of contiguity of neighboring pixels of similar intensity. This method is then compared to three existing corner detectors on test images and shows superior results.




GP 진화연산 이용 강풍특보 예측


This research introduces an evolutionary nonlinear regression based compensation technique for the short-range prediction of wind speed using AWS(Automatic Weather Station) data. Development of an efficient MOS(Model Output Statistics) is necessary to correct systematic errors of the model, but a linear regression based MOS is hard to manage an irregular nature of weather prediction. In order to solve the problem, a nonlinear and symbolic regression method using GP(Genetic Programming) is suggested for a development of MOS wind forecast guidance. Also FCM(Fuzzy C-Means) clustering is adopted to mitigate bias of wind speed data. The purpose of this study is to evaluate the accuracy of the estimation by a GP based nonlinear MOS for 3 days prediction of wind speed in South Korean regions. This method is then compared to the UM model and has shown superior results. Data for 2007-2009, 2011 is used for training, and 2012 is used for testing.