35,266 Hits in 3.9 sec

Contextual Domain Classification with Temporal Representations

Tzu-Hsiang Lin, Yipeng Shi, Chentao Ye, Yang Fan, Weitong Ruan, Emre Barut, Wael Hamza, Chengwei Su
2021 Proceedings of the 2021 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies: Industry Papers   unpublished
Experiments on the Contextual Domain Classification (CDC) task with various encoder architectures show that temporal representations combining both information outperforms only one of the two.  ...  We further demonstrate that our contextual Transformer is able to reduce 13.04% of classification errors compared to a non-contextual baseline.  ...  In this paper, we propose temporal representations to effectively leverage both recent and distant context on the Contextual Domain Classification (CDC) task.  ... 
doi:10.18653/v1/2021.naacl-industry.6 fatcat:iup7l5ovfbaljovnosvxydjoty

Action recognition with multiscale spatio-temporal contexts

Jiang Wang, Zhuoyuan Chen, Ying Wu
2011 CVPR 2011  
contextual domain.  ...  In this paper, we present a novel representation that captures contextual interactions between interest points, based on the density of all features observed in each interest point's mutliscale spatio-temporal  ...  For example, contextual domains with short spatial support and long temporal support can capture temporal evolution information.  ... 
doi:10.1109/cvpr.2011.5995493 dblp:conf/cvpr/WangCW11 fatcat:msqd4ll2abhqlc6k53y32m2zim

Contextual Max Pooling for Human Action Recognition

Zhong ZHANG, Shuang LIU, Xing MEI
2015 IEICE transactions on information and systems  
In this way, CMP explicitly considers the spatio-temporal contextual relationships among interest points and inherits the positive properties of max pooling.  ...  However, the spatio-temporal relationship among interest points has rarely been considered in the pooling step, which results in the imprecise representation of human actions.  ...  Computing the Probability p(x i , s i ) Given a spatio-temporal point (x i , s i ), its surrounding spatiotemporal area is called contextual domain [11] which is a cube with a predefined side length  ... 
doi:10.1587/transinf.2014edl8221 fatcat:oj4pmdzvzjc27dg5emnvhud37e

Building for Tomorrow: Assessing the Temporal Persistence of Text Classifiers [article]

Rabab Alkhalifa, Elena Kochkina, Arkaitz Zubiaga
2022 arXiv   pre-print
Findings from these experiments have important implications for the design of text classification models with the aim of preserving performance over time.  ...  characteristics can help predict the temporal stability of different models.  ...  Language models and classification algorithms We leverage a wide range of state-of-the-art static and contextual language representations, and neural classification models to evaluate by-design adaptability  ... 
arXiv:2205.05435v2 fatcat:xs7xdywlkffrrgvepqsorrbkwq

Comparative Analysis of Text Classification Approaches in Electronic Health Records [article]

Aurelie Mascio, Zeljko Kraljevic, Daniel Bean, Richard Dobson, Robert Stewart, Rebecca Bendayan, Angus Roberts
2020 arXiv   pre-print
ones based on contextual embeddings such as BERT.  ...  Recent advances in embedding methods have shown promising results for several clinical tasks, yet there is no exhaustive comparison of such approaches with other commonly used word representations and  ...  When combined with appropriate entity extraction tasks and specific domain embeddings, Bi-LSTM outperforms contextual embeddings.  ... 
arXiv:2005.06624v1 fatcat:pipw42sjpnavjaygcwmqngsmm4

Developing a Spatial-Temporal Contextual and Semantic Trajectory Clustering Framework [article]

Ivens Portugal, Paulo Alencar, Donald Cowan
2017 arXiv   pre-print
many domains.  ...  , the structure of space and time as well as the contextual and semantic trajectory properties.  ...  Another classification of the time representation deals with the way it is associated with the spatial-temporal objects.  ... 
arXiv:1712.03900v1 fatcat:ezz7gy5w6bafvdvp4fjvcut6te

Global Context-Aware Attention LSTM Networks for 3D Action Recognition

Jun Liu, Gang Wang, Ping Hu, Ling-Yu Duan, Alex C. Kot
2017 2017 IEEE Conference on Computer Vision and Pattern Recognition (CVPR)  
the assistance of global contextual information.  ...  we propose a new class of LST-M network, Global Context-Aware Attention LSTM (GCA-LSTM), for 3D action recognition, which is able to selectively focus on the informative joints in the action sequence with  ...  The ST-LSTM unit is equipped with an input gate (i j,t ), two forget gates corresponding to the two sources of contextual information (f (S) j,t for the spatial domain, and f (T ) j,t for the temporal  ... 
doi:10.1109/cvpr.2017.391 dblp:conf/cvpr/LiuWHDK17 fatcat:2gyrzku5lne47fnr43sd3lqbtu

Hierarchical Self-supervised Representation Learning for Movie Understanding [article]

Fanyi Xiao, Kaustav Kundu, Joseph Tighe, Davide Modolo
2022 arXiv   pre-print
We further demonstrate the effectiveness of our contextualized event features on LVU tasks [54], both when used alone and when combined with instance features, showing their complementarity.  ...  Most self-supervised video representation learning approaches focus on action recognition.  ...  The contextualizer can be lightweight and trained on a small amount of training data with stronger semantic and temporal structures (i.e., movies).  ... 
arXiv:2204.03101v1 fatcat:kl2xwoczfzedvd5tx452ecg2le

Time-Series Representation Learning via Temporal and Contextual Contrasting [article]

Emadeldeen Eldele, Mohamed Ragab, Zhenghua Chen, Min Wu, Chee Keong Kwoh, Xiaoli Li, Cuntai Guan
2021 arXiv   pre-print
Learning decent representations from unlabeled time-series data with temporal dynamics is a very challenging task.  ...  In this paper, we propose an unsupervised Time-Series representation learning framework via Temporal and Contextual Contrasting (TS-TCC), to learn time-series representation from unlabeled data.  ...  To address the above issues, we propose a Time-Series representation learning framework via Temporal and Contextual Contrasting (TS-TCC).  ... 
arXiv:2106.14112v1 fatcat:b45ts7bwungevd5jn375jnbmba

Ontology Patterns for Complex Activity Modelling [chapter]

Georgios Meditskos, Stamatia Dasiopoulou, Vasiliki Efstathiou, Ioannis Kompatsiaris
2013 Lecture Notes in Computer Science  
The aim is to allow the formal representation of activity interpretation models over activity classes that are generally characterized by intricate temporal associations, and where it is often the case  ...  The patterns implement the descriptions and situations (DnS) ontology pattern of DOLCE Ultra Lite, modelling activity classes of domain ontologies as instances.  ...  We use DnS to formally provide precise representations of contextualized situations and descriptions on activity concepts of the Domain Activity Ontology, describing the different activity types and temporal  ... 
doi:10.1007/978-3-642-39617-5_15 fatcat:qqogyludgbb4bfij6qftrcjwsu

Neural Temporality Adaptation for Document Classification: Diachronic Word Embeddings and Domain Adaptation Models

Xiaolei Huang, Michael J. Paul
2019 Proceedings of the 57th Annual Meeting of the Association for Computational Linguistics  
Second, we propose a time-driven neural classification model inspired by methods for domain adaptation. Experiments on six corpora show how these methods can make classifiers more robust over time.  ...  Language usage can change across periods of time, but document classifiers models are usually trained and tested on corpora spanning multiple years without considering temporal variations.  ...  We also observe that temporally closer domains share higher percentages of contextual words. The pattern aligns with our observations in the Section 2.2.  ... 
doi:10.18653/v1/p19-1403 dblp:conf/acl/HuangP19 fatcat:6otebknxaffyrn65drwwvel5aq

Contextualizing histogram

Bingbing Ni, Shuicheng Yan, Ashraf Kassim
2009 2009 IEEE Conference on Computer Vision and Pattern Recognition  
In this paper, we investigate how to incorporate spatial and/or temporal contextual information into classical histogram features with the aim of boosting visual classification performance.  ...  We evaluate these proposed methods on face recognition and group activity classification problems, and the results demonstrate that the contextualized histograms significantly boost the visual classification  ...  Temporal and Higher-order Extensions The above contextualized histogram is defined in the spatial domain only, but for applications related with videos, temporal contextual information is also critical  ... 
doi:10.1109/cvpr.2009.5206856 dblp:conf/cvpr/NiYK09a fatcat:ia6gzqowqveifks5zfggmozohe

Relations for Reusing (R4R) in A Shared Context: An Exploration on Research Publications and Cultural Objects

Andrea Wei-Ching Huang, Tyng-Ruey Chuang
2020 Figshare  
Will the rich domain knowledge from research publications and the implicit cross-domain metadata of cultural objects be compliant with each other?  ...  A contextual framework is proposed as dynamic and relational in supporting three different ontexts: Reusing, Publication and Curation, which are individually constructed but overlapped with major conceptual  ...  Here, we define a contextual setting as a sign with the triadic relation [13] :  The Representation is a representation of the activity context setting itself, and is the form that the setting takes.  ... 
doi:10.6084/m9.figshare.11987871.v2 fatcat:23bllc7rqbhxxaarlhuheybaae

SP-ACT: A hybrid framework for complex activity recognition combining OWL and SPARQL rules

Georgios Meditskos, Stamatia Dasiopoulou, Vasiliki Efstathiou, Ioannis Kompatsiaris
2013 2013 IEEE International Conference on Pervasive Computing and Communications Workshops (PERCOM Workshops)  
support for (i) temporal reasoning and (ii) new named individual assertions.  ...  The goal of the hybrid framework is to address the limitations of the ontology-based context modelling paradigm in domains that require the recognition of complex context elements, namely, the lack of  ...  Representation Layer The representation layer encapsulates a lightweight domain activity model for capturing information relevant to activities.  ... 
doi:10.1109/percomw.2013.6529451 dblp:conf/percom/MeditskosDEK13 fatcat:csofyz62drgdtemtlzox4udomu

On the Impact of Temporal Representations on Metaphor Detection [article]

Giorgio Ottolina, Matteo Palmonari, Mehwish Alam, Manuel Vimercati
2022 arXiv   pre-print
However, the results also suggest that temporal word embeddings may provide representations of the core meaning of the metaphor even too close to their contextual meaning, thus confusing the classifier  ...  To the best of our knowledge, this is the first study that examines the metaphor detection task with a detailed exploratory analysis where different temporal and static word embeddings are used to account  ...  These approaches use both (non-contextual) word embeddings and contextual word embeddings within a neural network with a final classification layer.  ... 
arXiv:2111.03320v2 fatcat:bze355a4r5ghnbiomlxmefo3iq
« Previous Showing results 1 — 15 out of 35,266 results