• 한국정보처리학회:학술대회논문집
• /
• 한국정보처리학회 2021년도 추계학술발표대회
• /
• pp.764-767
• /
• 2021

While Generative Adversarial Networks (GANs) have seen huge success in image synthesis tasks, synthesizing high-quality images from text descriptions is a challenging problem in computer vision. This paper proposes a method named Text-to-Face Generation Using Multi-Scale Gradients for Conditional Generative Adversarial Networks (T2F-MSGGANs) that combines GANs and a natural language processing model to create human faces has features found in the input text. The proposed method addresses two problems of GANs: model collapse and training instability by investigating how gradients at multiple scales can be used to generate high-resolution images. We show that T2F-MSGGANs converge stably and generate good-quality images.

• East Asian Economic Review
• /
• 제27권4호
• /
• pp.303-325
• /
• 2023

This paper examines how APEC has transformed itself into an international forum for the vision of regional integration. It aims to quantify the documentation produced by the international organization and provide quantifiable evidence that aligns with prior knowledge rather than relying solely on intuition. For this purpose, I use various text mining techniques to extract multi-dimensional features from the text of APEC Leaders' Declarations from 1993 to 2023. In terms of interest and expectations for APEC as a forum, it is found that members have experienced two major peaks and troughs over the last three decades. It is found that the change point coincides with the Asian financial crisis of 1997 and the tensions between the United States and China since 2017. To explore more various aspects of economic integration in the Asia-Pacific region, this study also considers how consistently APEC has been an international forum for addressing issues, which members are active, and how members have clustered based on their views of APEC.

• International Journal of Computer Science & Network Security
• /
• 제24권1호
• /
• pp.205-208
• /
• 2024

Artificial Intelligence (AI) technology has evolved rapidly in recent years and is used in everything from banking to email management to surgery, but without the help of the visible, most of the fun features of the Internet include visual impairment. It benefits people with disabilities. The main purpose of this study is to find ways to help people with visual impairments using AI technology. A visually impaired request is made for the visually impaired. For example, when a message arrives that the program will notify you by voice (reads the sender's name, read the message, and replies to it if necessary), this is a special program installed on your mobile phone. This program uses a customized algorithm developed in Python to convert written text to voice, read text, and convert voice to written text on a message when a visually impaired person wants to respond. Then it sends the response in the form of a text message. Therefore, the research should lead to programs for people with visual impairments. This program makes mobile phones easier and more comfortable to use and makes the daily life easier for visual impairments.

• KSII Transactions on Internet and Information Systems (TIIS)
• /
• 제8권1호
• /
• pp.213-231
• /
• 2014

Existing image search systems allow users to search images by keywords, or by example images through content-based image retrieval (CBIR). On the other hand, users might learn more relevant textual information about an image from its text captions or surrounding contexts within documents or Web pages. Without such contexts, it's difficult to extract semantic description directly from the image content. In this paper, we propose an annotation watermarking system for users to embed text descriptions, and retrieve more relevant textual information from similar images. First, tags associated with an image are converted by two-dimensional code and embedded into the image by discrete wavelet transform (DWT). Next, for images without annotations, similar images can be obtained by CBIR techniques and embedded annotations can be extracted. Specifically, we use global features such as color ratios and dominant sub-image colors for preliminary filtering. Then, local features such as Scale-Invariant Feature Transform (SIFT) descriptors are extracted for similarity matching. This design can achieve good effectiveness with reasonable processing time in practical systems. Our experimental results showed good accuracy in retrieving similar images and extracting relevant tags from similar images.

• 대한전자공학회:학술대회논문집
• /
• 대한전자공학회 2005년도 추계종합학술대회
• /
• pp.979-982
• /
• 2005

This paper is to be pre-processing that decides the text recognizability and quality contained in natural image. Differentiated with the existing studies, In this paper, it suggests the application of partially unified color models, Coiflet Wavelet and text extraction algorithm that uses the closed curve edge features of LoG (laplacian of gaussian)operator. The text image included in natural image such as signboard has the same hue, saturation and value, and there is a certain thickness as for their feature. Each color element is restructured into closed area by LoG operator, the 2nd differential operator. The text area is contracted by Hough Transform, logical AND-OR operator of each color model and Minimum-Distance classifier. This paper targets natural image into which text area is added regardless of the size and resolution of the image, and it is confirmed to have more excellent performance than other algorithms with many restrictions.

• 영어영문학
• /
• 제57권5호
• /
• pp.871-887
• /
• 2011

In Lacanian terms, the real, which is a non-representative Ding an sich, is indirectly approachable only in and through language. This 'speaking of the real' is made possible through a restoration of the missing link between one signifier, S1 and another signifier, S₂, as is manifested in the Lacanian formula of metaphor. In Freudian terms of textual metaphor, the missing link is restored by substituting a new edition for an old edition of one's historical text of life. This is what this essay means by the metaphorical/dualistic structure of the analytic/literary text. And this is a way of talking about an intertextuality between literature and psychoanalysis in the sense of the 'text as psyche' and the 'psyche as text.' Applying the 'signifying substitution' to the Oedipus complex, the Oedipal child can find a meaning(s), "my erotic indulgement with my Mom is wrong" by metaphorically substituting S₂: the Name of the Father for S1: the Desire of the Mother. This meaning leads to the constitution of the human subject and the formation of the incest taboo, one of the most significant distinctive features of the human being as distinguished from the animals. We can see a similar metaphorical structure of S1-S2 taking place in the literary texts such as Macbeth and "Dover Beach": in the course of the stage of life being substituted for the primal scene in the former, and the plain of Tucydides for a bed scene in the latter, respectively.

• 한국컴퓨터정보학회논문지
• /
• 제14권6호
• /
• pp.1-10
• /
• 2009

플래시 메모리는 비휘발성이고 저전력으로 동작하며 가볍고 내구성이 강하다. 이러한 특성으로 휴대용 멀티미디어 재생기(PMP)와 같은 모바일 컴퓨팅 환경에서의 저장 장치로 많이 사용되고 있다. 대용량의 플래시 메모리를 저장 장치로 가진 모바일 기기들은 비디오/오디오/사진등과 같은 다양한 종류의 멀티미디어 데이터를 저장하고 재생한다. 모바일 컴퓨팅 장치를 위한 기존의 인덱스 시스템은 노래 가사와 같은 텍스트 형태의 정보 검색에 비효육적이다. 본 논문에서는 대용량 플래시 메모리 기반 임베디드 텍스트 인덱스(Embedded Text Index: EMTEX) 시스템을 제안한다. EMTEX는 먼저 임베디드 시스템을 고려한 압축 알고리즘을 사용하며, 텍스트 인덱스가 구성된 필드에 삽입 및 삭제시 인덱스에 즉시 반영된다. 뿐만 아니라, 플래시 메모리의 특성을 고려한 효율적인 삽입, 삭제, 재구성 기능을 수행하며, DBMS의 상위 계층에서 독립적으로 동작한다는 장점을 갖는다. 제안한 시스템의 성능 평가를 위해 다양한 환경에서 실험을 수행하였다. 그 결과 EMTEX는 임베디드 환경에서 Oracle Text나 FT3와 같은 기존의 인덱스 시스템보다 더 좋은 성능을 보여주었다.

• Journal of information and communication convergence engineering
• /
• 제13권2호
• /
• pp.113-122
• /
• 2015

Feature subset selection is as a pre-processing step in learning algorithms. In this paper, we propose an efficient algorithm, ModifiedFAST, for feature subset selection. This algorithm is suitable for text datasets, and uses the concept of information gain to remove irrelevant and redundant features. A new optimal value of the threshold for symmetric uncertainty, used to identify relevant features, is found. The thresholds used by previous feature selection algorithms such as FAST, Relief, and CFS were not optimal. It has been proven that the threshold value greatly affects the percentage of selected features and the classification accuracy. A new performance unified metric that combines accuracy and the number of features selected has been proposed and applied in the proposed algorithm. It was experimentally shown that the percentage of selected features obtained by the proposed algorithm was lower than that obtained using existing algorithms in most of the datasets. The effectiveness of our algorithm on the optimal threshold was statistically validated with other algorithms.

• 한국산학기술학회논문지
• /
• 제16권8호
• /
• pp.5565-5570
• /
• 2015

암묵인용문 인식은 학술문헌의 본문 텍스트 내에서 명시적 인용표지가 누락된 인용문장을 자동 인식하는 것으로 인용 기반 논문 검색 및 요약의 핵심 기술이다. 기존 암묵인용문 인식의 최신 연구들은 단어 ngram, 단서어구, 명시인용문과의 거리, 기존 연구자의 성, 기존 방법의 명칭 등 다양한 자질을 활용하여 50% 이상 인식 수준을 보고하고 있다. 그러나 대부분의 기존 연구들은 영어에 대해 수행되었으며 한국어의 경우 최근 긍정/부정 단서어구 패턴을 활용한 규칙 기반 시도에서 42% 성능 수준이 보고되어 있어 추가 성능 향상이 요구되는 상황이다. 이 연구에서는 한국어 어휘 자질을 사용하여 한국어 암묵인용문의 기계학습 기반 인식을 시도하였다. 이를 위해 어절, 형태소, 음절 단위에 기반한 다양한 크기의 어휘 ngram 자질들의 인식 성능을 비교 평가하고 한국어 암묵인용문 인식에 적합한 어휘 자질로 형태소 1gram 및 음절 2gram 단위를 결정하였다. 또한 이들 어휘 자질들을 전후 명시인용문들과의 인접성을 표현한 위치 자질들과 결합하여 한국어 암묵인용문 인식 성능을 50% 이상 수준으로 대폭 향상시켰다.

• International Journal of Computer Science & Network Security
• /
• 제23권11호
• /
• pp.59-66
• /
• 2023

Text mining (TM) is most widely used to find patterns from various text documents. Cyber-bullying is the term that is used to abuse a person online or offline platform. Nowadays cyber-bullying becomes more dangerous to people who are using social networking sites (SNS). Cyber-bullying is of many types such as text messaging, morphed images, morphed videos, etc. It is a very difficult task to prevent this type of abuse of the person in online SNS. Finding accurate text mining patterns gives better results in detecting cyber-bullying on any platform. Cyber-bullying is developed with the online SNS to send defamatory statements or orally bully other persons or by using the online platform to abuse in front of SNS users. Deep Learning (DL) is one of the significant domains which are used to extract and learn the quality features dynamically from the low-level text inclusions. In this scenario, Convolutional neural networks (CNN) are used for training the text data, images, and videos. CNN is a very powerful approach to training on these types of data and achieved better text classification. In this paper, an Ensemble model is introduced with the integration of Term Frequency (TF)-Inverse document frequency (IDF) and Deep Neural Network (DNN) with advanced feature-extracting techniques to classify the bullying text, images, and videos. The proposed approach also focused on reducing the training time and memory usage which helps the classification improvement.