Institution
Naver Corporation
Company•Seongnam-si, South Korea•
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.
Papers published on a yearly basis
Papers
More filters
•
TL;DR: This paper proposed a data augmentation technique that leverages large-scale language models to generate realistic text samples from a mixture of real samples, effectively distilling knowledge from the language models and creating textual perturbations simultaneously.
Abstract: Large-scale language models such as GPT-3 are excellent few-shot learners, allowing them to be controlled via natural text prompts. Recent studies report that prompt-based direct classification eliminates the need for fine-tuning but lacks data and inference scalability. This paper proposes a novel data augmentation technique that leverages large-scale language models to generate realistic text samples from a mixture of real samples. We also propose utilizing soft-labels predicted by the language models, effectively distilling knowledge from the large-scale language models and creating textual perturbations simultaneously. We perform data augmentation experiments on diverse classification tasks and show that our method hugely outperforms existing text augmentation methods. Ablation studies and a qualitative analysis provide more insights into our approach.
12 citations
••
17 May 2020TL;DR: The authors transfer the knowledge from a concrete Transformer-based text LMs to an end-to-end spoken language understanding (SLU) module which can face a data shortage, based on recent cross-modal distillation methodologies.
Abstract: Speech is one of the most effective means of communication and is full of information that helps the transmission of utterer's thoughts. However, mainly due to the cumbersome processing of acoustic features, phoneme or word posterior probability has frequently been discarded in understanding the natural language. Thus, some recent spoken language understanding (SLU) modules have utilized end-to-end structures that preserve the uncertainty information. This further reduces the propagation of speech recognition error and guarantees computational efficiency. We claim that in this process, the speech comprehension can benefit from the inference of massive pre-trained language models (LMs). We transfer the knowledge from a concrete Transformer-based text LM to an SLU module which can face a data shortage, based on recent cross-modal distillation methodologies. We demonstrate the validity of our proposal upon the performance on Fluent Speech Command, an English SLU benchmark. Thereby, we experimentally verify our hypothesis that the knowledge could be shared from the top layer of the LM to a fully speech-based module, in which the abstracted speech is expected to meet the semantic representation.
12 citations
•
27 Feb 2013TL;DR: In this paper, a chat room page is displayed in response to selection of at least one of the listed status messages, chat room pages being configured to facilitate chatting, the chat rooms page being displayed including the at least selected status message automatically inserted into an input area of the chatroom page.
Abstract: An approach is provided to facilitate chatting services. A friend list page is displayed via a user terminal, the friend list page including an area configured to present a status message associated with a friend. A message page is displayed in response to selection of the status message, the message page including a list of status messages associated with the friend. A chat room page is displayed in response to selection of at least one of the listed status messages, the chat room page being configured to facilitate chatting, the chat room page being displayed including the at least one selected status message automatically inserted into an input area of the chat room page.
12 citations
•
17 Nov 2009TL;DR: In this article, the authors present a system and method for production of a multi-user network game that may produce and debug a multiuser network game and simply construct a multi user network game environment using a single game production tool to reduce a game production time.
Abstract: Provided is a system and method for production of a multi-user network game that may produce and debug a multi-user network game and simply construct a multi-user network game environment using a single game production tool to thereby reduce a game production time. A system for production of a multi-user network game being performed between a game server and a plurality of game clients via a network, may include: a game production module configured to produce and debug the multi-user network game and a multi-user network execution environment; and an emulation module configured to emulate an execution of the multi-user network game by constructing a virtual network execution environment that comprises at least one server virtual machine and at least one client virtual machine configured to execute the produced or debugged multi-user network game.
12 citations
•
25 Mar 2004TL;DR: Disclosed as mentioned in this paper is a method and system for managing web sites registered in a search engine that provides information about web sites on the Internet, wherein information about the Web sites registered by the search engine is analyzed to prevent the provision of search results different from essential contents contained in the web sites.
Abstract: Disclosed is a method and system for managing web sites registered in a search engine that provides information about web sites on the Internet, wherein information about the web sites registered in the search engine is analyzed to prevent the provision of search results different from essential contents contained in the web sites. In the method, information of the registered web site is received and recorded in a database after being classified by predetermined fields. A search robot is controlled to read a source file constituting a web page of the registered web site, and the read source file is then analyzed. It is determined based on a predetermined basis whether or not the registered web site is a deceptive site. Predetermined processing is performed on the registered web site if the web site is determined to be a deceptive site. The source file is preferably an HTML document.
12 citations
Authors
Showing all 4041 results
Name | H-index | Papers | Citations |
---|---|---|---|
Andrea Vedaldi | 89 | 305 | 63305 |
Sunghun Kim | 51 | 115 | 12994 |
Eric Gaussier | 41 | 231 | 8203 |
Un Ju Jung | 39 | 98 | 5696 |
Hyun-Soo Kim | 37 | 421 | 5650 |
Gabriela Csurka | 37 | 145 | 10959 |
Nojun Kwak | 34 | 234 | 6026 |
Young-Jin Park | 31 | 257 | 3759 |
Sung Joo Kim | 31 | 196 | 3078 |
Jae-Hoon Kim | 30 | 323 | 5847 |
Jung-Ryul Lee | 29 | 222 | 3322 |
Joon Son Chung | 28 | 73 | 4900 |
Ok-Hwan Lee | 27 | 163 | 2896 |
Diane Larlus | 27 | 69 | 4722 |
Jung Goo Lee | 26 | 142 | 1917 |