Hyperbolic Image Embeddings

Khrulkov, Valentin; Mirvakhabova, Leyla; Ustinova, Evgeniya; Oseledets, Ivan; Lempitsky, Victor

doi:10.1109/cvpr42600.2020.00645

Cited by 208 publications

(176 citation statements)

References 26 publications

Supporting

Mentioning

165

Contrasting

Order By: Relevance

“…Due to the specific properties of the hyperbolic space, however, the usual Euclidean mean is not applicable and thus a generalisation is needed. In hyperbolic geometry, the averaging of feature vectors is done by using the Einstein midpoint [20]. The Einstein midpoint takes its simplest form in Klein coordinates and is defined as follows:…”

Section: Implementation Of a Recommender System Using Hyperbolic Distancementioning

confidence: 99%

Incorporating Domain Knowledge into Health Recommender Systems Using Hyperbolic Embeddings

Peito

Han²

2021

Studies in Computational Intelligence

View full text Add to dashboard Cite

In contrast to many other domains, recommender systems in health services may benefit particularly from the incorporation of health domain knowledge, as it helps to provide meaningful and personalised recommendations catering to the individual's health needs. With recent advances in representation learning enabling the hierarchical embedding of health knowledge into the hyperbolic Poincaré space, this work proposes a content-based recommender system for patient-doctor matchmaking in primary care based on patients' health profiles, enriched by pre-trained Poincaré embeddings of the ICD-9 codes through transfer learning. The proposed model outperforms its conventional counterpart in terms of recommendation accuracy and has several important business implications for improving the patient-doctor relationship.

show abstract

Section: Implementation Of a Recommender System Using Hyperbolic Distancementioning

confidence: 99%

Incorporating Domain Knowledge into Health Recommender Systems Using Hyperbolic Embeddings

Peito

Han²

2021

Studies in Computational Intelligence

View full text Add to dashboard Cite

show abstract

“…The central assumption in this work is that the hyperbolic neural networks can improve GAN's performance, either in the generation or discrimination process, because the hyperbolic layers can leverage the hierarchical characteristics of the images [19]. The HGAN is a family of architectures derived from modifying the original GAN architecture proposed by Goodfellow et al [24].…”

Section: Hyperbolic Gan (Hgan)mentioning

confidence: 99%

“…The Hyperbolic space can leverage the hierarchy of ImageNet as shown in figure 2, where WordNet-ImageNet mammals tree embedded in a two-dimensional Poincaré Ball using the method proposed in [16]. This argument was first wielded in [19] where they claim that the hierarchical semantic structure of language concepts can also be present in the images of those concepts, as in our example with mammals in figure 2. In that work, they modify the last section of three network architectures adding hyperbolic embedding to a Poincaré space followed by hyperbolic layers for the few-shot learning task.…”

Section: Introductionmentioning

confidence: 99%

HGAN: Hyperbolic Generative Adversarial Network

2021

View full text Add to dashboard Cite

Recently, Hyperbolic Spaces in the context of Non-Euclidean Deep Learning have gained popularity because of their ability to represent hierarchical data. We propose that it is possible to take advantage of the hierarchical characteristic present in the images by using hyperbolic neural networks in a GAN architecture. In this study, different configurations using fully connected hyperbolic layers in the GAN, WGAN, CGAN, and the mapping network of the StyleGAN2 are tested in what we call the HGAN, HWGAN, HCGAN, and HStyleGAN, respectively. Furthermore, we test multiple values of curvature and introduce an exponential way to train it. The results are measured using the Inception Score (IS) and the Fréchet Inception Distance (FID) over the MNIST dataset and with FID over CIFAR-10. Depending on the configuration and space curvature, better results are achieved for each proposed hyperbolic version than their euclidean counterpart.

show abstract

“…Other studies proposed Poincaré embedding, which employs a hyperbolic space as the embedding space [9], [24]. In a hyperbolic space, the distance grows exponentially from the origin to the boundary.…”

Section: Conventional and Hierarchical Embeddingmentioning

confidence: 99%

Target-Oriented Deformation of Visual-Semantic Embedding Space

Matsubara

2021

IEICE Trans. Inf. & Syst.

View full text Add to dashboard Cite

Hyperbolic Image Embeddings

Cited by 208 publications

References 26 publications

Incorporating Domain Knowledge into Health Recommender Systems Using Hyperbolic Embeddings

Incorporating Domain Knowledge into Health Recommender Systems Using Hyperbolic Embeddings

HGAN: Hyperbolic Generative Adversarial Network

Target-Oriented Deformation of Visual-Semantic Embedding Space

Contact Info

Product

Resources

About