Disentangling Transfer and Interference in Multi-Domain Learning
Not on the map. Why?
Abstract (arXiv)
Humans are incredibly good at transferring knowledge from one domain to another, enabling rapid learning of new tasks. Likewise, transfer learning has enabled enormous success in many computer vision problems using pretraining. However, the benefits of transfer in multi-domain learning, where a network learns multiple tasks defined by different datasets, has not been adequately studied. Learning multiple domains could be beneficial, or these domains could interfere with each other given limited network capacity. Understanding how deep neural networks of varied capacity facilitate transfer across inputs from different distributions is a critical step towards open world learning. In this work, we decipher the conditions where interference and knowledge transfer occur in multi-domain learning. We propose new metrics disentangling interference and transfer, set up experimental protocols, and examine the roles of network capacity, task grouping, and dynamic loss weighting in reducing interference and facilitating transfer.
Key facts. Cited by 3 papers in the map (data through 7 Oct 2026). Builds on 54 of its 68 references that are in the map. First posted on arXiv 2 Jul 2021.
Builds on
54 of 68 references.
References · 54
- 2015Deep Residual Learning for Image Recognitioncited by 51,809
- 2015Batch Normalization: Accelerating Deep Network Training by Reducing Internal Covariate Shiftcited by 6,979
- 2017Gradient Episodic Memory for Continual Learningcited by 1,871
- 2019Similarity of Neural Network Representations Revisitedcited by 1,695
- 2018Group Normalizationcited by 1,514
- 2017Multi-Task Learning Using Uncertainty to Weigh Losses for Scene Geometry and Semanticscited by 1,345
- 2018Continual Lifelong Learning with Neural Networks: A Reviewcited by 1,313
- 2019A Comprehensive Survey on Transfer Learningcited by 1,154
- 2017GradNorm: Gradient Normalization for Adaptive Loss Balancing in Deep Multitask Networkscited by 816
- 2018Multi-Task Learning as Multi-Objective Optimizationcited by 809
- 2018Riemannian Walk for Incremental Learning: Understanding Forgetting and Intransigencecited by 734
- 2018Deep Visual Domain Adaptation: A Surveycited by 634
38 more
- 2018Taskonomy: Disentangling Task Transfer Learningcited by 631
- 2019Three scenarios for continual learningcited by 610
- 2017Learning multiple visual domains with residual adapterscited by 553
- 2018End-to-End Multi-Task Learning with Attentioncited by 533
- 2014Predicting Depth, Surface Normals and Semantic Labels with a Common Multi-Scale Convolutional Architecturecited by 483
- 2016Cross-stitch Networks for Multi-task Learningcited by 466
- 2019Which Tasks Should Be Learned Together in Multi-task Learning?cited by 367
- 2021Towards Open World Object Detectioncited by 297
- 2018Context Encoding for Semantic Segmentationcited by 289
- 2021Scaling Laws for Transfercited by 265
- 2014Towards Open World Recognitioncited by 226
- 2013OverFeat: Integrated Recognition, Localization and Detection using Convolutional Networkscited by 224
- 2016UberNet: Training a `Universal' Convolutional Neural Network for Low-, Mid-, and High-Level Vision using Diverse Datasets and Limited Memorycited by 200
- 2017Multi-task Self-Supervised Visual Learningcited by 200
- 2018Efficient parametrization of multi-domain deep neural networkscited by 168
- 2020Linear Mode Connectivity in Multitask and Continual Learningcited by 144
- 2019Mish: A Self Regularized Non-Monotonic Activation Functioncited by 141
- 2015Instance-aware Semantic Segmentation via Multi-task Network Cascadescited by 120
- 2018Don't forget, there is more than forgetting: new metrics for Continual Learningcited by 113
- 2020Understanding and Improving Information Transfer in Multi-Task Learningcited by 113
- 2017Incremental Learning Through Deep Adaptationcited by 108
- 2020A Comprehensive Study of Class Incremental Learning Algorithms for Visual Taskscited by 107
- 2018Auxiliary Tasks in Multi-task Learningcited by 106
- 2016Fully-adaptive Feature Sharing in Multi-Task Networks with Applications in Person Attribute Classificationcited by 93
- 2020Deep Visual Domain Adaptationcited by 91
- 2016A Deep Hierarchical Approach to Lifelong Learning in Minecraftcited by 90
- 2019Representation Similarity Analysis for Efficient Task taxonomy & Transfer Learningcited by 86
- 2020A Wholistic View of Continual Learning with Deep Neural Networks: Forgotten Lessons and the Bridge to Active and Open World Learningcited by 86
- 2018Real-Time Joint Semantic Segmentation and Depth Estimation Using Asymmetric Annotationscited by 51
- 2017Universal representations:The missing link between faces, text, planktons, and cat breedscited by 45
- 2018Open-world Learning and Application to Product Classificationcited by 39
- 2019Depthwise Convolution is All You Need for Learning Multiple Visual Domainscited by 36
- 2020Multi-Loss Weighting with Coefficient of Variationscited by 36
- 2016MOON: A Mixed Objective Optimization Network for the Recognition of Facial Attributescited by 34
- 2017BlitzNet: A Real-Time Deep Network for Scene Understandingcited by 29
- 2016Integrated perception with recurrent multi-task neural networkscited by 24
- 2018Adding New Tasks to a Single Network with Weight Transformations using Binary Maskscited by 21
- 2019Budget-Aware Adapters for Multi-Domain Learningcited by 18
50 most cited of 54 All 54 (JSON)
Built on by
Citing papers per year: 2022: 1; 2023: 1; 2024: 1.
3 papers.