Author
Jang-Won Park
Bio: Jang-Won Park is an academic researcher from Samsung. The author has contributed to research in topics: Communications system & Telecommunications link. The author has an hindex of 6, co-authored 15 publications receiving 130 citations.
Papers
More filters
••
01 Nov 2019TL;DR: An adversarial training framework for domain generalization in Question Answering (QA) task is utilized, where the two models constantly compete, so that QA model can learn domain-invariant features.
Abstract: Adapting models to new domain without finetuning is a challenging problem in deep learning. In this paper, we utilize an adversarial training framework for domain generalization in Question Answering (QA) task. Our model consists of a conventional QA model and a discriminator. The training is performed in the adversarial manner, where the two models constantly compete, so that QA model can learn domain-invariant features. We apply this approach in MRQA Shared Task 2019 and show better performance compared to the baseline model.
57 citations
•
19 Jul 2006TL;DR: In this article, an uplink scheduling system and method in a wireless broadband internet communication system includes calculating priorities of mobile stations (MSs) taking into account headroom size indicating transmission power additionally available for each of the MSs; and determining a transmission format of an MS selected to be preferentially allocated a slot as a result of the priority calculation, with a margin determined depending on a variation in channel state between the selected MS and a base station (BS).
Abstract: An uplink scheduling system and method in a wireless broadband internet communication system includes calculating priorities of mobile stations (MSs) taking into account headroom size indicating transmission power additionally available for each of the MSs; and determining a transmission format of an MS selected to be preferentially allocated a slot as a result of the priority calculation, taking into account headroom size reported by the selected MS and a margin determined depending on a variation in channel state between the selected MS and a base station (BS).
34 citations
•
03 Jan 2007TL;DR: In this paper, the authors proposed a scheduling method in a communication system including a plurality of mobile stations (MSs) and a base station (BS) for providing a communication service to the MSs.
Abstract: Disclosed is a scheduling method in a communication system including a plurality of mobile stations (MSs) and a base station (BS) for providing a communication service to the MSs. The scheduling method includes determining candidate transmission formats of the MSs according to channel status information fed back from the MSs and levels of transmission power of the MSs; and calculating priorities of the determined candidated transmission formats, and determining a transmission format having the highest priority among the candidated transmission formats, as a transmission format for each of the MSs.
16 citations
•
05 Feb 2008TL;DR: In this paper, an apparatus and method for uplink scheduling in a broadband wireless communication system are provided, in which channel conditions are divided into at least two areas, an area corresponding to channel information of a CSM mobile station received from the CSM-mobile station is determined, resources of a first pilot pattern are allocated to the mobile station, if the channel information belongs to an area with good channel condition, and if the cannel information belonging to an areas with poor channel condition.
Abstract: An apparatus and method for uplink scheduling in a broadband wireless communication system are provided, in which channel conditions are divided into at least two areas, an area corresponding to channel information of a CSM mobile station received from the CSM mobile station is determined, resources of a first pilot pattern are allocated to the CSM mobile station, if the channel information belongs to an area with good channel condition, and resources of a second pilot pattern are allocated to the CSM mobile station, if the cannel information belongs to an area with poor channel condition.
16 citations
•
TL;DR: The Korean Language Understanding Evaluation (KLUE) benchmark as mentioned in this paper is a collection of 8 Korean NLP tasks, including Topic Classification, SemanticTextual Similarity, Natural Language Inference, Named Entity Recognition, Relation Extraction, Dependency Parsing, Machine Reading Comprehension, and Dialogue State Tracking.
Abstract: We introduce Korean Language Understanding Evaluation (KLUE) benchmark. KLUE is a collection of 8 Korean natural language understanding (NLU) tasks, including Topic Classification, SemanticTextual Similarity, Natural Language Inference, Named Entity Recognition, Relation Extraction, Dependency Parsing, Machine Reading Comprehension, and Dialogue State Tracking. We build all of the tasks from scratch from diverse source corpora while respecting copyrights, to ensure accessibility for anyone without any restrictions. With ethical considerations in mind, we carefully design annotation protocols. Along with the benchmark tasks and data, we provide suitable evaluation metrics and fine-tuning recipes for pretrained language models for each task. We furthermore release the pretrained language models (PLM), KLUE-BERT and KLUE-RoBERTa, to help reproducing baseline models on KLUE and thereby facilitate future research. We make a few interesting observations from the preliminary experiments using the proposed KLUE benchmark suite, already demonstrating the usefulness of this new benchmark suite. First, we find KLUE-RoBERTa-large outperforms other baselines, including multilingual PLMs and existing open-source Korean PLMs. Second, we see minimal degradation in performance even when we replace personally identifiable information from the pretraining corpus, suggesting that privacy and NLU capability are not at odds with each other. Lastly, we find that using BPE tokenization in combination with morpheme-level pre-tokenization is effective in tasks involving morpheme-level tagging, detection and generation. In addition to accelerating Korean NLP research, our comprehensive documentation on creating KLUE will facilitate creating similar resources for other languages in the future. KLUE is available at this https URL.
7 citations
Cited by
More filters
•
26 Feb 2008TL;DR: In this paper, a base station apparatus for performing time and frequency scheduling in uplink packet access with an interference amount measurement part configured to measure an uplink interference amount for each interference measurement unit which comprises a predetermined period and a predetermined number of frequency blocks was provided.
Abstract: The object is achieved by providing a base station apparatus for performing time and frequency scheduling in uplink packet access with: an interference amount measurement part configured to measure an uplink interference amount for each interference amount measurement unit which comprises a predetermined period and a predetermined number of frequency blocks; an interference amount determination part configured to determine whether the uplink interference amount satisfies a predetermined condition; and an overload indicator reporting part configured to report an overload indicator to a neighboring cell when the predetermined condition is satisfied.
498 citations
•
26 Feb 2008TL;DR: In this article, a base station device which performs communication with a mobile station by using a downlink shared channel includes: reception means for receiving transmission acknowledgement information for a DL shared channel from the mobile station using a control channel which is frequency-multiplexed with an uplink shared channel; measurement means for measuring radio quality of the control channel; and judgment means for judging the reliability of the DL radio quality information according to the radio quality.
Abstract: A base station device which performs communication with a mobile station by using a downlink shared channel includes: reception means for receiving transmission acknowledgement information for a downlink shared channel from the mobile station by a control channel which is frequency-multiplexed with an uplink shared channel; measurement means for measuring radio quality of the control channel; and judgment means for judging the reliability of the downlink radio quality information according to the radio quality. Moreover, a communication control method in the base station device which performs communication with the mobile station by using the downlink shared channel includes: a step of receiving transmission acknowledgement information for the downlink shared channel from the mobile station by the control channel which is frequency-multiplexed with the uplink shared channel; a step of measuring radio quality of the control channel; and a step of judging whether the mobile station has transmitted the downlink radio quality information according to the radio quality.
282 citations
•
TL;DR: This paper proposes to decompose complex questions into a sequence of simple questions, and compute the final answer from the sequence of answers, and empirically demonstrates that question decomposition improves performance from 20.8 precision@1 to 27.5 precision @1 on this new dataset.
Abstract: Answering complex questions is a time-consuming activity for humans that requires reasoning and integration of information. Recent work on reading comprehension made headway in answering simple questions, but tackling complex questions is still an ongoing research challenge. Conversely, semantic parsers have been successful at handling compositionality, but only when the information resides in a target knowledge-base. In this paper, we present a novel framework for answering broad and complex questions, assuming answering simple questions is possible using a search engine and a reading comprehension model. We propose to decompose complex questions into a sequence of simple questions, and compute the final answer from the sequence of answers. To illustrate the viability of our approach, we create a new dataset of complex questions, ComplexWebQuestions, and present a model that decomposes questions and interacts with the web to compute an answer. We empirically demonstrate that question decomposition improves performance from 20.8 precision@1 to 27.5 precision@1 on this new dataset.
256 citations
••
01 Nov 2019
TL;DR: In this task, 18 distinct question answering datasets were adapted and unified into the same format and the best system achieved an average F1 score of 72.5 on the 12 held-out datasets.
Abstract: We present the results of the Machine Reading for Question Answering (MRQA) 2019 shared task on evaluating the generalization capabilities of reading comprehension systems. In this task, we adapted and unified 18 distinct question answering datasets into the same format. Among them, six datasets were made available for training, six datasets were made available for development, and the rest were hidden for final evaluation. Ten teams submitted systems, which explored various ideas including data sampling, multi-task learning, adversarial training and ensembling. The best system achieved an average F1 score of 72.5 on the 12 held-out datasets, 10.7 absolute points higher than our initial baseline based on BERT.
252 citations
•
13 Aug 2008TL;DR: In this paper, the authors describe a technique for sending control information in a wireless communication system where a UE spreads control information across frequency with a DFT and across time with an orthogonal sequence to obtain output data for the control information.
Abstract: Techniques for sending control information in a wireless communication system are described. In an aspect, a UE spreads control information across frequency with a DFT and across time with an orthogonal sequence to obtain output data for the control information. In one design, the UE receives codewords for N HARQ processes in N downlink subframes, determines an ACK value for each HARQ process, codes N ACK values for the N HARQ processes to obtain ACK information, generates output data for the ACK information, and sends the output data in one of M uplink subframes. In another aspect, first control information is processed based on a first coding and multiplexing scheme utilizing code division multiplexing in time and frequency domains. Second control information is processed based on a second coding and multiplexing scheme utilizing code division multiplexing in time domain and spreading in frequency domain.
99 citations