Filters








21,329 Hits in 3.5 sec

Discovering Latent Class Labels for Multi-Label Learning

Jun Huang, Linchuan Xu, Jing Wang, Lei Feng, Kenji Yamanishi
2020 Proceedings of the Twenty-Ninth International Joint Conference on Artificial Intelligence  
., Discovering Latent Class Labels for MLL) is proposed which can not only discover the latent labels in the training data but also predict new instances with the latent and known labels simultaneously  ...  Existing multi-label learning (MLL) approaches mainly assume all the labels are observed and construct classification models with a fixed set of target labels (known labels).  ...  The metric of BLEU suggested by the WebNLG challenge is multi-BLEU. For metrics BLEU and METEOR, the higher the better, while for metric TER, the lower the better.  ... 
doi:10.24963/ijcai.2020/419 dblp:conf/ijcai/GaoWHX20 fatcat:cz455fctnzao5dkfgoyziebffy

Latent Class Multi-Label Classification to Identify Subclasses of Disease for Improved Prediction

Awad Alsaid Alyousef, Svetlana Nihtyanova, Christopher P. Denton, Pietro Bosoni, Riccardo Bellazzi, Allan Tucker
2019 2019 IEEE 32nd International Symposium on Computer-Based Medical Systems (CBMS)  
Results show that the "Latent Class Multi-Label Classification Model" improves accuracy when compared with competitive similar methods.  ...  The new algorithm uses latent class models to cluster patients within groups that results in improved classification as well as aiding the understanding of the underlying differences of the discovered  ...  Results We now document the comparison of our Latent Class Model Multi Label Naïve Bayes Classifier to other methods when predicting the time to PF, time to PH and time to death for each discovered subgroup  ... 
doi:10.1109/cbms.2019.00109 dblp:conf/cbms/AlyousefNDBBT19 fatcat:brxuxvdwdjgwhafg2vgahif3se

Multi-Feedback Bandit Learning with Probabilistic Contexts

Luting Yang, Jianyi Yang, Shaolei Ren
2020 Proceedings of the Twenty-Ninth International Joint Conference on Artificial Intelligence  
We propose a kernelized learning algorithm based on upper confidence bound to choose the optimal arm in reproducing kernel Hilbert space for each context bundle.  ...  In this work, we focus on multi-feedback bandit learning with probabilistic contexts, where a bundle of contexts are revealed to the agent along with their corresponding probabilities at the beginning  ...  Discovering Latent Class Labels Problem Definition (Discovering Latent Class Labels for MLL).  ... 
doi:10.24963/ijcai.2020/423 dblp:conf/ijcai/HuangXWFY20 fatcat:ponhjxdxfjhlhjbaqyque2synm

Semi-supervised Multi-view Manifold Discriminant Intact Space Lear

2018 KSII Transactions on Internet and Information Systems  
Although some semi-supervised multi-view latent space learning methods have been presented, there is still much space for improvement: 1) How to learn latent discriminant intact feature representations  ...  Semi-supervised multi-view latent space learning is gaining considerable popularity recently in many machine learning applications due to the high cost and difficulty to obtain the large amount of label  ...  learning (MISL) [13] discovers a latent intact representation for each data of multiple views.  ... 
doi:10.3837/tiis.2018.09.011 fatcat:v5ghd5iuavegtmlawnvnous3si

Visual recognition by learning from web data: A weakly supervised domain generalization approach

Li Niu, Wen Li, Dong Xu
2015 2015 IEEE Conference on Computer Vision and Pattern Recognition (CVPR)  
We then extend our MIL formulation to learn one classifier for each class and each latent domain such that multiple classifiers from each class can be effectively integrated to achieve better generalization  ...  In this work, we formulate a new weakly supervised domain generalization approach for visual recognition by using loosely labeled web images/videos as training data.  ...  Finally, we incorporate the discovered latent domains into our multi-class multi-instance learning approach in order to learn the integrated classifiers, which are robust to any unseen target domain.  ... 
doi:10.1109/cvpr.2015.7298894 dblp:conf/cvpr/NiuLX15 fatcat:x5uwum4lzngvznliq3b4rrb64u

Learning Multimodal Latent Attributes

Yanwei Fu, Timothy M. Hospedales, Tao Xiang, Shaogang Gong
2014 IEEE Transactions on Pattern Analysis and Machine Intelligence  
model for learning multi-modal semi-latent attributes, which dramatically reduces requirements for an exhaustive accurate attribute ontology and expensive annotation effort.  ...  We show that our framework is able to exploit latent attributes to outperform contemporary approaches for addressing a variety of realistic multimedia sparse data learning tasks including: multi-task learning  ...  Learns a generative model for both class label and annotations given latent topics, in contrast to the attribute paradigm of expressing classes in terms of annotations/attributes.  ... 
doi:10.1109/tpami.2013.128 pmid:24356351 fatcat:tlchipuvl5evflsw6ewqs2oqyu

Discovering Hidden Factors of Variation in Deep Networks [article]

Brian Cheung, Jesse A. Livezey, Arjun K. Bansal, Bruno A. Olshausen
2015 arXiv   pre-print
Deep learning has enjoyed a great deal of success because of its ability to learn useful features for tasks such as classification.  ...  for categorization.  ...  We thank Nervana Systems for supporting Brian Cheung during the summer when this project originated and for their continued collaboration.  ... 
arXiv:1412.6583v4 fatcat:tw5zd5az75dspkiargqwzjmboe

Partially labeled topic models for interpretable text mining

Daniel Ramage, Christopher D. Manning, Susan Dumais
2011 Proceedings of the 17th ACM SIGKDD international conference on Knowledge discovery and data mining - KDD '11  
These models make use of the unsupervised learning machinery of topic models to discover the hidden topics within each label, as well as unlabeled, corpus-wide latent topics.  ...  Effective text mining in this setting requires models that can flexibly account for the textual patterns that underlie the observed labels while still discovering unlabeled topics.  ...  In contrast, supervised learning and (multi-) label prediction explicitly model the label space for the purpose of prediction (such as in [12, 18] ), but by design do not discover latent sub-structure  ... 
doi:10.1145/2020408.2020481 dblp:conf/kdd/RamageMD11 fatcat:iksnzhsbofctffqfy43l6ma2d4

Refining Image Categorization by Exploiting Web Images and General Corpus [article]

Yazhou Yao, Jian Zhang, Fumin Shen, Xiansheng Hua, Wankou Yang and Zhenmin Tang
2017 arXiv   pre-print
To suppress the search error induced noisy images, we then formulate image selection and classifier learning as a multi-class multi-instance learning problem and propose to solve the employed problem by  ...  The following two major challenges are well studied: 1) noise in the labels of subcategories derived from the general corpus; 2) noise in the labels of images retrieved from the web.  ...  Another one is based on subcategories discovering and multi-class multi-instance learning (which we refer to SDMML).  ... 
arXiv:1703.05451v1 fatcat:gjpbnhlog5gq5j4n25iqvrvedy

Overcoming data scarcity with transfer learning [article]

Maxwell L. Hutchinson, Erin Antono, Brenna M. Gibbons, Sean Paradiso, Julia Ling, Bryce Meredig
2017 arXiv   pre-print
Here, we describe and compare three techniques for transfer learning: multi-task, difference, and explicit latent variable architectures.  ...  For activation energies of steps in NO reduction, the explicit latent variable method is not only the most accurate, but also enjoys cancellation of errors in functions that depend on multiple tasks.  ...  Unlike multi-task learning and explicit latent variables, difference learning cannot be used directly for multi-class classification.  ... 
arXiv:1711.05099v1 fatcat:nbk535l4cfbtjgwn5xyfckmxse

Domain Agnostic Learning for Unbiased Authentication [article]

Jian Liang, Yuren Cao, Shuang Li, Bing Bai, Hao Li, Fei Wang, Kun Bai
2020 arXiv   pre-print
In our approach, the latent domains are discovered by learning the heterogeneous predictive relationships between inputs and outputs.  ...  However, for authentication, there could be a large number of domains shared by different identities/classes, and it is impossible to annotate these domains exhaustively.  ...  Each class set includes its unique classes. "Train"/"Test" denotes the data for training/testing. "Latent" suggests domain labels are absent.  ... 
arXiv:2010.05250v2 fatcat:h4lrernzuvh2ddxjhvlv7j32si

Learning Action Primitives for Multi-level Video Event Understanding [chapter]

Tian Lan, Lei Chen, Zhiwei Deng, Guang-Tong Zhou, Greg Mori
2015 Lecture Notes in Computer Science  
We learn action primitives and their interrelations in a multi-level spatiotemporal model for action recognition.  ...  In order to address this, we present an approach to discover action primitives, sub-categories of action classes, that allow us to model this intra-class variation.  ...  [26] use a probabilistic latent variable model for discovering action categories.  ... 
doi:10.1007/978-3-319-16199-0_7 fatcat:mxez3e3wr5cvblo2ibl65garsa

Attribute Learning for Understanding Unstructured Social Activity [chapter]

Yanwei Fu, Timothy M. Hospedales, Tao Xiang, Shaogang Gong
2012 Lecture Notes in Computer Science  
Recently, attribute learning has emerged as a promising paradigm for transferring learning to sparsely labelled classes in object or single-object short action classification.  ...  We show that our framework is able to exploit latent attributes to outperform contemporary approaches for addressing a variety of realistic multi-media sparse data learning tasks including: multi-task  ...  Discovering and learning those discriminative yet latent attributes thus becomes the key.  ... 
doi:10.1007/978-3-642-33765-9_38 fatcat:nsltcn7qyjcwfmlct67ce6h5du

Zero-shot multi-label learning via label factorisation

Hang Shao, Yuchen Guo, Guiguang Ding, Jungong Han
2018 IET Computer Vision  
The authors propose a novel learning framework based on label factorisation for this problem.  ...  The first is knowledge transfer that utilises information from seen classes to build recognition models for unseen classes.  ...  In multi-label learning, the matrix A is termed as decoding matrix or function that turns the latent representation of images into label matrix.  ... 
doi:10.1049/iet-cvi.2018.5131 fatcat:5shug5tmyvbqzncizpspcf5toi

Multi-Task Zero-Shot Action Recognition with Prioritised Data Augmentation [chapter]

Xun Xu, Timothy M. Hospedales, Shaogang Gong
2016 Lecture Notes in Computer Science  
Zero-Shot Learning (ZSL) promises to scale visual recognition by bypassing the conventional model training requirement of annotated examples for every category.  ...  Reusing the learned mapping to project target videos into an embedding space thus allows novel-classes to be recognised by nearest neighbour inference.  ...  learning problem with a lower-dimensional latent semantic embedding space for more effective matching  ... 
doi:10.1007/978-3-319-46475-6_22 fatcat:zmwhktds3vfndj6wh364qpsutu
« Previous Showing results 1 — 15 out of 21,329 results