• Journal of Ubiquitous Convergence Technology
• /
• v.1 no.1
• /
• pp.18-22
• /
• 2007

Multi-Instance Learning(MIL) performs well to deal with inherently ambiguity of images in multimedia retrieval. In this paper, an effective framework for Contented-Based Image Retrieval(CBIR) with MIL techniques is proposed, the effective mechanism is based on the image segmentation employing improved Mean Shift algorithm, and processes the segmentation results utilizing mathematical morphology, where the goal is to detect the semantic concepts contained in the query. Every sub-image detected is represented as a multiple features vector which is regarded as an instance. Each image is produced to a bag comprised of a flexible number of instances. And we apply a few number of MIL algorithms in this framework to perform the retrieval. Extensive experimental results illustrate the excellent performance in comparison with the existing methods of CBIR with MIL.

• Annual Conference on Human and Language Technology
• /
• 2012.10a
• /
• pp.47-52
• /
• 2012

본 논문에서는 소셜 네트워크 서비스에서 사용자가 작성한 텍스트로부터 그 사용자 프로파일 식별하는 문제를 다룬다. 프로파일 식별 관련 기존 연구에서는 개별 텍스트를 하나의 학습 단위로 간주하고 이를 기반으로 학습 모델을 구축한다. 프로파일을 식별하고자 하는 사용자의 텍스트들이 주어지면 각 텍스트마다 프로파일을 식별하고, 식별된 결과들을 합쳐 최종 프로파일로 선택한다. 하지만 SNS 특성상 프로파일을 식별하는 데에 영향을 끼치지 않는 텍스트들이 다수 존재하며, 기존 연구들은 이 텍스트들을 특별한 처리없이 학습 및 테스트에 사용함으로 인해 프로파일 식별 성능이 저하되는 문제점이 있다. 본 논문에서는 다중 인스턴스 학습(Multi-Instance Learning)을 기반으로 사용자 프로파일을 식별한다. 제안한 방법은 사용자가 작성한 텍스트 전체, 즉 텍스트 집합을 학습 단위로 간주하고 다중 인스턴스 학습 문제로 변환하여 프로파일을 식별한다. 다중 인스턴스 학습을 사용함으로써 프로파일 식별에 유의미한 텍스트들만이 고려되고 그 결과 프로파일 식별에 영향을 끼치지 않는 텍스트로부터의 성능 하락을 최소화할 수 있다. 실험을 통해 제안한 방법이 기존 학습 방법보다 성별, 나이, 결혼/연애 상태를 식별함에 있어서 더 좋은 성능을 보인다.

• Journal of KIISE
• /
• v.42 no.2
• /
• pp.227-234
• /
• 2015

This paper proposes a modified MI-SVM algorithm by considering data distribution. The previous MI-SVM algorithm seeks the margin by considering the "most positive" instance in a positive bag. Positive instances included in positive bags are located in a similar area in a feature space. In order to reflect this characteristic of positive instances, the proposed method selects the "most positive" instance by calculating the distance between each instance in the bag and a pivot point that is the intersection point of all positive instances. This paper suggests two ways to select the "most positive" pivot point in the training data. First, the algorithm seeks the "most positive" pivot point along the current predicted parameter, and then selects the nearest instance in the bag as a representative from the pivot point. Second, the algorithm finds the "most positive" pivot point by using a Diverse Density framework. Our experiments on 12 benchmark multi-instance data sets show that the proposed method results in higher performance than the previous MI-SVM algorithm.

• KSII Transactions on Internet and Information Systems (TIIS)
• /
• v.12 no.6
• /
• pp.2895-2921
• /
• 2018

Security has become one of the major concerns in mobile adhoc networks (MANETs). Data and voice communication amongst roaming battlefield entities (such as platoon of soldiers, inter-battlefield tanks and military aircrafts) served by MANETs throw several challenges. It requires complex securing strategy to address threats such as unauthorized network access, man in the middle attacks, denial of service etc., to provide highly reliable communication amongst the nodes. Intrusion Detection and Prevention System (IDPS) undoubtedly is a crucial ingredient to address these threats. IDPS in MANET is managed by Command Control Communication and Intelligence (C3I) system. It consists of networked computers in the tactical battle area that facilitates comprehensive situation awareness by the commanders for timely and optimum decision-making. Key issue in such IDPS mechanism is lack of Smart Learning Engine. We propose a novel behavioral based "Smart Multi-Instance Multi-Label Intrusion Detection and Prevention System (MIML-IDPS)" that follows a distributed and centralized architecture to support a Robust C3I System. This protocol is deployed in a virtually clustered non-uniform network topology with dynamic election of several virtual head nodes acting as a client Intrusion Detection agent connected to a centralized server IDPS located at Command and Control Center. Distributed virtual client nodes serve as the intelligent decision processing unit and centralized IDPS server act as a Smart MIML decision making unit. Simulation and experimental analysis shows the proposed protocol exhibits computational intelligence with counter attacks, efficient memory utilization, classification accuracy and decision convergence in securing C3I System in a Tactical Battlefield environment.

• IEMEK Journal of Embedded Systems and Applications
• /
• v.16 no.5
• /
• pp.209-214
• /
• 2021

Panoptic segmentation, which is now widely used in computer vision such as medical image analysis, and autonomous driving, helps understanding an image with holistic view. It identifies each pixel by assigning a unique class ID, and an instance ID. Specifically, it can classify 'thing' from 'stuff', and provide pixel-wise results of semantic prediction and object detection. As a result, it can solve both semantic segmentation and instance segmentation tasks through a unified single model, producing two different contexts for two segmentation tasks. Semantic segmentation task focuses on how to obtain multi-scale features from large receptive field, without losing low-level features. On the other hand, instance segmentation task focuses on how to separate 'thing' from 'stuff' and how to produce the representation of detected objects. With the advances of both segmentation techniques, several panoptic segmentation models have been proposed. Many researchers try to solve discrepancy problems between results of two segmentation branches that can be caused on the boundary of the object. In this survey paper, we will introduce the concept of panoptic segmentation, categorize the existing method into two representative methods and explain how it is operated on two methods: top-down method and bottom-up method. Then, we will analyze the performance of various methods with experimental results.

• Proceedings of the Korean Society of Broadcast Engineers Conference
• /
• 2022.11a
• /
• pp.171-174
• /
• 2022

Weakly-supervised learning is a widely adopted approach in video anomaly detection whereby only video labels are utilized instead of expensive frame-level annotations. Since the success of multi-instance learning (MIL), almost all recent approaches are based on maximizing the margin between the set of abnormal video snippets and those of normal video snippets. In this work, we present a simple contrastive approach for weakly supervised video anomaly detection (WS-VAD) with aims to enhance the performance of existing models. The method is generic in nature and introduces a loss function to encourage attraction of output features from the same video class and repel those from different video classes. Experimental results demonstrate our method can be applied to existing algorithms to improve detection accuracy in public video anomaly dataset.

• Journal of Korea Multimedia Society
• /
• v.10 no.11
• /
• pp.1516-1522
• /
• 2007

We had proposed the FPA(Fixed Partition Averaging) method in order to improve the storage requirement and classification rate of the Memory Based Reasoning. The algorithm worked not bad in many area, but it lead to some overhead for memory usage and lengthy computation in the multi classes area. We propose an Repetitive FPA algorithm which repetitively partitioning pattern space in the multi classes area. Our proposed methods have been successfully shown to exhibit comparable performance to k-NN with a lot less number of patterns and better result than EACH system which implements the NGE theory.

• Journal of Korea Multimedia Society
• /
• v.22 no.11
• /
• pp.1251-1258
• /
• 2019

Intelligent tutoring system enables users to effectively learn by utilizing various artificial intelligence techniques. For instance, it can recommend a proper curriculum or learning method to individual users based on their learning history. To do this effectively, user's characteristics need to be analyzed and classified based on various aspects such as interest, learning ability, and personality. Even though data labeled by the characteristics are required for more accurate classification, it is not easy to acquire enough amount of labeled data due to the labeling cost. On the other hand, unlabeled data should not need labeling process to make a large number of unlabeled data be collected and utilized. In this paper, we propose a semi-supervised learning method based on feedback variational auto-encoder(FVAE), which uses both labeled data and unlabeled data. FVAE is a variation of variational auto-encoder(VAE), where a multi-layer perceptron is added for giving feedback. Using unlabeled data, we train FVAE and fetch the encoder of FVAE. And then, we extract features from labeled data by using the encoder and train classifiers with the extracted features. In the experiments, we proved that FVAE-based semi-supervised learning was superior to VAE-based method in terms with accuracy and F1 score.

• Convergence Security Journal
• /
• v.22 no.2
• /
• pp.21-26
• /
• 2022

Wearing face masks is an effective measure to prevent COVID-19 infection. Infrared thermal image based temperature measurement and identity recognition system has been widely used in many large enterprises and universities in China, so it is totally necessary to research the face mask detection of thermal infrared imaging. Recently introduced MTCNN (Multi-task Cascaded Convolutional Networks)presents a conceptually simple, flexible, general framework for instance segmentation of objects. In this paper, we propose an algorithm for efficiently searching objects of images, while creating a segmentation of heat generation part for an instance which is a heating element in a heat sensed image acquired from a thermal infrared camera. This method called a mask MTCNN is an algorithm that extends MTCNN by adding a branch for predicting an object mask in parallel with an existing branch for recognition of a bounding box. It is easy to generalize the R-CNN to other tasks. In this paper, we proposed an infrared image detection algorithm based on R-CNN and detect heating elements which can not be distinguished by RGB images.

• Journal of the Korea Institute of Information and Communication Engineering
• /
• v.26 no.3
• /
• pp.367-373
• /
• 2022

As deep learning-based autonomous driving technology develops, artificial intelligence models for various purposes have been studied. Based on these studies, several models were used simultaneously to develop autonomous driving systems. It can occur by increasing hardware resource consumption. We propose a multi-tasks model using a shared backbone to solve this problem. This can solve the increase in the number of backbones for using AI models. As a result, in the proposed lightweight model, the model parameters could be reduced by more than 50% compared to the existing model, and the speed could be improved. In addition, each lane can be classified through lane detection using the instance segmentation method. However, further research is needed on the decrease in accuracy compared to the existing model.