• KSII Transactions on Internet and Information Systems (TIIS)
• /
• v.18 no.7
• /
• pp.1749-1773
• /
• 2024

Recent advances in deep neural networks (DNNs) have greatly improved the accuracy and universality of various intelligent applications, at the expense of increasing model size and computational demand. Since the resources of end devices are often too limited to deploy a complete DNN model, offloading DNN inference tasks to cloud servers is a common approach to meet this gap. However, due to the limited bandwidth of WAN and the long distance between end devices and cloud servers, this approach may lead to significant data transmission latency. Therefore, device-edge collaborative inference has emerged as a promising paradigm to accelerate the execution of DNN inference tasks where DNN models are partitioned to be sequentially executed in both end devices and edge servers. Nevertheless, collaborative inference in heterogeneous edge environments with multiple edge servers, end devices and DNN tasks has been overlooked in previous research. To fill this gap, we investigate the optimization problem of collaborative inference in a heterogeneous system and propose a scheme CIS, i.e., collaborative inference scheme, which jointly combines DNN partition, task offloading and scheduling to reduce the average weighted inference latency. CIS decomposes the problem into three parts to achieve the optimal average weighted inference latency. In addition, we build a prototype that implements CIS and conducts extensive experiments to demonstrate the scheme's effectiveness and efficiency. Experiments show that CIS reduces 29% to 71% on the average weighted inference latency compared to the other four existing schemes.

• Journal of KIISE:Software and Applications
• /
• v.37 no.9
• /
• pp.659-668
• /
• 2010

Collaborative filtering systems grasp behaviors for a new user and need new information for the user in order to recommend interesting items to the user. For the purpose of acquiring the information the collaborative filtering systems learn behaviors for users based on the previous data and can obtain new information from the results. In this paper, we propose an inductive inference method to obtain new information for users and rank items by using the new information in the proposed method. The proposed method clusters users into groups by learning users through NMF among inductive machine learning methods and selects the group features from the groups by using chi-square. Then, the method classifies a new user into a group by using the bayesian probability model as one of inductive inference methods based on the rating values for the new user and the features of groups. Finally, the method decides the ranks of items by applying the Rocchio algorithm to items with the missing values.

• KSII Transactions on Internet and Information Systems (TIIS)
• /
• v.18 no.5
• /
• pp.1223-1237
• /
• 2024

Autonomous vehicles use onboard sensors to sense the surrounding environment. In complex autonomous driving scenarios, the detection and recognition capabilities are constrained, which may result in serious accidents. An efficient way to enhance the detection and recognition capabilities is establishing collaborations with the neighbor vehicles. However, the collaborations introduce additional challenges in terms of the data heterogeneity, communication cost, and data privacy. In this paper, a novel personalized federated learning framework is proposed for addressing the challenges and enabling efficient collaborations in autonomous driving environment. For obtaining a global model, vehicles perform local training and transmit logits to a central unit instead of the entire model, and thus the communication cost is minimized, and the data privacy is protected. Then, the inference similarity is derived for capturing the characteristics of data heterogeneity. The vehicles are divided into clusters based on the inference similarity and a weighted aggregation is performed within a cluster. Finally, the vehicles download the corresponding aggregated global model and train a personalized model which is personalized for the cluster that has similar data distribution, so that accuracy is not affected by heterogeneous data. Experimental results demonstrate significant advantages of our proposed method in improving the efficiency of collaborative perception and reducing communication cost.

• Korean Journal of Computational Design and Engineering
• /
• v.11 no.1
• /
• pp.57-66
• /
• 2006

Modeling product configurations needs large amounts of knowledge about technical and marketing restrictions on the product. Previous attempts to automate product configurations concentrate on representations and management of the knowledge for specific domains in fixed and isolated computing environments. Since the knowledge about product configurations is subject to continuous change and hard to express, these attempts often failed to efficiently manage and exchange the knowledge in collaborative product development. In this paper, XML Topic Map (XTM) is introduced to represent and exchange the knowledge about product configurations in collaborative product development. A product configuration model based on XTM along with its merger and inference facilities enables configuration engineers In collaborative product development to manage and exchange their knowledge efficiently. An implementation of the proposed product configuration model is presented to demonstrate that the proposed approach enables enterprises to exchange the knowledge about product configurations during their collaborative product development.

• Proceedings of the IEEK Conference
• /
• 2009.05a
• /
• pp.384-386
• /
• 2009

This paper introduces an effectively optimized Fuzzy model identification by means of complex and nonlinear system applying PSO algorithm. In other words, we use PSO(Particle Swarm Optimization) for identification of Fuzzy model structure and parameter. PSO is an algorithm that follows a collaborative population-based search model. Each particle of swarm flies around in a multidimensional search space looking for the optimal solution. Then, Particles adjust their position according to their own and their neighboring-particles experience. This paper identifies the premise part parameters and the consequence structures that have many effects on Fuzzy system based on PSO. In the premise parts of the rules, we use triangular. Finally we evaluate the Fuzzy model that is widely used in the standard model of gas data and sew data.

• Annual Conference of KIPS
• /
• 2022.05a
• /
• pp.264-267
• /
• 2022

Recently, federated learning (FL) has increased prominence as a viable approach for enhancing user privacy and data security by allowing collaborative multi-party model learning without exchanging sensitive data. Despite this, most present FL systems still depend on a centralized aggregator to generate a global model by gathering all submitted models from users, which could expose user privacy and the risk of various threats from malicious users. To solve these issues, we suggested a safe FL framework that employs differential privacy to counter membership inference attacks during the collaborative FL model training process and empowers blockchain to replace the centralized aggregator server.

• KSII Transactions on Internet and Information Systems (TIIS)
• /
• v.8 no.2
• /
• pp.483-503
• /
• 2014

In this paper, human action recognition using pyramid histograms of oriented gradients and collaborative multi-task learning is proposed. First, we accumulate global activities and construct motion history image (MHI) for both RGB and depth channels respectively to encode the dynamics of one action in different modalities, and then different action descriptors are extracted from depth and RGB MHI to represent global textual and structural characteristics of these actions. Specially, average value in hierarchical block, GIST and pyramid histograms of oriented gradients descriptors are employed to represent human motion. To demonstrate the superiority of the proposed method, we evaluate them by KNN, SVM with linear and RBF kernels, SRC and CRC models on DHA dataset, the well-known dataset for human action recognition. Large scale experimental results show our descriptors are robust, stable and efficient, and outperform the state-of-the-art methods. In addition, we investigate the performance of our descriptors further by combining these descriptors on DHA dataset, and observe that the performances of combined descriptors are much better than just using only sole descriptor. With multimodal features, we also propose a collaborative multi-task learning method for model learning and inference based on transfer learning theory. The main contributions lie in four aspects: 1) the proposed encoding the scheme can filter the stationary part of human body and reduce noise interference; 2) different kind of features and models are assessed, and the neighbor gradients information and pyramid layers are very helpful for representing these actions; 3) The proposed model can fuse the features from different modalities regardless of the sensor types, the ranges of the value, and the dimensions of different features; 4) The latent common knowledge among different modalities can be discovered by transfer learning to boost the performance.

• Proceedings of the Korean Society of Precision Engineering Conference
• /
• 2005.10a
• /
• pp.440-443
• /
• 2005

The wide spread of internet business recently necessitates recommendation systems which can recommend the most suitable product fur customer demands. Currently the recommendation systems use content-based filtering and/or collaborative filtering methods, which are unable both to explain the reason for the recommendation and to reflect constantly changing requirements of the users. These methods guarantee good efficiency only if there is a lot of information about users. This paper proposes an algorithm called 'demand articulate & integration' which can perceive user's continuously varying intents and recommend proper contents. A method of knowledge classification which can be applicable to this algorithm is also developed in order to disassemble knowledge into basic units and articulate indices. The algorithm provides recommendation outputs that are close to expert's opinion through the tracing of articulate index. As a case study, a knowledge base for heritage information is constructed with the expert guide's knowledge. An intelligent recommendation system that can guide heritage tour as good as the expert guider is developed.

• Journal of KIISE:Software and Applications
• /
• v.31 no.8
• /
• pp.1054-1061
• /
• 2004

At recent, people can easily access to information by Internet to be rapidly evolving. And also, the amount is rapidly increasing. So the techniques, to automatically extract the required information are very important to reduce the time and the effort for retrieving information. In this paper, we describe a collaborative filtering system for automatically recommending high-quality information to users with similar interests on arbitrarily narrow information domains. It asks a user to rate a gauge set of items. It then evaluates the user's rates and suggests a recommendation set of items. We interpret the process of evaluation as an inference mechanism that maps a gauge set to a recommendation set. We accomplish the mapping with FAM (Fuzzy Associative Memory). We implemented the suggested system in a Web server and tested its performance in the domain of retrieval of technical papers, especially in the field of information technologies. The experimental results show that it may provide reliable recommendations.

• Journal of Korean Institute of Industrial Engineers
• /
• v.37 no.4
• /
• pp.323-330
• /
• 2011

Globalization of market and diversification of customers' needs make enterprises to collaboration of participants in supply chain. To establish collaboration, supply chain must have the flexibility and reconfigurability, which are supported by fractal based supply chain management (FrSCM). In this paper, base on the FrSCM, formulation of trust model among the enterprises in the supply chain, and development of profit sharing strategies in the supply chain based on the trust model are investigated. To evaluate trust model, generation of enterprise's goal and its description, extraction and systematic composition of trust factors and trust evaluation are investigated. Based on the developed model, we developed the fuzzy inference engine to evaluate the trust value in terms of numerical value. And then revenue sharing strategies are developed based on the fractal concept and trust model for the collaborative SCM. The fractal concept is used to obtain the optimal production and transportation plans. In addition, the trust model will be integrated into the RS model. In such an RS model, the supply chain will obtain the maximum total profit and profit of each participant depends on its trust value.