Filters








2,326 Hits in 7.7 sec

Zero-Shot Learning on Semantic Class Prototype Graph

Zhenyong Fu, Tao Xiang, Elyor Kodirov, Shaogang Gong
2018 IEEE Transactions on Pattern Analysis and Machine Intelligence  
Zero-Shot Learning (ZSL) for visual recognition is typically achieved by exploiting a semantic embedding space.  ...  To overcome these problems, a novel manifold distance computed on a semantic class prototype graph is proposed which takes into account the rich intrinsic semantic structure, i.e., semantic manifold, of  ...  CONCLUSION We have introduced a novel zero-shot learning approach based on measuring a manifold distance between a test image and an unseen class prototype on a semantic class prototype graph.  ... 
doi:10.1109/tpami.2017.2737007 pmid:28796607 fatcat:glg5peo2h5gytp6fyktg6j7bay

Transductive Multi-view Embedding for Zero-Shot Recognition and Annotation [chapter]

Yanwei Fu, Timothy M. Hospedales, Tao Xiang, Zhenyong Fu, Shaogang Gong
2014 Lecture Notes in Computer Science  
Most existing zero-shot learning approaches exploit transfer learning via an intermediate-level semantic representation such as visual attributes or semantic word vectors.  ...  That is, due to having disjoint and potentially unrelated classes, the projection functions learned from the auxiliary dataset/domain are biased when applied directly to the target dataset/domain.  ...  Thus the difference between zero-shot and N-shot learning lies only on the initial labelled instances: Zero-shot learning has the prototypes as labelled nodes; N-shot has instances as labelled nodes; and  ... 
doi:10.1007/978-3-319-10605-2_38 fatcat:clv6flbs4najfoxbhyfucgsxhy

Transductive Multi-class and Multi-label Zero-shot Learning [article]

Yanwei Fu, Yongxin Yang, Timothy M. Hospedales, Tao Xiang, Shaogang Gong
2015 arXiv   pre-print
Recently, zero-shot learning (ZSL) has received increasing interest.  ...  applied directly to map each instance into the same semantic representation space where a zero-shot classifier is used to recognise the unseen target class instances with a single known 'prototype' of  ...  Comparison with the state-of-the-art on zero-shot learning on AwA, USAA and CUB.  ... 
arXiv:1503.07884v1 fatcat:or4zahtjj5atpoxks5n4dilega

Transductive Multi-View Zero-Shot Learning

Yanwei Fu, Timothy M. Hospedales, Tao Xiang, Shaogang Gong
2015 IEEE Transactions on Pattern Analysis and Machine Intelligence  
The second limitation is the prototype sparsity problem which refers to the fact that for each target class, only a single prototype is available for zero-shot learning given a semantic representation.  ...  Most existing zero-shot learning approaches exploit transfer learning via an intermediate-level semantic representation shared between an annotated auxiliary dataset and a target dataset with different  ...  (a) zero-shot learning on AwA using only hand-crafted features; (b) zero-shot learning on AwA using hand-crafted and deep features together; (c) zero-shot learning on USAA.  ... 
doi:10.1109/tpami.2015.2408354 pmid:26440271 fatcat:eazqbmoc6vholji7ke6yyis5wq

Learning Robust Visual-semantic Mapping for Zero-shot Learning [article]

Jingcai Guo
2021 arXiv   pre-print
Zero-shot learning (ZSL) aims at recognizing unseen class examples (e.g., images) with knowledge transferred from seen classes.  ...  When inferring, given unseen class examples, the learned mapping function is reused to them and recognizes the class labels on some metrics among their semantic relations.  ...  This chapter presents our study on the conversion of zero-shot learning to the graph recognition task, which is a fine-grained zero-shot learning framework based on the example-level graph.  ... 
arXiv:2104.05668v1 fatcat:uq5msriuovettbwar46qzjfcbm

Zero-shot Learning via Shared-Reconstruction-Graph Pursuit [article]

Bo Zhao, Xinwei Sun, Yuan Yao, Yizhou Wang
2017 arXiv   pre-print
Zero-shot learning (ZSL) aims to recognize objects from novel unseen classes without any training data.  ...  Our method can be easily extended to the generalized zero-shot learning setting. Experiments on three popular datasets show that our method outperforms other methods on all datasets.  ...  The Shared Reconstruction Graph is proposed for alleviating the space shift problem in ZSL. Unseen image prototypes are synthesized for classifying testing instances using the learned SRG.  ... 
arXiv:1711.07302v1 fatcat:tnryqwtigzeozcmfjisuaqeb3a

Semantic Graph for Zero-Shot Learning [article]

Zhen-Yong Fu, Tao Xiang, Shaogang Gong
2015 arXiv   pre-print
Zero-shot learning aims to classify visual objects without any training data via knowledge transfer between seen and unseen classes.  ...  After incorporating one test image into the semantic graph, the absorbing probabilities from the test data to each unseen class can be effectively computed; and zero-shot classification can be achieved  ...  Conclusion In this work, we have introduced a novel zero-shot learning framework based on semantic graph.  ... 
arXiv:1406.4112v2 fatcat:otfb65m4czabdcgp54g5bk6pqy

Zero-shot object recognition by semantic manifold distance

Zhenyong Fu, Tao A Xiang, Elyor Kodirov, Shaogang Gong
2015 2015 IEEE Conference on Computer Vision and Pattern Recognition (CVPR)  
Object recognition by zero-shot learning (ZSL) aims to recognise objects without seeing any visual examples by learning knowledge transfer between seen and unseen object classes.  ...  In this paper we propose to model the semantic manifold in an embedding space using a semantic class label graph.  ...  The semantic graph in our approach is only related to the seen/unseen class prototypes. Once the semantic graph is constructed, it is fixed and used in the subsequent zero-shot learning process.  ... 
doi:10.1109/cvpr.2015.7298879 dblp:conf/cvpr/FuXKG15 fatcat:2ophdrj72vfcffimbzrjbighgi

Isometric Propagation Network for Generalized Zero-shot Learning [article]

Lu Liu, Tianyi Zhou, Guodong Long, Jing Jiang, Xuanyi Dong, Chengqi Zhang
2021 arXiv   pre-print
Zero-shot learning (ZSL) aims to classify images of an unseen class only based on a few attributes describing that class but no access to any training sample.  ...  Specifically, IPN learns to propagate the class representations on an auto-generated graph within each space.  ...  B EXPERIMENTS ON ZERO-SHOT LEARNING SETTING The comparison between IPN to other baselines on the setting of Zero-shot Learning is shown in Table 3 .  ... 
arXiv:2102.02038v1 fatcat:n25uggnrwzgqdbc23ftkknztie

Explanatory Multi-Scale Adversarial Semantic Embedding Space Learning for Zero-Shot Recognition

Huiting Li
2022 Open Journal of Applied Sciences  
The goal of zero-shot recognition is to classify classes it has never seen before, which needs to build a bridge between seen and unseen classes through semantic embedding space.  ...  Therefore, semantic embedding space learning plays an important role in zero-shot recognition. Among existing works, semantic embedding space is mainly taken by user-defined attribute vectors.  ...  In group 3, the learned attributes are only discriminative on training set, and some attributes on zero-shot classes are ignored.  ... 
doi:10.4236/ojapps.2022.123023 fatcat:vg36r25yz5abjkgu2kre66dify

Information Symmetry Matters: A Modal-Alternating Propagation Network for Few-Shot Learning [article]

Zhong Ji, Zhishen Hou, Xiyao Liu, Yanwei Pang, Jungong Han
2021 arXiv   pre-print
Semantic information provides intra-class consistency and inter-class discriminability beyond visual concepts, which has been employed in Few-Shot Learning (FSL) to achieve further gains.  ...  Specifically, the MAP-Net transfers the neighbor information by the graph propagation to generate the pseudo-semantics for unlabeled samples guided by the completed visual relationships and rectify the  ...  Based on the success of zero-shot learning, some methods utilizing auxiliary semantic information are proposed to boost the few-shot learning in recent years. Xing et al.  ... 
arXiv:2109.01295v1 fatcat:jrbcrmly5nd6tfk43il2cttan4

Independent Prototype Propagation for Zero-Shot Compositionality [article]

Frank Ruis, Gertjan Burghouts, Doina Bucur
2021 arXiv   pre-print
The method does not rely on any external data, such as class hierarchy graphs or pretrained word embeddings.  ...  Humans are good at compositional zero-shot reasoning; someone who has never seen a zebra before could nevertheless recognize one when we tell them it looks like a horse with black and white stripes.  ...  By initializing the node features with independent prototypes, the graph is able to learn the dependencies encoded by the compositional graph, including the novel zero-shot classes, instead of the biases  ... 
arXiv:2106.00305v2 fatcat:pde7b2ibtra5lh6jmon55e37oa

Relational Generalized Few-Shot Learning [article]

Xiahan Shi, Leonard Salewski, Martin Schiegg, Zeynep Akata, Max Welling
2020 arXiv   pre-print
Although this few-shot learning setup has received a lot of attention recently, most proposed methods focus on discriminating novel classes only.  ...  Our model Graph-convolutional Global Prototypical Networks (GcGPN) incorporates these inter-class relations using graph-convolution in order to embed novel class representations into the existing space  ...  TGG: Transferable graph generation for zero-shot and few-shot learning.  ... 
arXiv:1907.09557v2 fatcat:lzmwzrkupvbj5gljruejozt7xu

Recent Advances in Zero-shot Recognition [article]

Yanwei Fu, Tao Xiang, Yu-Gang Jiang, Xiangyang Xue, Leonid Sigal, and Shaogang Gong
2017 arXiv   pre-print
One approach to scaling up the recognition is to develop models capable of recognizing unseen categories without any training instances, or zero-shot recognition/ learning.  ...  We also overview related recognition tasks including one-shot and open set recognition which can be used as natural extensions of zero-shot recognition when limited number of class samples become available  ...  Yanwei Fu is supported by The Program for Professor of Special Appointment (Eastern Scholar) at Shanghai Institutions of Higher Learning.  ... 
arXiv:1710.04837v1 fatcat:u3mp6dgj2rgqrarjm4dcywegmy

Expanding Semantic Knowledge for Zero-shot Graph Embedding [article]

Zheng Wang, Ruihang Shao, Changping Wang, Changjun Hu, Chaokun Wang, Zhiguo Gong
2021 arXiv   pre-print
Zero-shot graph embedding is a major challenge for supervised graph learning.  ...  We show that its core part is a GNN prototypical model in which a class prototype is described by its mean feature vector.  ...  However, little work has considered the zero-shot graph embedding (ZGE) problem where some classes have no labeled data at the training time.  ... 
arXiv:2103.12491v1 fatcat:l7vybuupmrab5amwz5vqvbzs7y
« Previous Showing results 1 — 15 out of 2,326 results