• 한국정보통신학회:학술대회논문집
• /
• 한국정보통신학회 2022년도 춘계학술대회
• /
• pp.633-635
• /
• 2022

비디오 월 컨트롤러의 운용 중에 오류가 발생하면 제어 시스템은 로그 파일을 생성하고 로그를 기록한다. 로그 기록으로 인한 시스템의 부하를 줄이기 위해 로그레벨을 사용하는데, 평상시에는 로그레벨을 낮게 설정하여 가급적 로그를 기록하지 않고 오류가 발생하였을 때 로그레벨을 변경하여 상세한 로그를 기록하도록 운용하고 있다. 이로 인해 오류를 인지하더라도 즉각적인 원인 분석 및 대처가 불가능하고 로그레벨을 변경하기 위해서는 운영자의 개입이 불가피하다. 따라서 본 논문에서는 Contextual Bandit을 이용하여 운용 상황에 따라 로그레벨을 자동으로 설정하는 모델을 제안한다.

• KSII Transactions on Internet and Information Systems (TIIS)
• /
• 제14권8호
• /
• pp.3168-3186
• /
• 2020

Automatic topic essay generation (TEG) is a controllable text generation task that aims to generate informative, diverse, and topic-consistent essays based on multiple topics. To make the generated essays of high quality, a reasonable method should consider both diversity and topic-consistency. Another essential issue is the intrinsic link of the topics, which contributes to making the essays closely surround the semantics of provided topics. However, it remains challenging for TEG to fill the semantic gap between source topic words and target output, and a more powerful model is needed to capture the semantics of given topics. To this end, we propose a pretraining-based contextual self-attention (PC-SAN) model that is built upon the seq2seq framework. For the encoder of our model, we employ a dynamic weight sum of layers from BERT to fully utilize the semantics of topics, which is of great help to fill the gap and improve the quality of the generated essays. In the decoding phase, we also transform the target-side contextual history information into the query layers to alleviate the lack of context in typical self-attention networks (SANs). Experimental results on large-scale paragraph-level Chinese corpora verify that our model is capable of generating diverse, topic-consistent text and essentially makes improvements as compare to strong baselines. Furthermore, extensive analysis validates the effectiveness of contextual embeddings from BERT and contextual history information in SANs.

• Journal of Information Processing Systems
• /
• 제17권4호
• /
• pp.818-833
• /
• 2021

To address the problems of low precision rate, insufficient feature extraction, and poor contextual ability in existing text sentiment analysis methods, a mixed model account of a CNN-BiLSTM-TE (convolutional neural network, bidirectional long short-term memory, and topic extraction) model was proposed. First, Chinese text data was converted into vectors through the method of transfer learning by Word2Vec. Second, local features were extracted by the CNN model. Then, contextual information was extracted by the BiLSTM neural network and the emotional tendency was obtained using softmax. Finally, topics were extracted by the term frequency-inverse document frequency and K-means. Compared with the CNN, BiLSTM, and gate recurrent unit (GRU) models, the CNN-BiLSTM-TE model's F1-score was higher than other models by 0.0147, 0.006, and 0.0052, respectively. Then compared with CNN-LSTM, LSTM-CNN, and BiLSTM-CNN models, the F1-score was higher by 0.0071, 0.0038, and 0.0049, respectively. Experimental results showed that the CNN-BiLSTM-TE model can effectively improve various indicators in application. Lastly, performed scalability verification through a takeaway dataset, which has great value in practical applications.

• 지능정보연구
• /
• 제28권4호
• /
• pp.95-117
• /
• 2022

인터넷 기술의 발전으로 인해 다양한 미디어가 등장하면서 광고주들은 기업의 광고 전략에 적합한 미디어를 선택하는데 어려움을 경험하고 있다. 전통적인 광고 마케팅 전략을 바탕으로 광고 미디어를 선택하면 소비자의 상황 정보를 효과적으로 반영하는데 어려움이 존재한다. 이러한 상황에서 소비자의 과거 데이터를 분석하여 소비자가 필요하거나 관심 있는 정보를 바탕으로 광고주에게 맞춤형 미디어를 제공하는 추천 시스템이 필요하다. 전통적인 추천 시스템은 정량적 선호도 정보를 기반으로 추천 서비스를 제공하기 때문에 다양한 상황 정보를 반영하기 어려운 문제점이 존재한다. 본 연구에서는 딥러닝을 이용하여 소비자의 미디어 시청 시간, 거주 지역, 나이, 성별 등 상황 정보를 고려하여 광고주에게 맞춤형 미디어를 추천하는 방법론을 제안한다. 본 연구는 한국방송광고진흥공사에서 제공하는 소비자행태조사 데이터를 사용하여 추천 시스템을 구축하였다. 또한, 기존 연구에서 널리 사용되는 여러 벤치마크 모델과 비교하여 추천 성능을 검증하였다. 실험 결과, 본 연구에서 제안하는 소비자의 상황 정보를 반영한 추천 모델이 기존의 벤치마크 모델보다 높은 정확성을 나타내는 것을 확인하였다. 이 연구는 향후 광고주들이 소비자의 여러 상황 정보를 바탕으로 맞춤형 미디어 선택할 때 효과적인 의사결정을 내릴 수 있도록 도움을 주는데 기여를 할 수 있을 것으로 기대한다

• Asia Marketing Journal
• /
• 제15권3호
• /
• pp.159-172
• /
• 2013

In the present study, we apply the multiple cue probability learning (MCPL) paradigm to examine consumer learning from feedback in repeated trials. This paradigm is useful in investigating consumer learning, especially learning the relationships between the overall quality and attributes. With this paradigm, we can analyze what people learn from repeated trials by using the lens model, i.e., whether it is knowledge or consistency. In addition to introducing this paradigm, we aim to demonstrate that knowledge people gain from repeated trials with feedback is robust enough to weaken one of the most often examined contextual effects, the asymmetric dominance effect. The experiment consists of learning session and a choice task and stimuli are sport rafting boats with motor engines. During the learning session, the participants are shown an option with three attributes and are asked to evaluate its overall quality and type in a number between 0 and 100. Then an expert's evaluation, a number between 0 and 100, is provided as feedback. This trial is repeated fifteen times with different sets of attributes, which comprises one learning session. Depending on the conditions, the participants do one (low) or three (high) learning sessions or do not go through any learning session (no learning). After learning session, the participants then are provided with either a core or an extended choice set to make a choice to examine if learning from feedback would weaken the asymmetric dominance effect. The experiment uses a between-subjects experimental design (2 × 3; core set vs. extended set; no vs. low vs. high learning). The results show that the participants evaluate the overall qualities more accurately with learning. They learn the true trade-off rule between attributes (increase in knowledge) and become more consistent in their evaluations. Regarding the choice task, there is a significant decrease in the percentage of choosing the target option in the extended sets with learning, which clearly demonstrates that learning decreases the magnitude of the asymmetric dominance effect. However, these results are significant only when no learning condition is compared either to low or high learning condition. There is no significant result between low and high learning conditions, which may be due to fatigue or reflect the characteristics of learning curve. The present study introduces the MCPL paradigm in examining consumer learning and demonstrates that learning from feedback increases both knowledge and consistency and weakens the asymmetric dominance effect. The latter result may suggest that the previous demonstrations of the asymmetric dominance effect are somewhat exaggerated. In a single choice setting, people do not have enough information or experience about the stimuli, which may lead them to depend mostly on the contextual structure among options. In the future, more realistic stimuli and real experts' judgments can be used to increase the external validity of study results. In addition, consumers often learn through repeated choices in real consumer settings. Therefore, what consumers learn from feedback in repeated choices would be an interesting topic to investigate.

• ETRI Journal
• /
• 제43권2호
• /
• pp.299-312
• /
• 2021

It is necessary to achieve high performance in the task of zero anaphora resolution (ZAR) for completely understanding the texts in Korean, Japanese, Chinese, and various other languages. Deep-learning-based models are being employed for building ZAR systems, owing to the success of deep learning in the recent years. However, the objective of building a high-quality ZAR system is far from being achieved even using these models. To enhance the current ZAR techniques, we fine-tuned a pretrained bidirectional encoder representations from transformers (BERT). Notably, BERT is a general language representation model that enables systems to utilize deep bidirectional contextual information in a natural language text. It extensively exploits the attention mechanism based upon the sequence-transduction model Transformer. In our model, classification is simultaneously performed for all the words in the input word sequence to decide whether each word can be an antecedent. We seek end-to-end learning by disallowing any use of hand-crafted or dependency-parsing features. Experimental results show that compared with other models, our approach can significantly improve the performance of ZAR.

• Journal of Information Processing Systems
• /
• 제19권6호
• /
• pp.842-857
• /
• 2023

Currently, most sentiment classification models on microblogging platforms analyze sentence parts of speech and emoticons without comprehending users' emotional inclinations and grasping moral nuances. This study proposes a hybrid sentiment analysis model. Given the distinct nature of microblog comments, the model employs a combined stop-word list and word2vec for word vectorization. To mitigate local information loss, the TextCNN model, devoid of pooling layers, is employed for local feature extraction, while BiLSTM is utilized for contextual feature extraction in deep learning. Subsequently, microblog comment sentiments are categorized using a classification layer. Given the binary classification task at the output layer and the numerous hidden layers within BiLSTM, the Tanh activation function is adopted in this model. Experimental findings demonstrate that the enhanced TextCNN-BiLSTM model attains a precision of 94.75%. This represents a 1.21%, 1.25%, and 1.25% enhancement in precision, recall, and F1 values, respectively, in comparison to the individual deep learning models TextCNN. Furthermore, it outperforms BiLSTM by 0.78%, 0.9%, and 0.9% in precision, recall, and F1 values.

• 한국멀티미디어학회논문지
• /
• 제23권3호
• /
• pp.447-459
• /
• 2020

The Internet of Things(IoT) paradigm is flourishing strenuously for the last two decades. Researchers around the globe have their dreams to transmute every real-world object to the virtual object. Consequently, IoT devices are escalating exponentially. The abrupt evolution of these IoT devices has caused a major challenge i.e. object classification. In order to classify devices comprehensively and accurately, this paper proposes a context-aware based multi-classification model for devices, which classifies the smart devices according to people's contexts. However, the classification features of contextual data of different contexts are difficult to extract. The deep learning algorithm has the capability to solve this problem. This paper proposes a context-aware based multi-classification model of devices, which classifies the smart devices according to people's contexts.

• International Journal of Contents
• /
• 제18권3호
• /
• pp.11-20
• /
• 2022

Human Emotion Recognition is an exciting topic that has been attracting many researchers for a lengthy time. In recent years, there has been an increasing interest in exploiting contextual information on emotion recognition. Some previous explorations in psychology show that emotional perception is impacted by facial expressions, as well as contextual information from the scene, such as human activities, interactions, and body poses. Those explorations initialize a trend in computer vision in exploring the critical role of contexts, by considering them as modalities to infer predicted emotion along with facial expressions. However, the contextual information has not been fully exploited. The scene emotion created by the surrounding environment, can shape how people perceive emotion. Besides, additive fusion in multimodal training fashion is not practical, because the contributions of each modality are not equal to the final prediction. The purpose of this paper was to contribute to this growing area of research, by exploring the effectiveness of the emotional scene gist in the input image, to infer the emotional state of the primary target. The emotional scene gist includes emotion, emotional feelings, and actions or events that directly trigger emotional reactions in the input image. We also present an attention-based fusion network, to combine multimodal features based on their impacts on the target emotional state. We demonstrate the effectiveness of the method, through a significant improvement on the EMOTIC dataset.

• Journal of information and communication convergence engineering
• /
• 제22권1호
• /
• pp.44-55
• /
• 2024

High-utility itemset mining (HUIM) is a dominant technology that enables enterprises to make real-time decisions, including supply chain management, customer segmentation, and business analytics. However, classical support value-driven Apriori solutions are confined and unable to meet real-time enterprise demands, especially for large amounts of input data. This study introduces a groundbreaking model for top-N high utility itemset mining in real-time enterprise applications. Unlike traditional Apriori-based solutions, the proposed convolutional sequential embedding metrics-driven cosine-similarity-based multilayer perception learning model leverages global and contextual features, including semantic attributes, for enhanced top-N recommendations over sequential transactions. The MATLAB-based simulations of the model on diverse datasets, demonstrated an impressive precision (0.5632), mean absolute error (MAE) (0.7610), hit rate (HR)@K (0.5720), and normalized discounted cumulative gain (NDCG)@K (0.4268). The average MAE across different datasets and latent dimensions was 0.608. Additionally, the model achieved remarkable cumulative accuracy and precision of 97.94% and 97.04% in performance, respectively, surpassing existing state-of-the-art models. This affirms the robustness and effectiveness of the proposed model in real-time enterprise scenarios.