Blending-Target Domain Adaptation by Adversarial Meta-Adaptation Networks (2019-06-01T00:00:00.000000Z)

TL;DR

Empirical results show that BTDA is a quite challenging transfer setup for most existing DA algorithms, yet AMEAN significantly outperforms these state-of-the-art baselines and effectively restrains the negative transfer effects in BTDA.

Abstract

(Unsupervised) Domain Adaptation (DA) seeks for classifying target instances when solely provided with source labeled and target unlabeled examples for training. Learning domain-invariant features helps to achieve this goal, whereas it underpins unlabeled samples drawn from a single or multiple explicit target domains (Multi-target DA). In this paper, we consider a more realistic transfer scenario: our target domain is comprised of multiple sub-targets implicitly blended with each other so that learners could not identify which sub-target each unlabeled sample belongs to. This Blending-target Domain Adaptation (BTDA) scenario commonly appears in practice and threatens the validities of existing DA algorithms, due to the presence of domain gaps and categorical misalignments among these hidden sub-targets. To reap the transfer performance gains in this new scenario, we propose Adversarial Meta-Adaptation Network (AMEAN). AMEAN entails two adversarial transfer learning processes. The first is a conventional adversarial transfer to bridge our source and mixed target domains. To circumvent the intra-target category misalignment, the second process presents as ``learning to adapt'': It deploys an unsupervised meta-learner receiving target data and their ongoing feature-learning feedbacks, to discover target clusters as our ``meta-sub-target'' domains. This meta-sub-targets auto-design our meta-sub-target adaptation loss, which is capable to progressively eliminate the implicit category mismatching in our mixed target. We evaluate AMEAN and a variety of DA algorithms in three benchmarks under the BTDA setup. Empirical results show that BTDA is a quite challenging transfer setup for most existing DA algorithms, yet AMEAN significantly outperforms these state-of-the-art baselines and effectively restrains the negative transfer effects in BTDA.

References53 items

Unsupervised Multi-Target Domain Adaptation: An Information Theoretic Approach

AutoLoss: Learning Discrete Schedules for Alternate Optimization

Unbiasing Semantic Segmentation For Robot Perception using Synthetic Data Feature Transfer

Multi-target Unsupervised Domain Adaptation without Exactly Shared Categories

Boosting Domain Adaptation by Discovering Latent Domains

Domain Adaptive Faster R-CNN for Object Detection in the Wild

Deep Cocktail Network: Multi-source Unsupervised Domain Adaptation with Category Shift

A DIRT-T Approach to Unsupervised Domain Adaptation

Multiple Source Domain Adaptation with Adversarial Learning

Low-Shot Learning from Imaginary Data

Real-to-Virtual Domain Unification for End-to-End Autonomous Driving

Fine-Grained Recognition in the Wild: A Multi-task Domain Adaptation Approach

Improved Deep Embedded Clustering with Local Structure Preservation

Deep Hashing Network for Unsupervised Domain Adaptation

Virtual Adversarial Training: A Regularization Method for Supervised and Semi-Supervised Learning

Virtual to Real Reinforcement Learning for Autonomous Driving

Generate to Adapt: Aligning Domains Using Generative Adversarial Networks

Model-Agnostic Meta-Learning for Fast Adaptation of Deep Networks

Asymmetric Tri-training for Unsupervised Domain Adaptation

Adversarial Discriminative Domain Adaptation

Central Moment Discrepancy (CMD) for Domain-Invariant Representation Learning

Unsupervised Pixel-Level Domain Adaptation with Generative Adversarial Networks

FCNs in the Wild: Pixel-level Adversarial and Constraint-based Adaptation

Neural Architecture Search with Reinforcement Learning

Sim-to-Real Robot Learning from Pixels with Progressive Nets

Deep Reconstruction-Classification Networks for Unsupervised Domain Adaptation

Coupled Generative Adversarial Networks

CryptoNets: Applying Neural Networks to Encrypted Data with High Throughput and Accuracy

Learning to learn by gradient descent by gradient descent

Deep Transfer Learning with Joint Adaptation Networks

Unsupervised Domain Adaptation with Residual Transfer Networks

Deep Residual Learning for Image Recognition

Unsupervised Deep Embedding for Clustering Analysis

Simultaneous Deep Transfer Across Domains and Tasks

Domain-Adversarial Training of Neural Networks

Learning Transferable Features with Deep Adaptation Networks

Unsupervised Domain Adaptation by Backpropagation

ImageNet classification with deep convolutional neural networks

Domain adaptation for object recognition: An unsupervised approach

A Survey on Transfer Learning

Adapting Visual Category Models to New Domains

A theory of learning from different domains

Covariate Shift by Kernel Mean Matching

Domain Adaptation with Multiple Sources

Et al

Cross-domain video concept detection using adaptive svms

Semi-supervised Learning by Entropy Minimization

GENERATIVE ADVERSARIAL NETS

Using Simulation and Domain Adaptation to Improve Efficiency of Deep Robotic Grasping

Reading Digits in Natural Images with Unsupervised Feature Learning

GradientBased Learning Applied to Document Recognition

Gradient-based learning applied to document recognition

, a adversarial transfer framework incorporating meta-learner dynamically inducing meta-sub-targets to auto-design adversarial adaptation losses, which effectively achieve transfers in BTDA