• Proceedings of the Korean Society of Computer Information Conference
• /
• 2021.07a
• /
• pp.313-315
• /
• 2021

Timely and accurate diagnosis of lung diseases using Chest X-ray images has been gained much attention from the computer vision and medical imaging communities. Although previous studies have presented the capability of deep convolutional neural networks by achieving competitive binary classification results, their models were seemingly unreliable to effectively distinguish multiple disease groups using a large number of x-ray images. In this paper, we aim to build an advanced approach, so-called Ensemble Knowledge Distillation (EKD), to significantly boost the classification accuracies, compared to traditional KD methods by distilling knowledge from a cumbersome teacher model into an ensemble of lightweight student models with parallel branches trained with ground truth labels. Therefore, learning features at different branches of the student models could enable the network to learn diverse patterns and improve the qualify of final predictions through an ensemble learning solution. Although we observed that experiments on the well-established ChestX-ray14 dataset showed the classification improvements of traditional KD compared to the base transfer learning approach, the EKD performance would be expected to potentially enhance classification accuracy and model generalization, especially in situations of the imbalanced dataset and the interdependency of 14 weakly annotated thorax diseases.

• Journal of Digital Convergence
• /
• v.19 no.5
• /
• pp.307-313
• /
• 2021

In this paper, a lightweight network with fewer parameters compared to the existing object detection method is proposed. In the case of the currently used detection model, the network complexity has been greatly increased to improve accuracy. Therefore, the proposed network uses EfficientNet as a feature extraction network, and the subsequent layers are formed in a pyramid structure to utilize low-level detailed features and high-level semantic features. An attention process was applied between pyramid structures to suppress unnecessary noise for prediction. All computational processes of the network are replaced by depth-wise and point-wise convolutions to minimize the amount of computation. The proposed network was trained and evaluated using the PASCAL VOC dataset. The features fused through the experiment showed robust properties for various objects through a refinement process. Compared with the CNN-based detection model, detection accuracy is improved with a small amount of computation. It is considered necessary to adjust the anchor ratio according to the size of the object as a future study.

• KSII Transactions on Internet and Information Systems (TIIS)
• /
• v.15 no.11
• /
• pp.4065-4083
• /
• 2021

Super-resolution can improve the clarity of low-resolution (LR) images, which can increase the accuracy of high-level compute vision tasks. Portable devices have low computing power and storage performance. Large-scale neural network super-resolution methods are not suitable for portable devices. In order to save the computational cost and the number of parameters, Lightweight image processing method can improve the processing speed of portable devices. Therefore, we propose the Enhanced Information Multiple Distillation Network (EIMDN) to adapt lower delay and cost. The EIMDN takes feedback mechanism as the framework and obtains low level features through high level features. Further, we replace the feature extraction convolution operation in Information Multiple Distillation Block (IMDB), with Ghost module, and propose the Enhanced Information Multiple Distillation Block (EIMDB) to reduce the amount of calculation and the number of parameters. Finally, coordinate attention (CA) is used at the end of IMDB and EIMDB to enhance the important information extraction from Spaces and channels. Experimental results show that our proposed can achieve convergence faster with fewer parameters and computation, compared with other lightweight super-resolution methods. Under the condition of higher peak signal-to-noise ratio (PSNR) and higher structural similarity (SSIM), the performance of network reconstruction image texture and target contour is significantly improved.

• Proceedings of the Korea Information Processing Society Conference
• /
• 2019.05a
• /
• pp.435-437
• /
• 2019

모바일 환경에서 다양한 AI 관련 응용을 수행하기 위해, 정확도에 기반한 크고 깊은 신경망 이외에, 정확도를 비교적 유지하면서 좀더 효율적인 신경망 구조에 대한 다양한 연구가 진행중이다. 본 논문에서는 모바일 딥러닝을 위한 다양한 임베디드 장치 및 모바일 폰에서의 성능 평가를 통해 경량 신경망의 비교 분석에 대한 연구를 담고 있다.

• Journal of Broadcast Engineering
• /
• v.24 no.2
• /
• pp.357-360
• /
• 2019

We propose an efficient single convolutional neural network (CNN) for pedestrian detection on embedded GPUs. We first determine the optimal number of the convolutional layers and hyper-parameters for a lightweight CNN. Then, we employ a multi-scale approach to make the network robust to the sizes of the pedestrians in images. Experimental results demonstrate that the proposed algorithm is capable of real-time operation, while providing higher detection performance than conventional algorithms.

• Proceedings of the Korean Society of Broadcast Engineers Conference
• /
• fall
• /
• pp.80-81
• /
• 2021

딥 러닝 모델 사용에 있어서, 일반적인 사용자가 이용할 수 있는 하드웨어 리소스는 제한적이기 때문에 기존 모델을 경량화 할 수 있는 프루닝 방법을 통해 제한적인 리소스를 효과적으로 활용할 수 있도록 한다. 그 방법으로, 여러 딥 러닝 모델들 중 비교적 파라미터 수가 많은 것으로 알려진 GAN 아키텍처에 네트워크 프루닝을 적용함으로써 비교적 무거운 모델을 적은 파라미터를 통해 학습할 수 있는 방법을 제시한다. 또한, 본 논문을 통해 기존의 SRGAN 논문에서 가장 효과적인 결과로 제시했던 16 개의 residual block 의 개수를 실제로 줄여 봄으로써 기존 논문에서 제시했던 결과와의 차이에 대해 서술한다.

• Proceedings of the Korean Society of Computer Information Conference
• /
• 2023.07a
• /
• pp.39-40
• /
• 2023

본 논문은 딥러닝 기반의 소리 인식 모델을 기반으로 실내에서 발생하는 다양한 소리를 시각적인 정보로 제공하는 시스템을 위해 경량화된 CNN ResNet 구조의 인공지능 모델을 제시한다. 적용하는 경량화 기법은 모델의 크기와 연산량을 최적화하여 자원이 제한된 장치에서도 효율적으로 동작할 수 있도록 한다. 이를 위해 마이크로 컴퓨터나 휴대용 기기와 같은 임베디드 장치에서도 원활한 인공지능 추론을 가능하게 하는 모델을 양자화 기법을 적용한 경량화 방법들을 실험적으로 비교한다.

• Proceedings of the Korean Institute of Information and Commucation Sciences Conference
• /
• 2022.10a
• /
• pp.87-89
• /
• 2022

Recently, as more consumers spend more time at home due to COVID-19, the time spent on digital consumption such as SNS and OTT, which can be easily used non-face-to-face, naturally increased. Since 2019, when COVID-19 occurred, digital consumption has doubled from 44% to 82%, and it is important to quickly and accurately grasp and apply trends by analyzing consumers' emotions due to the rapidly changing digital characteristics. However, there are limitations in actually implementing services using emotional analysis in small systems rather than large-scale systems, and there are not many cases where they are actually serviced. However, if even a small system can easily analyze consumer trends, it will help the rapidly changing modern society. In this paper, we propose a lightweight trend analysis system that builds a learning network through Transfer Learning (Fine Tuning) of the BERT Model and interlocks Crawler for real-time data collection.

• KIPS Transactions on Computer and Communication Systems
• /
• v.6 no.3
• /
• pp.159-168
• /
• 2017

Smartphone malware has increased because Smartphone users has increased and smartphones are widely used in everyday life. Since 2012, Android has been the most mobile operating system. Owing to the open nature of Android, countless malware are in Android markets that seriously threaten Android security. Most of Android malware detection program does not detect malware to which bypass techniques apply and also does not detect unknown malware. In this paper, we propose lightweight method for detection of Android malware using static analysis and deep learning techniques. For experiments we crawl 7,000 apps from the Google Play Store and collect 6,120 malwares. The result show that proposed method can achieve 98.05% detection accuracy. Also, proposed method can detect about unknown malware families with good performance. On smartphones, the method requires 10 seconds for an analysis on average.

• Journal of IKEEE
• /
• v.26 no.3
• /
• pp.364-372
• /
• 2022

This paper proposes the design and implementation results for human and object classification systems utilizing frequency modulated continuous wave (FMCW) radar sensor. Such a system requires the process of radar sensor signal processing for multi-target detection and the process of deep learning for the classification of human and object. Since deep learning requires such a great amount of computation and data processing, the lightweight process is utmost essential. Therefore, binary neural network (BNN) structure was adopted, operating convolution neural network (CNN) computation in a binary condition. In addition, for the real-time operation, a hardware accelerator was implemented and verified via FPGA platform. Based on performance evaluation and verified results, it is confirmed that the accuracy for multi-target classification of 90.5%, reduced memory usage by 96.87% compared to CNN and the run time of 5ms are achieved.