Do Different Deep Metric Learning Losses Lead to Similar Learned Features?

• URL: http://arxiv.org/abs/2205.02698v1
• Date: Thu, 5 May 2022 15:07:19 GMT
• Title: Do Different Deep Metric Learning Losses Lead to Similar Learned Features?
• Authors: Konstantin Kobs, Michael Steininger, Andrzej Dulny, Andreas Hotho
• Abstract summary: We compare 14 pretrained models from a recent study and find that, even though all models perform similarly, different loss functions can guide the model to learn different features. Our analysis also shows that some seemingly irrelevant properties can have significant influence on the resulting embedding.
• Score: 4.043200001974071
• License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/
• Abstract: Recent studies have shown that many deep metric learning loss functions perform very similarly under the same experimental conditions. One potential reason for this unexpected result is that all losses let the network focus on similar image regions or properties. In this paper, we investigate this by conducting a two-step analysis to extract and compare the learned visual features of the same model architecture trained with different loss functions: First, we compare the learned features on the pixel level by correlating saliency maps of the same input images. Second, we compare the clustering of embeddings for several image properties, e.g. object color or illumination. To provide independent control over these properties, photo-realistic 3D car renders similar to images in the Cars196 dataset are generated. In our analysis, we compare 14 pretrained models from a recent study and find that, even though all models perform similarly, different loss functions can guide the model to learn different features. We especially find differences between classification and ranking based losses. Our analysis also shows that some seemingly irrelevant properties can have significant influence on the resulting embedding. We encourage researchers from the deep metric learning community to use our methods to get insights into the features learned by their proposed methods.

Related papers

• Learning the Relation between Similarity Loss and Clustering Loss in Self-Supervised Learning [27.91000553525992]
  Self-supervised learning enables networks to learn discriminative features from massive data itself. We analyze the relation between similarity loss and feature-level cross-entropy loss. We provide theoretical analyses and experiments to show that a suitable combination of these two losses can get state-of-the-art results.
  arXiv  Detail & Related papers  (2023-01-08T13:30:39Z)
• Taxonomizing local versus global structure in neural network loss landscapes [60.206524503782006]
  We show that the best test accuracy is obtained when the loss landscape is globally well-connected. We also show that globally poorly-connected landscapes can arise when models are small or when they are trained to lower quality data.
  arXiv  Detail & Related papers  (2021-07-23T13:37:14Z)
• Projected Distribution Loss for Image Enhancement [15.297569497776374]
  We show that aggregating 1D-Wasserstein distances between CNN activations is more reliable than the existing approaches. In imaging applications such as denoising, super-resolution, demosaicing, deblurring and JPEG artifact removal, the proposed learning loss outperforms the current state-of-the-art on reference-based perceptual losses.
  arXiv  Detail & Related papers  (2020-12-16T22:13:03Z)
• Intriguing Properties of Contrastive Losses [12.953112189125411]
  We study three intriguing properties of contrastive learning. We study if instance-based contrastive learning can learn well on images with multiple objects present. We show that, for contrastive learning, a few bits of easy-to-learn shared features can suppress, and even fully prevent, the learning of other sets of competing features.
  arXiv  Detail & Related papers  (2020-11-05T13:19:48Z)
• Stereopagnosia: Fooling Stereo Networks with Adversarial Perturbations [71.00754846434744]
  We show that imperceptible additive perturbations can significantly alter the disparity map. We show that, when used for adversarial data augmentation, our perturbations result in trained models that are more robust.
  arXiv  Detail & Related papers  (2020-09-21T19:20:09Z)
• Learning Condition Invariant Features for Retrieval-Based Localization from 1M Images [85.81073893916414]
  We develop a novel method for learning more accurate and better generalizing localization features. On the challenging Oxford RobotCar night condition, our method outperforms the well-known triplet loss by 24.4% in localization accuracy within 5m.
  arXiv  Detail & Related papers  (2020-08-27T14:46:22Z)
• Towards Visually Explaining Similarity Models [29.704524987493766]
  We present a method to generate gradient-based visual attention for image similarity predictors. By relying solely on the learned feature embedding, we show that our approach can be applied to any kind of CNN-based similarity architecture. We show that our resulting attention maps serve more than just interpretability; they can be infused into the model learning process itself with new trainable constraints.
  arXiv  Detail & Related papers  (2020-08-13T17:47:41Z)
• Few-shot Visual Reasoning with Meta-analogical Contrastive Learning [141.2562447971]
  We propose to solve a few-shot (or low-shot) visual reasoning problem, by resorting to analogical reasoning. We extract structural relationships between elements in both domains, and enforce them to be as similar as possible with analogical learning. We validate our method on RAVEN dataset, on which it outperforms state-of-the-art method, with larger gains when the training data is scarce.
  arXiv  Detail & Related papers  (2020-07-23T14:00:34Z)
• Unsupervised Landmark Learning from Unpaired Data [117.81440795184587]
  Recent attempts for unsupervised landmark learning leverage synthesized image pairs that are similar in appearance but different in poses. We propose a cross-image cycle consistency framework which applies the swapping-reconstruction strategy twice to obtain the final supervision. Our proposed framework is shown to outperform strong baselines by a large margin.
  arXiv  Detail & Related papers  (2020-06-29T13:57:20Z)
• Distilling Localization for Self-Supervised Representation Learning [82.79808902674282]
  Contrastive learning has revolutionized unsupervised representation learning. Current contrastive models are ineffective at localizing the foreground object. We propose a data-driven approach for learning in variance to backgrounds.
  arXiv  Detail & Related papers  (2020-04-14T16:29:42Z)
• Disentangling Image Distortions in Deep Feature Space [20.220653544354285]
  We take a step in the direction of a broader understanding of perceptual similarity by analyzing the capability of deep visual representations to intrinsically characterize different types of image distortions. A dimension-reduced representation of the features extracted from a given layer permits to efficiently separate types of distortions in the feature space. Each network layer exhibits a different ability to separate between different types of distortions, and this ability varies according to the network architecture.
  arXiv  Detail & Related papers  (2020-02-26T11:02:13Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.