In order to process video data efficiently, a segmentation technique through scene change detection must be required. This is fundamental operation used in many digital applications such as libraries, on demand (VOD), etc. Many of these advanced require manipulations compressed signals. So, the achieved by analyzing directly domain, thereby avoiding overhead decompressing into individual frames pixel domain. this paper, we propose fast algorithm using direct feature extraction from MPEG...

In recent decades, rapid growth has been observed in the incorporation of sustainability into marketing. Accordingly, contrasting relationships between them have carefully studied to assess relevance one idea other and vice versa. response this change, scholars practitioners tasked with exploring trends Therefore, purpose study is investigate existing literatures on both all levels marketing, determine research provide implications applying for future practices. This investigated only title,...

Recent research in visual saliency has established a computational measure of perceptual importance. In this paper we present visual-saliency-based operator to enhance selected regions volume. We show how use such an on user-specified field compute emphasis field. further discuss the can be integrated into visualization pipeline through its modifications regional luminance and chrominance. Finally, validate our work using eye-tracking-based user study that new enhancement is more effective...

Mesh saliency has been proposed as a computational model of perceptual importance for meshes, and it used in graphics abstraction, simplification, segmentation, illumination, rendering, illustration. Even though this technique is inspired by models low-level human vision, not yet validated with respect to performance. Here, we present user study that compares the previous mesh approaches eye movements. To quantify correlation between fixation locations 3D rendered images, introduce...

Owing to the prevalence of coronavirus disease (COVID-19), coping with clinical issues at individual level has become important healthcare system. Accordingly, precise initiation treatment after a hospital visit is required for expedited processes and effective diagnoses outpatients. To achieve this, artificial intelligence in medical natural language processing (NLP), such as chatbot or decision support system, can be suitable tools an advanced Furthermore, decisions on specialty from...

Despite several improvements in the drug development pipeline over past decade, failures due to unexpected adverse effects have rapidly increased at all stages of clinical trials. To improve success rate trials, it is necessary identify potential loser candidates that may fail Therefore, we need develop reliable models for predicting outcomes trials candidates, which guide discovery process. In this study, propose an outer product–based convolutional neural network (OPCNN) model integrates...

We propose a new multi-view clustering method which uses results obtained on each view as voting pattern in order to construct set of clusters. Our experiments multilingual corpus documents show that performance increases significantly over simple concatenation and another technique.

Artists, illustrators, photographers, and cinematographers have long used the principles of contrast composition to guide visual attention. In this paper we introduce geometry modification as a tool persuasively direct We build upon recent advances in mesh saliency develop techniques alter elicit greater Eye-tracking-based user studies show that our approach successfully guides attention statistically significant manner. Our operates directly on geometry, therefore, produces view-independent...

The data collected from wireless sensor network indicate the system status, environment or health condition of human being, and we can use to carry out appropriate work by processing it. In recent years, using deep learning, it is possible construct a more intelligent context-aware predicting future situations as well monitoring current state. this article, propose framework for streaming analysis based on learning. particular, in an where time requirements are strictly enforced, results...

This paper presents a Web-based user evaluation of system for classifying and presenting political viewpoints blog posts. The is based on classification model trained using supervised learning algorithm, the data set consists recent posts from blogs that are self-identified as liberal or conservative viewpoint. We first discuss process. Then, with prototype retrieving blogs, we look at how results can be presented to users in order improve search experience. describe an online study 15...

In this paper we propose an extension of the PLSA model in which extra latent variable allows to co-cluster documents and terms simultaneously. We show on three datasets that our extended produces statistically significant improvements with respect two clustering measures over original multinomial mixture MM models.

KorQuAD 2.0은 총 100,000+ 쌍으로 구성된 한국어 질의응답 데이터셋이다. 기존 표준 데이터인 1.0과의 차이점은 크게 세가지가 있는데 첫 번째는 주어지는 지문이 한두 문단이 아닌 위키백과 한 페이지 전체라는 점이다. 두 번째로 지문에 표와 리스트도 포함되어 있기 때문에 HTML tag로 구조화된 문서에 대한 이해가 필요하다. 마지막으로 답변이 단어 혹은 구의 단위뿐 아니라 문단, 표, 리스트 전체를 포괄하는 긴 영역이 될 수 있다. Baseline 모델로 공개된 구글 BERT를 활용하여 F1 스코어 46.0%의 성능을 확인하였다. 이는 사람의 점수 85.7%에 비해 매우 낮은 점수로, 본 데이터가 도전적인 과제임을 알 추가적으로 답을 찾을 없는 경우에 학습 데이터 증강 방식을 통해 높였다. 데이터의 공개를 평문에 국한되어 있던 질의응답의 대상을 다양한 길이와 형식을 가진 과제로 확장하고자 한다.

The comprehensibility of large and complex 3D models can be greatly enhanced by guiding viewer's attention to important regions. Lighting is crucial our perception shape. Careful use lighting has been widely used in art, scientific illustration, computer graphics guide visual attention. In this paper, we explore how the saliency objects emphasize regions suppress less ones.

We propose a new pattern matching algorithm for composite context-aware services. The algorithm, RETE-ADH, extends RETE to enhance systems that are based on the service architecture. RETE-ADH increases speed of by searching only subset rules can be matched. In addition, is scalable and suitable parallelization. describe design proposed present experimental results from simulated smart office environment compare with other algorithms, showing outperforms original 85%.

Sign Language Translation (SLT) is a task that has not been studied relatively much compared to the study of Recognition (SLR). However, SLR recognizes unique grammar sign language, which different from spoken language and problem non-disabled people cannot easily interpret. So, we're going solve translating directly in video. To this end, we propose new keypoint normalization method for performing translation based on skeleton point signer robustly normalizing these points translation. It...

While many FSCIL studies have been undertaken, achieving satisfactory performance, especially during incremental sessions, has remained challenging. One prominent challenge is that the encoder, trained with an ample base session training set, often underperforms in sessions. In this study, we introduce a novel framework for FSCIL, capitalizing on generalizability of Contrastive Language-Image Pre-training (CLIP) model to unseen classes. We achieve by formulating image-object-specific (IOS)...

Automatic bibliographic reference annotation involves the tokenization and identification of fields. Recent methods use machine learning techniques such as Conditional Random Fields to tackle this problem. On other hand, state art always learn evaluate their systems with a well structured data having simple format bibliography at end scientific articles. And that is reason why parsing new different from regular does not work well. In our previous work, we have established standard for...

Antithesis relationship between marketing and sustainability has been reviewed carefully to identify the significance of one idea other vice versa. Authors this paper are willing answer question on how have researched in past 10 years. This study systematizes scientific knowledge research trends created through debate from researchers. uses a methodological approach topic modeling based Latent Dirichlet Allocation (LDA) model. choice is driven by opportunity its trend over time that was...