AdaptGCD: Multi-Expert Adapter Tuning for Generalized Category Discovery
Not on the map. Why?
Abstract (arXiv)
Different from the traditional semi-supervised learning paradigm that is constrained by the close-world assumption, Generalized Category Discovery (GCD) presumes that the unlabeled dataset contains new categories not appearing in the labeled set, and aims to not only classify old categories but also discover new categories in the unlabeled data. Existing studies on GCD typically devote to transferring the general knowledge from the self-supervised pretrained model to the target GCD task via some fine-tuning strategies, such as partial tuning and prompt learning. Nevertheless, these fine-tuning methods fail to make a sound balance between the generalization capacity of pretrained backbone and the adaptability to the GCD task. To fill this gap, in this paper, we propose a novel adapter-tuning-based method named AdaptGCD, which is the first work to introduce the adapter tuning into the GCD task and provides some key insights expected to enlighten future research. Furthermore, considering the discrepancy of supervision information between the old and new classes, a multi-expert adapter structure equipped with a route assignment constraint is elaborately devised, such that the data from old and new classes are separated into different expert groups. Extensive experiments are conducted on 7 widely-used datasets. The remarkable improvements in performance highlight the effectiveness of our proposals.
Key facts. Cited by 2 papers in the map (data through 7 Oct 2026). Builds on 75 of its 106 references that are in the map. First posted on arXiv 29 Oct 2024.
Builds on
75 of 106 references.
References · 75
- 2021Learning Transferable Visual Models From Natural Language Supervisioncited by 33,456
- 2021LoRA: Low-Rank Adaptation of Large Language Modelscited by 14,084
- 2020A Simple Framework for Contrastive Learning of Visual Representationscited by 11,979
- 2023DINOv2: Learning Robust Visual Features without Supervisioncited by 7,764
- 2021Masked Autoencoders Are Scalable Vision Learnerscited by 7,339
- 2021Emerging Properties in Self-Supervised Vision Transformerscited by 6,893
- 2021Prefix-Tuning: Optimizing Continuous Prompts for Generationcited by 3,878
- 2017Outrageously Large Neural Networks: The Sparsely-Gated Mixture-of-Experts Layercited by 3,413
- 2021Switch Transformers: Scaling to Trillion Parameter Models with Simple and Efficient Sparsitycited by 3,014
- 2020Supervised Contrastive Learningcited by 2,803
- 2020Improved Baselines with Momentum Contrastive Learningcited by 2,282
- 2020FixMatch: Simplifying Semi-Supervised Learning with Consistency and Confidencecited by 2,080
38 more
- 2013Fine-Grained Visual Classification of Aircraftcited by 1,701
- 2022Visual Prompt Tuningcited by 1,674
- 2019MixMatch: A Holistic Approach to Semi-Supervised Learningcited by 1,390
- 2019Contrastive Multiview Codingcited by 1,297
- 2016Temporal Ensembling for Semi-Supervised Learningcited by 1,054
- 2022Few-Shot Parameter-Efficient Fine-Tuning is Better and Cheaper than In-Context Learningcited by 886
- 2022AdaptFormer: Adapting Vision Transformers for Scalable Visual Recognitioncited by 714
- 2024Parameter-Efficient Fine-Tuning for Large Models: A Comprehensive Surveycited by 641
- 2017Learning multiple visual domains with residual adapterscited by 553
- 2021FlexMatch: Boosting Semi-Supervised Learning with Curriculum Pseudo Labelingcited by 537
- 2019Pseudo-Labeling and Confirmation Bias in Deep Semi-Supervised Learningcited by 507
- 2023LoraHub: Efficient Cross-Task Generalization via Dynamic LoRA Compositioncited by 359
- 2020K-Adapter: Infusing Knowledge into Pre-Trained Models with Adapterscited by 318
- 2021In Defense of Pseudo-Labeling: An Uncertainty-Aware Pseudo-label Selection Framework for Semi-Supervised Learningcited by 301
- 2022LST: Ladder Side-Tuning for Parameter and Memory Efficient Transfer Learningcited by 228
- 2022Generalized Category Discoverycited by 209
- 2021Open-World Semi-Supervised Learningcited by 165
- 2021A Unified Objective for Novel Class Discoverycited by 161
- 2023S-LoRA: Serving Thousands of Concurrent LoRA Adapterscited by 146
- 2021AutoNovel: Automatically Discovering and Learning Novel Visual Categoriescited by 116
- 2022Parametric Classification for Generalized Category Discovery: A Baseline Studycited by 113
- 2023When MOE Meets LLMs: Parameter Efficient Fine-tuning for Multi-task Medical Applicationscited by 110
- 2024Higher Layers Need More LoRA Expertscited by 101
- 2023Dynamic Conceptional Contrastive Learning for Generalized Category Discoverycited by 90
- 2022PromptCAL: Contrastive Affinity Learning via Auxiliary Prompts for Generalized Novel Category Discoverycited by 85
- 2023One-for-All: Generalized LoRA for Parameter-Efficient Fine-tuningcited by 84
- 2023LoRAMoE: Alleviate World Knowledge Forgetting in Large Language Models via MoE-Style Plugincited by 82
- 2022Parameter-efficient Model Adaptation for Vision Transformerscited by 81
- 2023Punica: Multi-Tenant LoRA Servingcited by 80
- 2021Cross-domain Few-shot Learning with Task-specific Adapterscited by 79
- 2023Towards Efficient Visual Adaption via Structural Re-parameterizationcited by 73
- 2023Learning Semi-supervised Gaussian Mixture Models for Generalized Category Discoverycited by 72
- 2023E^2VPT: An Effective and Efficient Approach for Visual Prompt Tuningcited by 70
- 2023Conditional Adapters: Parameter-efficient Transfer Learning with Fast Inferencecited by 69
- 2023VMT-Adapter: Parameter-Efficient Transfer Learning for Multi-Task Dense Scene Understandingcited by 62
- 2023No Representation Rules Them All in Category Discoverycited by 61
- 2019The Herbarium Challenge 2019 Datasetcited by 59
- 2021How Well Do Sparse Imagenet Models Transfer?cited by 50
50 most cited of 75 All 75 (JSON)
Built on by
Citing papers per year: 2026: 2.
2 papers.