Representation Learning by Learning to Count

Noroozi, Mehdi; Pirsiavash, Hamed; Favaro, Paolo

doi:10.1109/iccv.2017.628

Cited by 362 publications

(262 citation statements)

References 39 publications

Supporting

Mentioning

262

Contrasting

Order By: Relevance

“…Unsupervised learning through learning representation space that successfully reconstructs samples is widely used for a variety of tasks, including classification [26], denosing [57], and in-painting [43]. Conventional methods for unsupervised representation learning are usually based on a pretext task such as reconstruction of static images [38] or videos [59]. Learning to reconstruct data was used for tasks like de-nsoing [57], in-painting [42], image refinement for defense against adversarial example [53], and for one-class classifiers [46,51,52].…”

Section: Related Workmentioning

confidence: 99%

“…Other types of pretext tasks proposed for unsupervised learning include understanding the correct order of video frames [6,36] or predicting the spatial relation between image patches [12], e.g., jigsaw puzzle solving as a pretext task was exploited by Noorozi and Favaro [37]. In another work, Noroozi et al [38] proposed to train an unsupervised model by counting the primitive elements of images. Pathak et al [41] proposed a model to segment an image into foreground and background.…”

Section: Related Workmentioning

confidence: 99%

See 1 more Smart Citation

Self-Supervised Representation Learning via Neighborhood-Relational Encoding

Sabokrou

Khalooei

Adeli

2019

2019 IEEE/CVF International Conference on Computer Vision (ICCV)

View full text Add to dashboard Cite

In this paper, we propose a novel self-supervised representation learning by taking advantage of a neighborhoodrelational encoding (NRE) among the training data. Conventional unsupervised learning methods only focused on training deep networks to understand the primitive characteristics of the visual data, mainly to be able to reconstruct the data from a latent space. They often neglected the relation among the samples, which can serve as an important metric for self-supervision. Different from the previous work, NRE aims at preserving the local neighborhood structure on the data manifold. Therefore, it is less sensitive to outliers. We integrate our NRE component with an encoder-decoder structure for learning to represent samples considering their local neighborhood information. Such discriminative and unsupervised representation learning scheme is adaptable to different computer vision tasks due to its independence from intense annotation requirements. We evaluate our proposed method for different tasks, including classification, detection, and segmentation based on the learned latent representations. In addition, we adopt the auto-encoding capability of our proposed method for applications like defense against adversarial example attacks and video anomaly detection. Results confirm the performance of our method is better or at least comparable with the state-of-the-art for each specific application, but with a generic and self-supervised approach.

show abstract

Section: Related Workmentioning

confidence: 99%

Section: Related Workmentioning

confidence: 99%

Self-Supervised Representation Learning via Neighborhood-Relational Encoding

Sabokrou

Khalooei

Adeli

2019

2019 IEEE/CVF International Conference on Computer Vision (ICCV)

View full text Add to dashboard Cite

show abstract

“…The empirical analysis of [33,20] suggests that the performance of recent deep networks is not yet saturated with respect to the size of training data. For this reason, learning methods from semi-supervised learning [42,39,33,20] to unsupervised learning [1,7,58,38] are attracting attention along with weakly-labeled or unlabeled large-scale data.…”

Section: Introductionmentioning

confidence: 99%

Learning Loss for Active Learning

Yoo¹,

Kweon

2019

2019 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR)

554

520

View full text Add to dashboard Cite

The performance of deep neural networks improves with more annotated data. The problem is that the budget for annotation is limited. One solution to this is active learning, where a model asks human to annotate data that it perceived as uncertain. A variety of recent methods have been proposed to apply active learning to deep networks but most of them are either designed specific for their target tasks or computationally inefficient for large networks. In this paper, we propose a novel active learning method that is simple but task-agnostic, and works efficiently with the deep networks. We attach a small parametric module, named "loss prediction module," to a target network, and learn it to predict target losses of unlabeled inputs. Then, this module can suggest data that the target model is likely to produce a wrong prediction. This method is task-agnostic as networks are learned from a single loss regardless of target tasks. We rigorously validate our method through image classification, object detection, and human pose estimation, with the recent network architectures. The results demonstrate that our method consistently outperforms the previous methods over the tasks.

show abstract

“…A very common solution is that of measuring the source prediction uncertainty on the target data with an entropy loss which is minimized during training [16,18]. A recent stream of works has introduced techniques to extract self-supervisory signals from unlabeled data as the patch relative position [9,22], counting primitives [23], or image coloring [33]. They capture invariances and regularities that allow to train models useful as fine-tuning priors, and those information appear also independent from the specific visual domain of the data from which they are obtained.…”

Section: Related Workmentioning

confidence: 99%

Tackling Partial Domain Adaptation with Self-supervision

Bucci

D’Innocente

Tommasi

2019

Lecture Notes in Computer Science

View full text Add to dashboard Cite

Domain adaptation approaches have shown promising results in reducing the marginal distribution difference among visual domains. They allow to train reliable models that work over datasets of different nature (photos, paintings etc.), but they still struggle when the domains do not share an identical label space. In the partial domain adaptation setting, where the target covers only a subset of the source classes, it is challenging to reduce the domain gap without incurring in negative transfer. Many solutions just keep the standard domain adaptation techniques by adding heuristic sample weighting strategies. In this work we show how the self-supervisory signal obtained from the spatial co-location of patches can be used to define a side task that supports adaptation regardless of the exact label sharing condition across domains. We build over a recent work that introduced a jigsaw puzzle task for domain generalization: we describe how to reformulate this approach for partial domain adaptation and we show how it boosts existing adaptive solutions when combined with them. The obtained experimental results on three datasets supports the effectiveness of our approach.

show abstract

Representation Learning by Learning to Count

Cited by 362 publications

References 39 publications

Self-Supervised Representation Learning via Neighborhood-Relational Encoding

Self-Supervised Representation Learning via Neighborhood-Relational Encoding

Learning Loss for Active Learning

Tackling Partial Domain Adaptation with Self-supervision

Contact Info

Product

Resources

About