Digital Library[ Search Result ]
Improving Retrieval Models through Reinforcement Learning with Feedback
Min-Taek Seo, Joon-Ho Lim, Tae-Hyeong Kim, Hwi-Jung Ryu, Du-Seong Chang, Seung-Hoon Na
http://doi.org/10.5626/JOK.2024.51.10.900
Open-domain question answering involves the process of retrieving clues through search to solve problems. In such tasks, it is crucial that the search model provides appropriate clues, as this directly impacts the final performance. Moreover, information retrieval is an important function frequently used in everyday life. This paper recognizes the significance of these challenges and aims to improve performances of search models. Just as the recent trend involves adjusting outputs in decoder models using Reinforcement Learning from Human Feedback (RLHF), this study seeks to enhance search models through the use of reinforcement learning. Specifically, we defined two rewards: the loss of the answer model and the similarity between the retrieved documents and the correct document. Based on these, we applied reinforcement learning to adjust the probability score of the top-ranked document in the search model's document probability distribution. Through this approach, we confirmed the generality of the reinforcement learning method and its potential for further performance improvements.
PrefixLM for Korean Text Summarization
Kun-Hui Lee, Seung-Hoon Na, Joon-Ho Lim, Tae-Hyeong Kim, Du-Seong Chang
http://doi.org/10.5626/JOK.2022.49.6.475
In this paper, we examine the effectiveness of PrefixLM that consists of half of the parameters of the T5"s encoder-decoder architecture for Korean text generation tasks. Different from T5 where input and output sequences are separately provided, the transformer block of PrefixLM takes a single sequence that concatenates both input and output sequences. By designing the attention mask, PrefixLM performs uni- and bi-directional attentions on input and output sequences, respectively, thereby enabling to perform two roles of encoder and decoder with a single transformer block. Experiment results on Korean abstractive document summarization task show that PrefixLM leads to performance increases of 2.17 and 2.78 more than 2 in Rouge-F1 score over BART and T5, respectively, implying that the PrefixLM is promising in Korean text generation tasks.
Radar Signal Processor for High-Resolution Target Detection
http://doi.org/10.5626/JOK.2022.49.5.369
Recently, as the technology of multi-function radar is developed, the radar deception technology of ballistic missiles is also developing. For some ballistic missiles, the propellant explodes in the air after the stage is separated, causing the warhead and many fragments to fly together, which lowers the multi-function radar’s ability to engage ballistic missiles. Thus, there is a need for a radar system capable of operating a broadband waveform to intercept a warhead, by quickly discriminating between a high-speed warhead and fragments while retaining the existing target detection/tracking function. It is possible to find and intercept warhead among fragments by extracting the length of a target, using a broadband waveform and performing warhead classification using this. In this paper, we describe the process of performing the target detection/tracking function using a narrowband waveform such as doppler processing, pulse compression, threshold processing, and target processing and high-resolution target length extraction and phase diffraction correction for accurate length extraction using a wideband waveform to create a radar system that satisfies these requirements. Also, it shows the results of designing and implementing these functions with signal processing software and performing tests.
Component-based Software Architecture Design Method for Defense Software
Sungwon Lee, Jonghwan Shin, Taehyung Kim
http://doi.org/10.5626/JOK.2019.46.11.1113
Component-based software engineering is widely used in a variety of embedded software developments. However, most methodologies for component-based software engineering have certain limitations in coping with the software configuration structure governed by Korean regulations for weapon system software development. The software configuration structure by rule assumes that the software development is based on object-oriented language and tries to present different perspectives in one diagram. In this paper, we propose a component-based software architecture design method for defense software that can be used in software development with non-object-oriented language. Further, the proposal aims to compose a software configuration structure and desired documentation products such as diagrams through a design process. To help comprehend each step of the proposed design method, real samples of ongoing projects are presented.
Regularizing Korean Conversational Model by Applying Denoising Mechanism
Tae-Hyeong Kim, Yunseok Noh, Seong-Bae Park, Se-Yeong Park
http://doi.org/10.5626/JOK.2018.45.6.572
A conversation system is a system that responds appropriately to input utterances. Recently, the sequence-to-sequence framework has been widely used as a conversation-learning model. However, the conversation model learned in such a way often generates a safe and dull response that does not provide appropriate information or sophisticated meaning. In addition, this model is also useless for input utterances appearing in various forms, such as with changed ending words or changed word order. To solve these problems, we propose a denoising response generation model applying a denoising mechanism. By injecting noise into original input, the proposed method creates a model that will stochastically experience new input made up of items that were not included in the original data during the training process. This data augmentation effect regularizes the model and allows the realization of a robust model. We evaluate our model using 90k input utterances-responses from Korean conversation pair data. The proposed model achieves better results compared to a baseline model on both ROUGE F1 score and qualitative evaluations by human annotators.
Search
Journal of KIISE
- ISSN : 2383-630X(Print)
- ISSN : 2383-6296(Electronic)
- KCI Accredited Journal
Editorial Office
- Tel. +82-2-588-9240
- Fax. +82-2-521-1352
- E-mail. chwoo@kiise.or.kr