mlsanityData to 14 Oct 2026

Joint Class-Time Learning for Video Classification with Multi-Instance Partial-Label Learning

arXiv 2610.06234 ↗First posted 5 Oct 2026cs.CV cs.AIcited by 0

Not on the map. Why?

Abstract (arXiv)

Multi-instance partial-label learning (MIPL) addresses inexact supervision in both the instance and label spaces, which can be applied to video classification. However, bag-level labels do not explicitly supervise the correspondence between candidate classes and temporal evidence. We propose ours, which couples label disambiguation with temporal evidence allocation through a joint class--time assignment. Occupancy-regularized spherical matching associates contextualized video features while learning nonuniform temporal mass and discouraging excessive concentration. During training, candidate-restricted inference recomputes the assignment within the candidate label set. A dual-marginal KL projection then constructs a structured teacher that incorporates momentum-refined class beliefs while preserving the proposal's temporal occupancy. A single plan-level KL objective aligns the full-space predictor with this teacher. Our analysis characterizes when candidate re-solving differs from masking and shows that, under the stated construction, the joint objective decomposes into class-marginal and class-conditional temporal supervision. We construct VCMIPL benchmarks from Breakfast, DoTA, and FineAction using model-generated candidate labels and evaluate the method across four feature representations. Extensive experimental results demonstrate that PIVOTMIPL outperforms existing MIPL algorithms in both effectiveness and efficiency.

Key facts. No citing papers in the map yet (data through 14 Oct 2026). Builds on 20 of its 31 references that are in the map. First posted on arXiv 5 Oct 2026.

Builds on

20 of 31 references.

References · 20

  1. 2015Deep Residual Learning for Image Recognitioncited by 52,096
  2. 2018SlowFast Networks for Video Recognitioncited by 1,867
  3. 2024Transformers are SSMs: Generalized Models and Efficient Algorithms Through Structured State Space Dualitycited by 1,336
  4. 2023VideoMAE V2: Scaling Video Masked Autoencoders with Dual Maskingcited by 528
  5. 2017UntrimmedNets for Weakly Supervised Action Recognition and Detectioncited by 130
  6. 2021FineAction: A Fine-Grained Video Dataset for Temporal Action Localizationcited by 61
  7. 2022ASM-Loc: Action-aware Segment Modeling for Weakly-Supervised Temporal Action Localizationcited by 40
  8. 2022SoLar: Sinkhorn Label Refinery for Imbalanced Partial-Label Learningcited by 31
  9. 2023Weakly-Supervised Temporal Action Localization by Inferring Salient Snippet-Featurecited by 22
  10. 2023Proposal-Based Multiple Instance Learning for Weakly-Supervised Temporal Action Localizationcited by 20
  11. 2024Candidate Pseudolabel Learning: Enhancing Vision-Language Models by Prompt Tuning with Unlabeled Datacited by 20
  12. 2025Realistic Evaluation of Deep Partial-Label Learning Algorithmscited by 12
8 more
  1. 2023Disambiguated Attention Embedding for Multi-Instance Partial-Label Learningcited by 10
  2. 2024Weakly Supervised Temporal Action Localization via Dual-Prior Collaborative Learning Guided by Multimodal Large Language Modelscited by 10
  3. 2022Multi-Instance Partial-Label Learning: Towards Exploiting Dual Inexact Supervisioncited by 9
  4. 2025Rethinking Pseudo-Label Guided Learning for Weakly Supervised Temporal Action Localization from the Perspective of Noise Correctioncited by 8
  5. 2025Bridge the Gap: From Weak to Full Supervision for Temporal Action Localization with PseudoFormercited by 6
  6. 2025Multi-Instance Partial-Label Learning with Margin Adjustmentcited by 5
  7. 2024Exploiting Conjugate Label Information for Multi-Instance Partial-Label Learningcited by 3
  8. 2025Calibratable Disambiguation Loss for Multi-Instance Partial-Label Learningcited by 1

Built on by

Not cited yet.