Pith. Sign in

RECORD

Deep Feature Factorization For Concept Discovery

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
Edo Collins, Radhakrishna Achanta, Sabine Süsstrunk

Machine Learning · Submitted 26 Jun 2018

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

The authors' main claim DFF can also be used to perform co-segmentation and co-localization, and we report state-of-the-art results on these tasks.

how it is stated · asserted 0.95 read the record →

Comment (0) Read on arXiv ↗
Cite ▾
arXiv ID1806.10206 (version 5)
Pith IDVSRVXCOI, Pith's permanent record of this paper and its review
CitationEdo Collins, Radhakrishna Achanta, Sabine S\"usstrunk. "Deep Feature Factorization For Concept Discovery." arXiv:1806.10206 (2018). https://pith.science/paper/VSRVXCOI
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, 3,609 input tokens.

What the abstract itself states, as probabilities: claims first 0.04 · states limitation 0.05 · human participants 0.04 · controlled comparison 0.09 · randomized assignment 0.01 · states quantitative result 0.07 · compares with prior results 0.63 · mentions code or data availability 0.03.

The abstract, sentence by sentence

arXiv listing abstract v5. Results and the main claim are underlined; .

We propose Deep Feature Factorization (DFF), a method capable of localizing similar semantic concepts within an image or a set of images. We use DFF to gain insight into a deep convolutional neural network's learned features, where we detect hierarchical cluster structures in feature space. This is visualized as heat maps, which highlight semantically matching regions across a set of images, revealing what the network `perceives' as similar. DFF can also be used to perform co-segmentation and co-localization, and we report state-of-the-art results on these tasks.

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 We propose Deep Feature Factorization (DFF), a method capable of localizing similar semantic concepts within an image or a set of images. contribution 0.96 0.87 0.25 asserted 0.96
s1 We use DFF to gain insight into a deep convolutional neural network's learned features, where we detect hierarchical cluster structures in feature space. contribution 0.57 0.93 0.13 asserted 1.00
s2 This is visualized as heat maps, which highlight semantically matching regions across a set of images, revealing what the network `perceives' as similar. interpretation 0.45 0.85 0.13 asserted 1.00
s3 DFF can also be used to perform co-segmentation and co-localization, and we report state-of-the-art results on these tasks. result 0.84 0.95 0.26 asserted 0.95

What kind of paper the classifier reports this is: method or tool 1.00 · evidence from observation or experiment 0.87.

What was examined, and what was not

sources read
  • arXiv listing abstract v5
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

We propose Deep Feature Factorization (DFF), a method capable of localizing similar semantic concepts within an image or a set of images. We use DFF to gain insight into a deep convolutional neural network's learned features, where we detect hierarchical cluster structures in feature space. This is visualized as heat maps, which highlight semantically matching regions across a set of images, revealing what the network `perceives' as similar. DFF can also be used to perform co-segmentation and co-localization, and we report state-of-the-art results on these tasks.

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

What it cites

Uses the method or tool of 1 paper
Cites as background 1 paper
  • There has been work extending NMF to multiple layers [ref] , implementing NMF using neural networks [ref] and using NMF approximations as input to a neural network [ref] .

    Neural Network Matrix Factorization · Non-negative Matrix Factorization · 0.96

Discussion

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