Distance-based Hyperspherical Classification for Multi-source Open-Set Domain Adaptation

  title={Distance-based Hyperspherical Classification for Multi-source Open-Set Domain Adaptation},
  author={Silvia Bucci and Francesco Cappio Borlino and Barbara Caputo and Tatiana Tommasi},
  journal={2022 IEEE/CVF Winter Conference on Applications of Computer Vision (WACV)},
Vision systems trained in closed-world scenarios fail when presented with new environmental conditions, new data distributions, and novel classes at deployment time. How to move towards open-world learning is a long-standing research question. The existing solutions mainly focus on specific aspects of the problem (single domain Open-Set, multi-domain Closed-Set), or propose complex strategies which combine several losses and manually tuned hyperparameters. In this work, we tackle multi-source… 

Figures and Tables from this paper

Interpretable Open-Set Domain Adaptation via Angular Margin Separation

This work proposes a representation learning framework termed Angular Margin Separation (AMS) that unveils the power of discriminative and robust representation for both open-set domain adaptation and cross-domain semantic recovery and proposes a post-training re-projection that boosts the performance of seen classes interpretation without deterioration on unseen classes.

Simplifying Open-Set Video Domain Adaptation with Contrastive Learning

This work proposes to address OUVDA with an unified contrastive learning framework that learns discriminative and well-clustered features and proposes a video-oriented temporal contrastive loss that enables the method to better cluster the feature space by exploiting the freely available temporal information in video data.

Contrastive Learning for Cross-Domain Open World Recognition

This work proposes the first learning approach that deals with all the previously mentioned challenges at once by exploiting a single contrastive objective and shows how it learns a feature space perfectly suitable to incrementally include new classes and is able to capture knowledge which generalizes across a variety of visual domains.

How to exploit hyperspherical embeddings for out-of-distribution detection?

This work proposes CIDER, a novel representation learning framework that exploits hyperspherical embeddings for OOD detection and jointly optimizes two losses to promote strong ID-OOD separability: a dispersion loss that promotes large angular distances among different class prototypes, and a compactness loss that encourages samples to be close to their class prototypes.



Progressive Graph Learning for Open-Set Domain Adaptation

This paper introduces an end-to-end Progressive Graph Learning (PGL) framework where a graph neural network with episodic training is integrated to suppress underlying conditional shift and adversarial learning is adopted to close the gap between the source and target distributions.

Exploring Category-Agnostic Clusters for Open-Set Domain Adaptation

This paper addresses the problem of domain adaptation from closed-set to open-set situation by augmenting the state-of-the-art domain adaptation technique, Self-Ensembling, with category-agnostic clusters in target domain with SE-CC.

Open Set Domain Adaptation

This work learns a mapping from the source to the target domain by jointly solving an assignment problem that labels those target instances that potentially belong to the categories of interest present in the source dataset.

Multi-source Open-Set Deep Adversarial Domain Adaptation

A novel adversarial learningdriven approach to deal with multi-source openset unsupervised domain adaptation (MS-OSDA) which model a shared feature space for all the domains which explicitly mitigates the domaingap among the source-domains.

Open Set Domain Adaptation by Backpropagation

This paper proposes a method for an open set domain adaptation scenario, which utilizes adversarial training, and allows to extract features that separate unknown target from known target samples.

On the Effectiveness of Image Rotation for Open Set Domain Adaptation

This work proposes a novel method to addresses both problems of OSDA and reproducibility using the self-supervised task of rotation recognition using a new open set metric that properly balances the contribution of recognizing the known classes and rejecting the unknown samples.

Separate to Adapt: Open Set Domain Adaptation via Progressive Separation

The approach adopts a coarse-to-fine weighting mechanism to progressively separate the samples of unknown and known classes, and simultaneously weigh their importance on feature distribution alignment, which allows openness-robust open set domain adaptation, which can be adaptive to a variety of openness in the target domain.

Towards Inheritable Models for Open-Set Domain Adaptation

This work formalizes knowledge inheritability as a novel concept and proposes a simple yet effective solution to realize inheritable models suitable for the above practical DA paradigm.

Revisiting Batch Normalization For Practical Domain Adaptation

This paper proposes a simple yet powerful remedy, called Adaptive Batch Normalization (AdaBN) to increase the generalization ability of a DNN, and demonstrates that the method is complementary with other existing methods and may further improve model performance.

Your Classifier can Secretly Suffice Multi-Source Domain Adaptation

This work uses pseudo-labeled target samples and enforce a classifier agreement on the pseudo-labels, a process called Self-supervised Implicit Alignment (SImpAl), and finds that SImpAl readily works even under category-shift among the source domains.