Pith. Sign in

RECORD

Hard Negative Sample Mining for Whole Slide Image Classification

PaperPDF from arXiv

data files

Paper breakdownThe authors' claims, conditions and measurements, as JSON Signed receiptProof this review has not been altered
X Bluesky LinkedIn Reddit Hacker News
Wentao Huang, Xiaoling Hu, Shahira Abousamra, Prateek Prasanna, Chao Chen

Computer Vision and Pattern Recognition · Submitted 3 Oct 2024

abstract compiled · 2026-09-23 · no language-model review

The authors' main claim Experiments on two public datasets demonstrate the efficacy of these proposed ideas.

how it is stated · asserted 1.00 read the record →

Comment (0) Read on arXiv ↗
Cite ▾
arXiv ID2410.02212 (version 1)
Pith IDMISPAKSM, Pith's permanent record of this paper and its review
CitationChao Chen, Prateek Prasanna, Shahira Abousamra, Wentao Huang, Xiaoling Hu. "Hard Negative Sample Mining for Whole Slide Image Classification." arXiv:2410.02212 (2024). https://pith.science/paper/MISPAKSM
Claim paper

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

Compiled by classifier-1.13.0 via TypeSafe API, question bank abstract-p1-20260923, 2026-09-23: 1 request, 6,012 input tokens.

What the abstract itself states, as probabilities: claims first 0.08 · states limitation 0.07 · human participants 0.18 · controlled comparison 0.13 · randomized assignment 0.01 · states quantitative result 0.04 · compares with prior results 0.24 · mentions code or data availability 0.98.

The abstract, sentence by sentence

arXiv listing abstract v1. Results and the main claim are underlined; .

Weakly supervised whole slide image (WSI) classification is challenging due to the lack of patch-level labels and high computational costs. State-of-the-art methods use self-supervised patch-wise feature representations for multiple instance learning (MIL). Recently, methods have been proposed to fine-tune the feature representation on the downstream task using pseudo labeling, but mostly focusing on selecting high-quality positive patches. In this paper, we propose to mine hard negative samples during fine-tuning. This allows us to obtain better feature representations and reduce the training cost. Furthermore, we propose a novel patch-wise ranking loss in MIL to better exploit these hard negative samples. Experiments on two public datasets demonstrate the efficacy of these proposed ideas. Our codes are available at https://github.com/winston52/HNM-WSI

background contribution method result interpretation main claim

What the classifier reports for each sentence. The number is the probability it gave the label.
#sentencerolethe authors' own workstates the main conclusionstrength
s0 Weakly supervised whole slide image (WSI) classification is challenging due to the lack of patch-level labels and high computational costs. background 1.00 0.05 0.02 asserted 1.00
s1 State-of-the-art methods use self-supervised patch-wise feature representations for multiple instance learning (MIL). background 1.00 0.04 0.02 asserted 1.00
s2 Recently, methods have been proposed to fine-tune the feature representation on the downstream task using pseudo labeling, but mostly focusing on selecting high-quality positive patches. gap 0.63 0.04 0.03 asserted 0.98
s3 In this paper, we propose to mine hard negative samples during fine-tuning. contribution 0.95 0.85 0.14 asserted 0.96
s4 This allows us to obtain better feature representations and reduce the training cost. result 0.37 0.71 0.30 asserted 0.93
s5 Furthermore, we propose a novel patch-wise ranking loss in MIL to better exploit these hard negative samples. contribution 0.70 0.89 0.08 asserted 0.96
s6 Experiments on two public datasets demonstrate the efficacy of these proposed ideas. result 0.97 0.91 0.47 asserted 1.00
s7 Our codes are available at https://github.com/winston52/HNM-WSI contribution 0.61 0.47 0.02 asserted 0.95

What kind of paper the classifier reports this is: method or tool 1.00 · evidence from observation or experiment 0.94.

What was examined, and what was not

sources read
  • arXiv listing abstract v1
not examined
  • full text
  • figures and tables
  • references and citing sentences
  • relations to other papers
levels run
  • P abstract
not run
  • S sections
  • T body sentences
  • C citations
  • R cross-paper relations

Every quoted sentence on this record is the authors'. Every label is from Pith's closed vocabulary, with the probability shown beside it. Every number in the measurements table is the authors', checked by code. Record as JSON

The paper
Original abstract

Weakly supervised whole slide image (WSI) classification is challenging due to the lack of patch-level labels and high computational costs. State-of-the-art methods use self-supervised patch-wise feature representations for multiple instance learning (MIL). Recently, methods have been proposed to fine-tune the feature representation on the downstream task using pseudo labeling, but mostly focusing on selecting high-quality positive patches. In this paper, we propose to mine hard negative samples during fine-tuning. This allows us to obtain better feature representations and reduce the training cost. Furthermore, we propose a novel patch-wise ranking loss in MIL to better exploit these hard negative samples. Experiments on two public datasets demonstrate the efficacy of these proposed ideas. Our codes are available at https://github.com/winston52/HNM-WSI

Human review

How Pith checks papers → · Full audit of this paper →

How the literature uses it

What this paper does with the arXiv papers it cites

Each citing sentence is taken from the citing paper's LaTeX source and typed by Pith's classifier. A label appears at probability 0.8 or more (0.95 for support), a contest at 0.7 or more; the quoted words are the citing authors'. How each label was checked · JSON

Discussion

Know something about this paper? Leave a comment, or highlight any passage above to comment on it.