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.

Solving Continual Combinatorial Selection via Deep Reinforcement Learning

Abstract: We consider the Markov Decision Process (MDP) of selecting a subset of items at each step, termed the Select-MDP (S-MDP). The large state and action spaces of S-MDPs make them intractable to solve with typical reinforcement learning (RL) algorithms especially when the number of items is huge. In this paper, we present a deep RL algorithm to solve this issue by adopting the following key ideas. First, we convert the original S-MDP into an Iterative Select-MDP (IS-MDP), which is equivalent to the S-MDP in terms of optimal actions. IS-MDP decomposes a joint action of selecting K items simultaneously into K iterative selections resulting in the decrease of actions at the expense of an exponential increase of states. Second, we overcome this state space explo-sion by exploiting a special symmetry in IS-MDPs with novel weight shared Q-networks, which prov-ably maintain sufficient expressive power. Various experiments demonstrate that our approach works well even when the item space is large and that it scales to environments with item spaces different from those used in training.

Bilingual Lexicon Extraction via Pivot Language and Word Alignment Tool

Abstract: This paper presents a simple and effective method for automatic bilingual lexicon extraction from less-known language pairs. To do this, we bring in a bridge language named the pivot language and adopt information retrieval techniques combined with natural language processing techniques. Moreover, we use a freely available word aligner: Anymalign (Lardilleux et al., 2011) for constructing context vectors. Unlike the previous works, we obtain context vectors via a pivot language. Therefore, we do not require to translate context vectors by using a seed dictionary and improve the accuracy of low frequency word alignments that is weakness of statistical model by using Anymalign. In this paper, experiments have been conducted on two different language pairs that are bi-directional KoreanSpanish and Korean-French, respectively. The experimental results have demonstrated that our method for high-frequency words shows at least 76.3 and up to 87.2% and for the lowfrequency words at least 43.3% and up to 48.9% within the top 20 ranking candidates, respectively.

Development of Air Force Winter Service Uniform Shirt Pattern and Automatic Pattern Drafting Program for MTM Production

Abstract: This study improves the fitness of Air Force winter service uniforms through the development of a shirt pattern drafting method and automatic pattern drafting program for MTM production. A calculation formula is formed through a correlation analysis and regression analysis using Size Korea 2004 3D measurement data after analyzing 4 kinds of existing shirt pattern drafting methods and 3 types of shirt patterns currently used for the Air Force service uniform. The results of this study are as follows: The developed pattern drafting method has 4 parts that use calculated dimensions: neck base width, front interscye, back interscye and scye depth. Other body measuring parts that have a high correlation with calculation parts are inserted into regression analysis as independent variables to create dimension calculation formulas. The result of the final study patterns were better than existing winter service uniforms in nearly all items for the appearance evaluation and motion adaptability evaluations. The method was converted into an automatic pattern drafting program using C++ after the completion of pattern drafting method development.

Complete atrioventricular block-induced Torsade de pointes, manifested by epilepsy.

Abstract: Complete atrioventricular (AV) block is frequently regarded as a cause of informed syncopal attacks, even though the escape rhythm is maintained. Torsade de pointes (TdP) may be a significant complication of AV block associated with QT prolongation. Here, we report the case of a 42-year-old female who was referred to our hospital due to recurrent seizure-like attacks while taking anti-convulsant drugs at a psychiatric hospital. TdP with a long QT interval (corrected QT = 0.591 seconds) was observed on an electrocardiogram (ECG) taken in the emergency department. The patient's drug history revealed olanzapine as the suspicious agent. Even after the medication was stopped, however, the QT interval remained within an abnormal range and multiple episodes of TdP and related seizure-like symptoms were found via ECG monitoring. A permanent pacemaker was thus implanted, and the ventricular rate was set at over 80 beats/min. There was no recurrence of tachyarrhythmia or other symptoms.

Pororobot: A Deep Learning Robot That Plays Video Q&A Games

Abstract: Recent progress in machine learning has lead to great advancements in robot intelligence and human-robot interaction (HRI). It is reported that robots can deeply understand visual scene information and describe the scenes in natural language using object recognition and natural language processing methods. Imagebased question and answering (Q&A) systems can be used for enhancing HRI. However, despite these successful results, several key issues still remain to be discussed and improved. In particular, it is essential for an agent to act in a dynamic, uncertain, and asynchronous environment for achieving human-level robot intelligence. In this paper, we propose a prototype system for a video Q&A robot “Pororobot”. The system uses the state-of-the-art machine learning methods such as a deep concept hierarchy model. In our scenario, a robot and a child plays a video Q&A game together under real world environments. Here we demonstrate preliminary results of the proposed system and discuss some directions as future works.