Crafting Better Contrastive Views for Siamese Representation Learning

Peng, Xiangyu; Wang, Kai; Zhu, Zheng; Wang, Mang; You, Yang

doi:10.1109/cvpr52688.2022.01556

Cited by 72 publications

(18 citation statements)

References 24 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…For example, D w was designed to be large enough for self-supervised learning, which has received surprisingly little attention in the WSOL community [9]. We are also interested in using iNatLoc500 to study whether self-supervised learning methods can be improved by using WSOL methods to select crops [40], especially in the context of fine-grained data [15]. For the object detection community, the clean boxes in iNatLoc500 can (i) serve as a test set for object detectors trained on the noisy iNat17 boxes, (ii) be used to study the problem of learning multi-instance detectors from one box per image, and (iii) be used to analyze the role of label granularity in object detection.…”

Section: Discussionmentioning

confidence: 99%

On Label Granularity and Object Localization

Cole

Wilber

Horn

et al. 2022

Lecture Notes in Computer Science

View full text Add to dashboard Cite

Weakly supervised object localization (WSOL) aims to learn representations that encode object location using only image-level category labels. However, many objects can be labeled at different levels of granularity. Is it an animal, a bird, or a great horned owl? Which image-level labels should we use? In this paper we study the role of label granularity in WSOL. To facilitate this investigation we introduce iNat-Loc500, a new large-scale fine-grained benchmark dataset for WSOL. Surprisingly, we find that choosing the right training label granularity provides a much larger performance boost than choosing the best WSOL algorithm. We also show that changing the label granularity can significantly improve data efficiency.

show abstract

Section: Discussionmentioning

confidence: 99%

On Label Granularity and Object Localization

Cole

Wilber

Horn

et al. 2022

Lecture Notes in Computer Science

View full text Add to dashboard Cite

show abstract

“…This was followed by papers that simplified this approach [13][14][15]37] and proposed non-contrastive variants [16,33]. While those approaches have been suc-cessful, the utility of augmentation-based self-supervised learning has been questioned [65,102] with follow-up work proposing the use of objectness [63,72] and saliency [77] to alleviate some of those concerns. While we share the motivation for visual representation learning, we also question the reliance on image augmentation and propose using language sampling to learn better invariances.…”

Section: Related Workmentioning

confidence: 99%

Learning Visual Representations via Language-Guided Sampling

Banani¹,

Desai²,

Johnson³

2023

Preprint

View full text Add to dashboard Cite

Although an object may appear in numerous contexts, we often describe it in a limited number of ways. This happens because language abstracts away visual variation to represent and communicate concepts. Building on this intuition, we propose an alternative approach to visual learning: using language similarity to sample semantically similar image pairs for contrastive learning. Our approach deviates from image-based contrastive learning by using language to sample pairs instead of hand-crafted augmentations or learned clusters. Our approach also deviates from image-text contrastive learning by relying on pre-trained language models to guide the learning rather than minimize a cross-modal similarity. Through a series of experiments, we show that language-guided learning can learn better features than both image-image and image-text representation learning approaches.

show abstract

“…We follow the code implementation and hyper-parameter of [68]. For small datasets (i.e., CIFAR-10/100 and Tiny ImageNet), we use the same training setup in all experiments.…”

Section: Implementation Detailsmentioning

confidence: 99%

“…Since the data size is larger, we train SimCLR and our method with the batch size of 1024 and cosine-annealed learning rate of 0.6 for faster convergence. We adopt the same setting as in [68] for training the linear classifier.…”

Section: Implementation Detailsmentioning

confidence: 99%

Hyperbolic Contrastive Learning

Yen¹,

Lin²,

Yamada³

et al. 2023

Preprint

View full text Add to dashboard Cite

Learning good image representations that are beneficial to downstream tasks is a challenging task in computer vision. As such, a wide variety of self-supervised learning approaches have been proposed. Among them, contrastive learning has shown competitive performance on several benchmark datasets. The embeddings of contrastive learning are arranged on a hypersphere that results in using the inner (dot) product as a distance measurement in Euclidean space. However, the underlying structure of many scientific fields like social networks, brain imaging, and computer graphics data exhibit highly non-Euclidean latent geometry. We propose a novel contrastive learning framework to learn semantic relationships in the hyperbolic space. Hyperbolic space is a continuous version of trees that naturally owns the ability to model hierarchical structures and is thus beneficial for efficient contrastive representation learning. We also extend the proposed Hyperbolic Contrastive Learning (HCL) to the supervised domain and studied the adversarial robustness of HCL. The comprehensive experiments show that our proposed method achieves better results on selfsupervised pretraining, supervised classification, and higher robust accuracy than baseline methods.

show abstract

Crafting Better Contrastive Views for Siamese Representation Learning

Cited by 72 publications

References 24 publications

On Label Granularity and Object Localization

On Label Granularity and Object Localization

Learning Visual Representations via Language-Guided Sampling

Hyperbolic Contrastive Learning

Contact Info

Product

Resources

About