# PACO: Proxy-Task Alignment and Online Calibration for On-the-Fly Category Discovery

arXiv: https://arxiv.org/abs/2604.11484 · HTML: https://mlsanity.com/paper/2604.11484/

## Key facts

- No citing papers in the map yet (data through 14 Oct 2026).
- Builds on 25 of its 47 references that are in the map.
- First posted on arXiv 13 Apr 2026.

## Builds on (most cited first, 25 of 25)

- [Attention Is All You Need](https://mlsanity.com/paper/1706.03762/) (2017)
- [Deep Residual Learning for Image Recognition](https://mlsanity.com/paper/1512.03385/) (2015)
- [An Image is Worth 16x16 Words: Transformers for Image Recognition at Scale](https://mlsanity.com/paper/2010.11929/) (2020)
- [ImageNet Large Scale Visual Recognition Challenge](https://mlsanity.com/paper/1409.0575/) (2014)
- [Emerging Properties in Self-Supervised Vision Transformers](https://mlsanity.com/paper/2104.14294/) (2021)
- [Densely Connected Convolutional Networks](https://mlsanity.com/paper/1608.06993/) (2016)
- [Learning to Generalize: Meta-Learning for Domain Generalization](https://mlsanity.com/paper/1710.03463/) (2017)
- [Towards Open Set Deep Networks](https://mlsanity.com/paper/1511.06233/) (2015)
- [Open-World Semi-Supervised Learning](https://mlsanity.com/paper/2102.03526/) (2021)
- [Learning Placeholders for Open-Set Recognition](https://mlsanity.com/paper/2103.15086/) (2021)
- [AutoNovel: Automatically Discovering and Learning Novel Visual Categories](https://mlsanity.com/paper/2106.15252/) (2021)
- [Parametric Classification for Generalized Category Discovery: A Baseline Study](https://mlsanity.com/paper/2211.11727/) (2022)
- [Novel Visual Category Discovery with Dual Ranking Statistics and Mutual Knowledge Distillation](https://mlsanity.com/paper/2107.03358/) (2021)
- [No Representation Rules Them All in Category Discovery](https://mlsanity.com/paper/2311.17055/) (2023)
- [Joint Representation Learning and Novel Category Discovery on Single- and Multi-modal Data](https://mlsanity.com/paper/2104.12673/) (2021)
- [SPTNet: An Efficient Alternative Framework for Generalized Category Discovery with Spatial Prompt Tuning](https://mlsanity.com/paper/2403.13684/) (2024)
- [Textual Knowledge Matters: Cross-Modality Co-Teaching for Generalized Visual Class Discovery](https://mlsanity.com/paper/2403.07369/) (2024)
- [GET: Unlocking the Multi-modal Potential of CLIP for Generalized Category Discovery](https://mlsanity.com/paper/2403.09974/) (2024)
- [HiLo: A Learning Framework for Generalized Category Discovery Robust to Domain Shifts](https://mlsanity.com/paper/2408.04591/) (2024)
- [Prototypical Hash Encoding for On-the-Fly Fine-Grained Category Discovery](https://mlsanity.com/paper/2410.19213/) (2024)
- [Novel Class Discovery for Long-tailed Recognition](https://mlsanity.com/paper/2308.02989/) (2023)
- [Boosting Novel Category Discovery Over Domains with Soft Contrastive Learning and All-in-One Classifier](https://mlsanity.com/paper/2211.11262/) (2022)
- [Generate, Refine, and Encode: Leveraging Synthesized Novel Samples for On-the-Fly Fine-Grained Category Discovery](https://mlsanity.com/paper/2507.04051/) (2025)
- [Labeled Data Selection for Category Discovery](https://mlsanity.com/paper/2406.04898/) (2024)
- [Learning for Transductive Threshold Calibration in Open-World Recognition](https://mlsanity.com/paper/2305.12039/) (2023)

## Abstract (arXiv)

On-the-Fly Category Discovery (OCD) requires a model, trained on an offline support set, to recognize known classes while discovering new ones from an online streaming sequence. Existing methods focus heavily on offline training. They aim to learn discriminative representations on the support set so that novel classes can be separated at test time. However, their discovery mechanism at inference is typically reduced to a single threshold. We argue that this paradigm is fundamentally flawed as OCD is not a static classification problem, but a dynamic process. The model must continuously decide 1) whether a sample belongs to a known class, 2) matches an existing novel category, or 3) should initiate a new one. Moreover, prior methods treat the support set as fixed knowledge. They do not update their decision boundaries as new evidence arrives during inference. This leads to unstable and inconsistent category formation. Our experiments confirm these issues. With properly calibrated and adaptive thresholds, substantial improvements can be achieved, even without changing the representation. Motivated by this, we propose PACO, a support-set-calibrated, tree-structured online decision framework. The framework models inference as a sequence of hierarchical decisions, including known-class routing, birth-aware novel assignment, and attach-versus-create operations over a dynamic prototype memory. Furthermore, we simulate the proxy discovery process to initialize the thresholds during offline training to align with inference. Thresholds are continuously updated during inference using mature novel prototypes. Importantly, PACO requires no heavy training and no dataset-specific tuning. It can be directly integrated into existing OCD pipelines as an inference-time module. Extensive experiments show significant improvements over SOTA baselines across seven benchmarks.