Filters








57,822 Hits in 3.3 sec

Few-Shot Class-Incremental Learning by Sampling Multi-Phase Tasks [article]

Da-Wei Zhou, Han-Jia Ye, De-Chuan Zhan
2022 arXiv   pre-print
The task of recognizing few-shot new classes without forgetting old classes is called few-shot class-incremental learning (FSCIL).  ...  The calibration module also adaptively contextualizes the instance-specific embedding with a set-to-set function.  ...  Incremental Tasks 𝒮𝒮 1 Fake-Incremental Task 1 Base Task Training Set 𝒟𝒟 0 𝒮𝒮 2 𝒬𝒬 1 𝒬𝒬 2 Learning Generalizable Features Eq. 4 adapts to new classes without harming current embedding.  ... 
arXiv:2203.17030v1 fatcat:7ooyl4xn6fdnfku62a5euxzwpu

Rectification-based Knowledge Retention for Continual Learning [article]

Pravendra Singh, Pratik Mazumder, Piyush Rai, Vinay P. Namboodiri
2021 arXiv   pre-print
The task incremental learning problem becomes even more challenging when the test set contains classes that are not part of the train set, i.e., a task incremental generalized zero-shot learning problem  ...  Deep learning models suffer from catastrophic forgetting when trained in an incremental learning setting.  ...  CADA-VAE learns a common latent embedding space for both the image/visual features and the class/attribute embeddings and brings the latent embeddings of the image features and class embeddings closer  ... 
arXiv:2103.16597v1 fatcat:fi56ub3kaffihpg2apixsvzmg4

Expanding object detector's Horizon: Incremental learning framework for object detection in videos

Alina Kuznetsova, Sung Ju Hwang, Bodo Rosenhahn, Leonid Sigal
2015 2015 IEEE Conference on Computer Vision and Pattern Recognition (CVPR)  
Our detection model consists of an embedding space and multiple class prototypes in that embedding space, that represent object classes; distance to those prototypes allows us to reason about multi-class  ...  To this end, we develop a new scalable and accurate incremental object detection algorithm, based on several extensions of large-margin embedding (LME).  ...  In continuous domain adaptation [15] (and incremental learning [28, 29] ) the goal is to continuously adapt (or learn) a fixed complexity model to perform as accurately as possible on the arriving target  ... 
doi:10.1109/cvpr.2015.7298597 dblp:conf/cvpr/KuznetsovaHRS15 fatcat:yx2ao3zzpfe7ni4zv5cjfcn4gm

Semantic Drift Compensation for Class-Incremental Learning

Lu Yu, Bartlomiej Twardowski, Xialei Liu, Luis Herranz, Kai Wang, Yongmei Cheng, Shangling Jui, Joost van de Weijer
2020 2020 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR)  
Class-incremental learning of deep networks sequentially increases the number of classes to be classified.  ...  Embedding networks have the advantage that new classes can be naturally included into the network without adding new weights. Therefore, we study incremental learning for embedding networks.  ...  Continual Learning for Embeddings We consider a class-incremental learning setup where a network learns several tasks, each task containing a number of new classes.  ... 
doi:10.1109/cvpr42600.2020.00701 dblp:conf/cvpr/0004TLHWCJ020 fatcat:fpdphkv72jcmha2nhvhs365onu

Incremental user embedding modeling for personalized text classification [article]

Ruixue Lian, Che-Wei Huang, Yuqing Tang, Qilong Gu, Chengyuan Ma, Chenlei Guo
2022 arXiv   pre-print
Adaptive user representation learning by utilizing user personalized information has become increasingly challenging due to ever-growing history data.  ...  In this work, we propose an incremental user embedding modeling approach, in which embeddings of user's recent interaction histories are dynamically integrated into the accumulated history vectors via  ...  Incremental User Embedding Learning The core idea of incremental embedding learning method is to dynamically integrate recent comment history embeddings into the accumulated history vectors via an upper-layer  ... 
arXiv:2202.06369v1 fatcat:x33ttieprnfxdipccioikz4zlu

Lifelong Adaptive Machine Learning for Sensor-based Human Activity Recognition Using Prototypical Networks [article]

Rebecca Adaimi, Edison Thomaz
2022 arXiv   pre-print
Moreover, analysis has so far focused on task-incremental or class-incremental learning paradigms where task boundaries are known.  ...  Online learning is further facilitated using contrastive loss to enforce inter-class separation.  ...  adapted for incremental continual learning (Figure 2 ).  ... 
arXiv:2203.05692v1 fatcat:xgtaxzixsrdhhd7xcs7h4y7o7u

Semantic Drift Compensation for Class-Incremental Learning [article]

Lu Yu, Bartłomiej Twardowski, Xialei Liu, Luis Herranz, Kai Wang, Yongmei Cheng, Shangling Jui, Joost van de Weijer
2020 arXiv   pre-print
Class-incremental learning of deep networks sequentially increases the number of classes to be classified.  ...  Embedding networks have the advantage that new classes can be naturally included into the network without adding new weights. Therefore, we study incremental learning for embedding networks.  ...  Continual Learning for Embeddings We consider a class-incremental learning setup where a network learns several tasks, each task containing a number of new classes.  ... 
arXiv:2004.00440v1 fatcat:fw3e7rh5njgfddkvdtxsyzihse

MgSvF: Multi-Grained Slow vs. Fast Framework for Few-Shot Class-Incremental Learning [article]

Hanbin Zhao, Yongjian Fu, Mintong Kang, Qi Tian, Fei Wu, Xi Li
2021 arXiv   pre-print
As a challenging problem, few-shot class-incremental learning (FSCIL) continually learns a sequence of tasks, confronting the dilemma between slow forgetting of old knowledge and fast adaptation to new  ...  concentrate on this "slow vs. fast" (SvF) dilemma to determine which knowledge components to be updated in a slow fashion or a fast fashion, and thereby balance old-knowledge preservation and new-knowledge adaptation  ...  Incremental learning is usually conducted under the task-incremental [70] , [71] or the class-incremental learning scenarios [2] , [68] , [72] - [75] .  ... 
arXiv:2006.15524v3 fatcat:74vw2x2trba65ddiq2jii2ee2q

Bayesian Embeddings for Few-Shot Open World Recognition [article]

John Willes, James Harrison, Ali Harakeh, Chelsea Finn, Marco Pavone, Steven Waslander
2021 arXiv   pre-print
We combine Bayesian non-parametric class priors with an embedding-based pre-training scheme to yield a highly flexible framework which we refer to as few-shot learning for open world recognition (FLOWR  ...  In this work we extend embedding-based few-shot learning algorithms to the open-world recognition setting.  ...  [39] consider a few-shot approach to incremental learning in which a base classifier is paired with a standard few-shot learner.  ... 
arXiv:2107.13682v1 fatcat:odbdstuw2vavvhuue5bwoj2kae

Incremental Few-Shot Object Detection for Robotics [article]

Yiting Li, Haiyue Zhu, Sichao Tian, Fan Feng, Jun Ma, Chek Sing Teo, Cheng Xiang, Prahlad Vadakkepat, Tong Heng Lee
2022 arXiv   pre-print
Firstly, to best preserve performance on the pre-trained base classes, we propose a novel Dual-Embedding-Space (DES) architecture which decouples the representation learning of base and novel categories  ...  Incremental few-shot learning is highly expected for practical robotics applications.  ...  Specifically, for the n-th learning task, consider its training set D n = {(I i , c i )} where I i denotes the image sample set for new class c i ∈ C t .  ... 
arXiv:2005.02641v2 fatcat:6dfgqbwj5ncwlo36fpblslirwa

SemGIF: A Semantics Guided Incremental Few-shot Learning Framework with Generative Replay

S. Divakar Bhat, Biplab Banerjee, Subhasis Chaudhuri
2021 British Machine Vision Conference  
Considering the importance of modeling a discriminative feature space in IFSL for separating the base and the novel classes, we propose a feature augmentation strategy where the visual embeddings are supplemented  ...  Existing techniques prefer to preserve some base class samples to tackle forgetting, which does not comply with the intention of incremental learning.  ...  The notion of incremental or continual learning [26, 37] is deduced to handle such non-stationary setups where the model is required to continuously adapt to new tasks while judiciously controlling the  ... 
dblp:conf/bmvc/BhatBC21 fatcat:fvh2rivqbrbcjjdiqkpbt5hdzu

Forward Compatible Few-Shot Class-Incremental Learning [article]

Da-Wei Zhou, Fu-Yun Wang, Han-Jia Ye, Liang Ma, Shiliang Pu, De-Chuan Zhan
2022 arXiv   pre-print
This scenario becomes more challenging when new class instances are insufficient, which is called few-shot class-incremental learning (FSCIL).  ...  Current methods handle incremental learning retrospectively by making the updated model similar to the old one.  ...  The graph model learns to adapt the embeddings of old class prototypes and new class prototypes, and such ability is generalizable to the incremental learning process.  ... 
arXiv:2203.06953v1 fatcat:u3hadxewirbaxmag2bpjzijvm4

Towards Generalized and Incremental Few-Shot Object Detection [article]

Yiting Li, Haiyue Zhu, Jun Ma, Chek Sing Teo, Cheng Xiang, Prahlad Vadakkepat, Tong Heng Lee
2021 arXiv   pre-print
Real-world object detection is highly desired to be equipped with the learning expandability that can enlarge its detection classes incrementally.  ...  In this paper, to address the above incremental few-shot learning issues, a novel Incremental Few-Shot Object Detection (iFSOD) method is proposed to enable the effective continual learning from few-shot  ...  Conventional incremental learning approaches Compared with the general incremental learning method, our method is more suitable for the sequential adaptation with small training sets.  ... 
arXiv:2109.11336v1 fatcat:6cjeqmznazbutnxniog6b5rmh4

Toward Fine-grained Image Retrieval with Adaptive Deep Learning for Cultural Heritage Image

Sathit Prasomphan
2023 Computer systems science and engineering  
In this step, the original class is not necessary for re-training which we call an adaptive deep learning technique.  ...  The goal of the proposed method is to perform a retrieval task for classes. Incremental learning for new classes was conducted to reduce the re-training process.  ...  However, old groups of images were not considered in the hash code learning process.  ... 
doi:10.32604/csse.2023.025293 fatcat:ggtexahsv5fr5ax2cpz2hnsa6u

Self-Promoted Prototype Refinement for Few-Shot Class-Incremental Learning [article]

Kai Zhu, Yang Cao, Wei Zhai, Jie Cheng, Zheng-Jun Zha
2021 arXiv   pre-print
Few-shot class-incremental learning is to recognize the new classes given few samples and not forget the old classes.  ...  To address this problem, we propose a novel incremental prototype learning scheme.  ...  class-incremental learning (CIL).  ... 
arXiv:2107.08918v1 fatcat:xflgrzwl3bbwhaditcadipag4u
« Previous Showing results 1 — 15 out of 57,822 results