RECORD
Sm: enhanced localization in Multiple Instance Learning for medical imaging classification
Computer Vision and Pattern Recognition · Submitted 4 Oct 2024
abstract compiled · 2026-09-23 · no language-model review
The authors' main claim A thorough empirical validation shows that our module leads to state-of-the-art performance in localization while being competitive or superior in classification.
The compiled record
The paper's own sentences, arranged. Each label is drawn from a closed vocabulary; the number beside it is the probability the classifier gave that label. Numbers in the tables are the authors', checked by code. No sentence in this record was written by a model.
Open the compiled record
The abstract, sentence by sentence
Multiple Instance Learning (MIL) is widely used in medical imaging classification to reduce the labeling effort. While only bag labels are available for training, one typically seeks predictions at both bag and instance levels (classification and localization tasks, respectively). Early MIL methods treated the instances in a bag independently. Recent methods account for global and local dependencies among instances. Although they have yielded excellent results in classification, their performance in terms of localization is comparatively limited. We argue that these models have been designed to target the classification task, while implications at the instance level have not been deeply investigated. Motivated by a simple observation -- that neighboring instances are likely to have the same label -- we propose a novel, principled, and flexible mechanism to model local dependencies. It can be used alone or combined with any mechanism to model global dependencies (e.g., transformers). A thorough empirical validation shows that our module leads to state-of-the-art performance in localization while being competitive or superior in classification. Our code is at https://github.com/Franblueee/SmMIL.
background contribution method result interpretation main claim
| # | sentence | role | the authors' own work | states the main conclusion | strength |
|---|---|---|---|---|---|
| s0 | Multiple Instance Learning (MIL) is widely used in medical imaging classification to reduce the labeling effort. | background 1.00 | 0.04 | 0.02 | asserted 1.00 |
| s1 | While only bag labels are available for training, one typically seeks predictions at both bag and instance levels (classification and localization tasks, respectively). | background 1.00 | 0.06 | 0.02 | asserted 0.74 |
| s2 | Early MIL methods treated the instances in a bag independently. | background 1.00 | 0.03 | 0.01 | asserted 1.00 |
| s3 | Recent methods account for global and local dependencies among instances. | background 1.00 | 0.04 | 0.02 | asserted 1.00 |
| s4 | Although they have yielded excellent results in classification, their performance in terms of localization is comparatively limited. | gap 0.84 | 0.08 | 0.07 | asserted 0.99 |
| s5 | We argue that these models have been designed to target the classification task, while implications at the instance level have not been deeply investigated. | gap 0.99 | 0.13 | 0.06 | asserted 0.89 |
| s6 | Motivated by a simple observation -- that neighboring instances are likely to have the same label -- we propose a novel, principled, and flexible mechanism to model local dependencies. | contribution 0.86 | 0.67 | 0.10 | asserted 0.91 |
| s7 | It can be used alone or combined with any mechanism to model global dependencies (e.g., transformers). | method 0.79 | 0.40 | 0.05 | asserted 0.87 |
| s8 | A thorough empirical validation shows that our module leads to state-of-the-art performance in localization while being competitive or superior in classification. | result 1.00 | 0.96 | 0.83 | asserted 1.00 |
| s9 | Our code is at https://github.com/Franblueee/SmMIL. | contribution 0.77 | 0.35 | 0.02 | asserted 0.83 |
What was examined, and what was not
- arXiv listing abstract v3
- full text
- figures and tables
- references and citing sentences
- relations to other papers
- P abstract
- S sections
- T body sentences
- C citations
- R cross-paper relations
The paper
Original abstract
Multiple Instance Learning (MIL) is widely used in medical imaging classification to reduce the labeling effort. While only bag labels are available for training, one typically seeks predictions at both bag and instance levels (classification and localization tasks, respectively). Early MIL methods treated the instances in a bag independently. Recent methods account for global and local dependencies among instances. Although they have yielded excellent results in classification, their performance in terms of localization is comparatively limited. We argue that these models have been designed to target the classification task, while implications at the instance level have not been deeply investigated. Motivated by a simple observation -- that neighboring instances are likely to have the same label -- we propose a novel, principled, and flexible mechanism to model local dependencies. It can be used alone or combined with any mechanism to model global dependencies (e.g., transformers). A thorough empirical validation shows that our module leads to state-of-the-art performance in localization while being competitive or superior in classification. Our code is at https://github.com/Franblueee/SmMIL.
Human review
How the literature uses it
What this paper does with the arXiv papers it cites
What it cites · 5 sentences without a confident label
Uses the method or tool of 1 paper
-
This constraint can be achieved using spectral normalization [ref] , and we study its influence in <ref>.
Spectral Normalization for Generative Adversarial Networks · Modelling the smoothness · 0.93
Cites as support 1 paper
-
As studied by [ref] , it is enough to use a small number of iterations $T$ to approximate the exact solution.
Predict then Propagate: Graph Neural Networks meet Personalized PageRank · The smooth operator · 0.96
Cites as background 2 papers
-
Based on it, other transformer-based approaches have emerged, also focusing on global dependencies [ref] .
Diagnose Like a Pathologist: Transformer-Enabled Hierarchical Attention-Guided Multiple Instance Learning for Whole Slide Image Classification · Introduction · 0.83
-
Moreover, the Dirichlet energy has been studied in the context of GNNs to analyze the over-smoothing phenomenon [ref] .
A Note on Over-Smoothing for Graph Neural Networks · Related work · 0.90
Discussion
Know something about this paper? Leave a comment, or highlight any passage above to comment on it.