Naver Corporation

About: Naver Corporation is a company organization based out in Seongnam-si, South Korea. It is known for research contribution in the topics: Terminal (electronics) & Computer science. The organization has 4038 authors who have published 4294 publications receiving 35045 citations. The organization is also known as: NAVER Corporation & NAVER.

Research on an Authentication Algorithm for an Electronic Attendance System in the Constructing of a Smart Campus

Abstract: This paper presents a proposal for a large-scale lecture authentication algorithm for an electronic attendance system The utilization rate for existing electronic attendance systems is low Because it couldn't solve the problems that occur by special conditions of university such as vicarious attendance, bottlenecking during attendance check-in, and attendees departure at intermission The most important things are not the authentication problems or errors on the technical issues Instead, issues are related to attempts to check attendance simultaneously by a number of students, failure to prevent vicarious attendance, and the uncertainties of students remaining in the classroom following check-in The authentication algorithm of the proposed system uses smartphone in attendance check So it will save time and prevent bottlenecks as large numbers of students will not need to stand in line to check-in their attendance in the wall-mounted attendance device And, it will minimize, if not eliminate, students who might leave at intermission Because the instructor can verify attendance many times with the simplicity of the attendance check verification procedure and the short authentication time Thus, the proposed algorithm will facilitate the implementation of a smart electronic attendance system in the construction of the smart campus

Community providing system, community providing method and computer readable storage medium storing execution program for that method

Abstract: Disclosed are a community providing system, a community providing method, and a computer-readable recording medium for storing a program capable of carrying out the method. In one embodiment, community configuration items including at least one of a template and a background image of a community are provided, and corresponding URL information is sorted and stored for each user when the user specifies a desired community configuration item. A community environment to which the requested community configuration item is applied is provided after receiving the URL information on the requested community configuration item according to the request made by the user. Therefore, system usage may be increased because of the provision of various community environments where the user can show his personality through the establishment of various environments.

Personal icon providing system and method thereof

Abstract: Disclosed is a personal icon providing system and method. When a user is accessed through a network, information on personal icons provided to the user is collected. The information on the collected personal icons is transmitted to the user's terminal to be stored therein, and the user's personal icon is displayed on the network by using the information stored in the terminal. In this instance, the personal icon may have a single-layer structure. According to the present invention, the user's icon can be easily displayed at any position on the network, and in particular, a personal icon for identifying the user in cooperation with the user's movement can be quickly and easily provided.

Graph Attention Networks for Speaker Verification

Abstract: This work presents a novel back-end framework for speaker verification using graph attention networks. Segment-wise speaker embeddings extracted from multiple crops within an utterance are interpreted as node representations of a graph. The proposed framework inputs segment-wise speaker embeddings from an enrollment and a test utterance and directly outputs a similarity score. We first construct a graph using segment-wise speaker embeddings and then input these to graph attention networks. After a few graph attention layers with residual connections, each node is projected into a one-dimensional space using affine transform, followed by a readout operation resulting in a scalar similarity score. To enable successful adaptation for speaker verification, we propose techniques such as separating trainable weights for attention map calculations between segment-wise speaker embeddings from different utterances. The effectiveness of the proposed framework is validated using three different speaker embedding extractors trained with different architectures and objective functions. Experimental results demonstrate consistent improvement over various baseline back-end classifiers, with an average equal error rate improvement of 20% over the cosine similarity back-end without test time augmentation.