Pith. sign in

REVIEW 3 major objections 5 minor 39 references

To Beta or Not To Beta: Information Bottleneck for DigitaL Image Forensics

T0 review · 3 major / 5 minor · reviewed 2026-08-14 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗ Review JSON

Pith's one-line read An information-bottleneck CNN trained on camera identity can localize image splices by exposing inconsistent camera fingerprints, outperforming existing methods on standard benchmarks.

desk verdict A credible but incremental application of variational IB to splice localization, undercut by beta tuning on the DSO-1 test set. read the letter →

arxiv 1908.03864 v1 pith:NTSHQ4US submitted 2019-08-11 cs.CV

Aurobrata Ghosh , Zheng Zhong , Steve Cruz , Subbu Veeravasarapu , Terrance E Boult , Maneesh Singh This is my paper · ORCID
classification cs.CV
keywords imageforensicssplicelocalizationinformationbottleneckvariationalinferencecamera-modelidentificationnoiseresidualGANinpaintingdetectiondeeplearning
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

This paper proposes InfoPrint, a deep network that treats image forensics as a representation learning problem: train a CNN to identify the camera model of small image patches while forcing the internal representation through an information bottleneck, so that the network keeps only low-level statistical fingerprints and discards semantic content. Because spliced or inpainted regions come from a different image-formation process, their fingerprints differ from the host image, and a two-component Gaussian mixture over patch representations can localize the tampering. The paper reports that InfoPrint outperforms SpliceBuster and EX-SC on three standard splice-localization datasets and also flags regions altered by three inpainting GANs. It further claims that the variational IB solution is accurate and far cheaper to train than an earlier numerical mutual-information method, making long and deep training practical. If this is right, it shows that deliberately suppressing semantics in favor of device fingerprints is a viable route to robust localization of image forgeries.

What carries the argument

The carrying mechanism is the variational information bottleneck, applied here in reverse of its usual use. Instead of extracting high-level semantics, the objective $L = I(Z,Y) - \beta I(Z,X)$ is optimized with variational bounds so that $Z$ retains only the information in a patch that is predictive of the camera model; since object content is not predictive of camera identity, that information is discarded. The encoder begins with a constrained convolution layer whose filter weights sum to zero, $R(k)=w_k(0,0)+\sum_{i,j\ne(0,0)}w_k(i,j)=0$, which acts as a learned high-pass noise-residual extractor. A succession of residual blocks ends in a stochastic layer, and the decoder is deliberately a simple logistic-regression softmax. The authors select $\beta$ by inspecting the rate-distortion plane, then segment test-image patch encodings with a two-component EM Gaussian mixture.

What would settle it

Compile a test set of images where each splice is taken from another photo shot with the same camera model and the same processing pipeline, then run InfoPrint; if its localization F1 on same-model splices is no better than chance, the paper's core assumption that inserted regions carry a different low-level fingerprint is falsified for that scenario.

Watch

Extended reading notes

Core claim

InfoPrint is a 27-layer CNN whose encoder maps each $49\times49\times3$ patch to a stochastic code $Z\sim\mathcal{N}(\mu_x,\mathrm{diag}(\sigma_x))$, trained on camera-model identification from the Dresden Image Database with the variational IB loss $J_{\mathrm{IB}} = \frac{1}{N}\sum_{i=1}^N \mathbb{E}_{z\sim p(z|x_i)}[-\log q(y_i|z)] + \beta\,\mathrm{KL}[p(z|x_i)\|r(z)]$, where $r(z)$ is a standard Gaussian. The central claim is that this bottleneck suppresses semantic content and preserves each camera model's low-level noise-residual fingerprint. At test time the network computes fingerprint encodings for overlapping patches, and a two-component Gaussian mixture separates host from inserted regions. On DSO-1, NC16, and NC17-dev1, InfoPrint reaches F1 scores of 0.72, 0.42, and 0.44 and AUC scores of 0.92, 0.83, and 0.82, above SpliceBuster and EX-SC; it also localizes regions inpainted by three GANs.

Load-bearing premise

The paper assumes that any inserted region has a different low-level statistical fingerprint from the host — a splice from a different camera or an inpainting algorithm — and that this difference survives the image's processing history; if an attacker inserts same-camera content or re-compresses to homogenize noise, the inconsistency vanishes and localization fails.

Editorial extensions

If this is right

  • If InfoPrint's representation is a true camera fingerprint, the same network should localize splices from cameras never seen in training, because training on 27 Dresden camera models teaches separation of low-level noise patterns.
  • The variational IB solution is fast enough (about 14 hours on one GPU) to make long training and deeper architectures practical, while still matching or beating the numerically expensive mutual-information model.
  • Because the method marks inconsistency rather than semantic content, it can expose GAN-inpainted regions even in images that have been resized or re-compressed, as long as the hallucinated pixels leave a distinct low-level signature.
  • The model outputs a probability mask but not a binary verdict: the two-class GMM always finds two classes, so forgery detection must be a separate step.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

  • An immediate extension is to feed the patch-level fingerprint vectors into a binary forged/pristine classifier, which would let the same representation answer the detection question the paper explicitly leaves open.
  • A stress test worth running is same-camera splicing: if the assumption holds only for different-camera input, InfoPrint should be benchmarked on a same-camera spliced dataset to map the boundary of its validity.
  • The rate-distortion curve suggests a testable prediction: at lower rate (larger $\beta$), the representation should become more robust to post-processing such as re-compression, because it has been forced to discard content-dependent details; this could be measured directly.
  • Because the bottleneck is agnostic to image type, the same scheme could be tried on video frames or on non-visible-light sensor outputs, where device fingerprints have different but predictable statistical structure.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Referee Report

3 major / 5 minor

Summary. The paper introduces InfoPrint, a 27-layer CNN trained for camera-model identification with a variational information bottleneck (IB) objective, and applies it to splice localization by segmenting IB-based fingerprint representations with a two-class GMM. The network is trained on the Dresden Image Database, and the authors report F1, MCC, and AUC on DSO-1, NC16, and NC17-dev1, comparing against their own non-IB baseline (NoMI), their numerical mutual-information baseline (MI), and two published methods (SpliceBuster, EX-SC). The paper also shows qualitative results for detecting inpainting by three GANs and discusses failure cases in small and saturated images.

Significance. If the results hold, the paper makes an interesting and useful contribution by casting camera-model fingerprint modeling as a variational IB representation-learning problem and showing that a variational solution can be cheaper than a numerical MI alternative while improving localization. The experimental reporting is reasonably thorough: several metrics are used, Otsu thresholding is reported alongside optimal thresholds, and the paper discusses explicit failure cases. However, the central claim of outperforming the state of the art depends heavily on the DSO-1 column, where the regularization parameter was selected directly on the DSO-1 test set; this introduces selection bias. In addition, the closely related Noiseprint method is cited but not compared, and the GAN-inpainting contribution is only qualitative. These issues need to be addressed before the performance claims can be accepted.

major comments (3)
  1. [Section 5, beta-selection paragraph and Tables 2-3] The regularization parameter beta is selected on the DSO-1 evaluation set itself: the text states, "To select beta for the forensic task, we compute F1 scores on DSO-1 for all values of beta till 0 and find a peak," and then Tables 2-3 report DSO-1 results for the two chosen values beta=1e-3 and 5e-4. Because the same set is used both to choose beta and to produce the reported DSO-1 metrics, the DSO-1 F1/MCC/AUC numbers are not unbiased estimates of InfoPrint's performance, and the headline improvement over SpliceBuster and EX-SC (F1 0.72 vs 0.66 and 0.57) may be inflated by selection bias. This is load-bearing, since the DSO-1 column shows the largest margins; on NC17-dev1 EX-SC ties or beats one of the InfoPrint variants. The authors should select beta on a held-out validation set (for example, a split of DSO-1 or the Dresden validation set) and then report DSO-1 as an untouched test set, ideally with variability over beta choices.
  2. [Sections 2 and 5, baseline selection] Noiseprint [14] is described in Related Work as a closely related "novel approach" that uses a denoising CNN to estimate noise-residual properties for forgery discovery, yet it is not included in the quantitative comparisons in Tables 2 and 3. Given that Noiseprint is arguably the most similar published method to InfoPrint, omitting it weakens the claim that InfoPrint "outperforms the state-of-the-art." The authors should add a Noiseprint comparison, or explicitly justify its exclusion with quantitative or architectural reasons.
  3. [Section 5, GAN inpainting experiments and Figure 3] The claimed ability to detect alterations made by three inpainting GANs is supported only by qualitative examples. The paper acknowledges that no standard dataset exists, but since this is stated as a contribution in the abstract and introduction, the claim is not substantiated without a quantitative evaluation. A controlled experiment with synthesized inpainted images and ground-truth masks, reported with at least one metric (e.g., F1 or AUC), would be needed to support the claim; otherwise the claim should be explicitly downgraded to a qualitative demonstration.
minor comments (5)
  1. [Abstract and Section 5] The abstract says "up to 5% points" improvement, while Section 5 reports "up to 6% points" over SpliceBuster and "15% points" over EX-SC; these numbers should be reconciled.
  2. [Tables 2-3] The tables rely on black and blue text colors to distinguish threshold choices; these colors may not survive all print or accessibility settings. Please use a separate textual marker or symbol in addition to color.
  3. [Section 3, reference [5]] The MI baseline [5] is cited as "Anonymous" and is not publicly available. Since the comparison against this baseline is a central part of the ablation, the authors should provide implementation details, code, or a public version of that work to make the comparison reproducible.
  4. [Section 5, beta selection] The phrase "for all values of beta till 0" is vague; please specify the grid of beta values tested and the number of runs per value, especially since beta=1e-3 is called an anomaly attributed to stochastic training.
  5. [Figure 4] The caption mentions "Log-probability maps of proposed methods," but the figure itself is not described in the text; please clarify what is plotted and how the log-probability maps were derived.

Circularity Check

1 steps flagged · score 4.0 of 10

DSO-1 comparison is a fitted prediction: beta is selected by optimizing F1 on DSO-1, and the same DSO-1 F1 is then reported as evidence of superiority.

  1. fitted input called prediction [Section 5, Experiments & Results (beta selection and Tables 2-3)]
    "To selectβ for the forensic task, we compute F1 scores on DSO-1 for all values ofβ till 0 and find a peak from 2e-3 to 1e-4 (1e-3 is an anomaly we attribute to stochastic training). Hence we conduct our experiments for two central values,β = 1e− 3, 5e− 4."

    The paper selects the regularization hyperparameter β by maximizing F1 on DSO-1, then reports the DSO-1 F1/MCC columns of Tables 2-3 as the headline result ('F1 scores indicate up to 6% points improvement over SB ... on DSO-1'). The reported DSO-1 performance is therefore a model-selection statistic, not an independent test-set prediction: the chosen model is literally the one that scored best on that dataset, and the optimal-threshold F1/MCC is computed from the same ground-truth masks. The non-tuned datasets (NC16 and NC17-dev1) show much smaller or mixed advantages, so the paper's largest claimed margin is substantially an artifact of fitting beta to the test set rather than an unbiased estimate of InfoPrint's performance.

full rationale

The core derivation—variational IB objective, encoder-decoder training on Dresden, constrained-convolution noise residuals, and GMM segmentation of feature inconsistencies—is self-contained and not definitionally circular. Equation 6 is a standard variational IB loss, and the localization pipeline has independent content. The only load-bearing step that reduces to its own input is the evaluation protocol for DSO-1: Section 5 fits beta to DSO-1 F1 and then Tables 2-3 report that same dataset's F1/MCC as the headline result. This is a fitted-parameter-called-prediction pattern, making the DSO-1 margin optimistically biased. However, the NC16/NC17 results, the AUC trends, and the qualitative GAN-inpainting demonstrations do not involve beta tuning and provide independent evidence, so the central claim is only partially circular rather than wholly reducible to the fit. The self-referential MI baseline [5] is transparently described and is not load-bearing for the method's derivation.

Assumptions & free parameters 3 free parameters · 4 assumptions · 0 invented entities

The central method rests on the standard variational IB objective and the assumption that tampering changes low-level fingerprints. The only fitted constants are beta, lambda, omega1, omega2, of which beta is tuned directly on a benchmark test set.

free parameters (3)
  • beta (IB regularization weight) = 1e-3, 5e-4 (also tested 1e-4 to 2e-3)
    Selected using RD curve and F1 scores on DSO-1 test set, then used for all results; tuning on the evaluation set inflates DSO-1 scores.
  • lambda (constrained-conv penalty weight) = 1
    Chosen empirically (Section 4C) to enforce the high-pass filter constraint.
  • omega1, omega2 (L1/L2 weight decay) = 1e-4 each
    Chosen empirically (Section 4C); sensitivity not analyzed.
assumptions (4)
  • domain assumption Inserted (spliced or inpainted) regions have different low-level statistical fingerprints than the host image.
    Section 4 states 'we assume that we can determine such forgeries by inspecting low-level pixel statistics'; the whole two-class segmentation depends on it.
  • domain assumption The untampered region is the largest part of the test image.
    Section 4B: 'Assuming that the untampered region is the largest part of the image like in [14,13]', used for the two-component GMM prior.
  • domain assumption Semantic content has no obvious relation to camera-model identity.
    Section 4A uses this to justify training on camera-model classification so IB suppresses semantics.
  • standard math Variational IB bound and Markov chain Y -> X -> Z are valid.
    Section 3 follows Alemi et al. [4]; assumes the standard IB factorization and variational approximations for intractable marginals.

how reviews work

0 comments
Cite this review

Pith. "Pith review of To Beta or Not To Beta: Information Bottleneck for DigitaL Image Forensics." pith.science (2026). https://pith.science/paper/NTSHQ4US

@misc{pith2026190803864,
  author       = {Pith},
  title        = {Pith review of: To Beta or Not To Beta: Information Bottleneck for DigitaL Image Forensics},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/NTSHQ4US}},
  note         = {Machine review of arXiv:1908.03864}
}
read the original abstract

We consider an information theoretic approach to address the problem of identifying fake digital images. We propose an innovative method to formulate the issue of localizing manipulated regions in an image as a deep representation learning problem using the Information Bottleneck (IB), which has recently gained popularity as a framework for interpreting deep neural networks. Tampered images pose a serious predicament since digitized media is a ubiquitous part of our lives. These are facilitated by the easy availability of image editing software and aggravated by recent advances in deep generative models such as GANs. We propose InfoPrint, a computationally efficient solution to the IB formulation using approximate variational inference and compare it to a numerical solution that is computationally expensive. Testing on a number of standard datasets, we demonstrate that InfoPrint outperforms the state-of-the-art and the numerical solution. Additionally, it also has the ability to detect alterations made by inpainting GANs.

Figures

Figures reproduced from arXiv: 1908.03864 by the authors.

Figure 1
Figure 1. β selection from RD curve. Left: RD curve in red, right: F1 metric on DSO-1. Low distortion values are attained for β ≤ 5e − 3, while a peak in the F1 is observed from 2e-3 to 1e-4 [PITH_FULL_IMAGE:figures/full_fig_p007_1.png] view at source ↗
Figure 2
Figure 2. Qualitative results showing the superiority of InfoPrint over published methods. [PITH_FULL_IMAGE:figures/full_fig_p007_2.png] view at source ↗
Figure 3
Figure 3. InfoPrint vs inpainting GANs: Liu et al. [26], Yu et al. [37], Nazeri et al. [30]. [PITH_FULL_IMAGE:figures/full_fig_p008_3.png] view at source ↗
Figures from the paper (1 more)
Figure 4
Figure 4. Figure 4: Failure examples. All methods tend to fail when the input image is small: e.g. [PITH_FULL_IMAGE:figures/full_fig_p008_4.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

39 extracted references · 34 canonical work pages

  1. [5]

    SpliceRadar: A learned method for blind image forensics

    Anonymous. SpliceRadar: A learned method for blind image forensics. In Supplied as supplementary material , 2019

    work page 2019

  2. [14]

    Noiseprint: a CNN-based camera model fingerprint

    Davide Cozzolino and Luisa Verdoliva. Noiseprint: a CNN-based camera model fingerprint. arXiv, 2018

    work page 2018

  3. [1]

    Achille and S

    A. Achille and S. Soatto. Information dropout: Learning optimal representations through noisy computation. IEEE Transactions on Pattern Analysis and Machine Intelligence , PP(99), 2018

    work page 2018

  4. [2]

    Agarwal and H

    S. Agarwal and H. Farid. Photo forensics from JPEG dimples. In 2017 IEEE Workshop on Information F orensics and Security (WIFS), pages 1–6, 12 2017

    work page 2017

  5. [3]

    Saurous, and Kevin Murphy

    Alexander Alemi, Ben Poole, Ian Fischer, Joshua Dillon, Rif A. Saurous, and Kevin Murphy. Fixing a broken ELBO. In Proceedings of the 35th International Conference on Machine Learning (ICML) , Jul 2018

    work page 2018

  6. [4]

    Alemi, Ian Fischer, Joshua V

    Alexander A. Alemi, Ian Fischer, Joshua V . Dillon, and Kevin Murphy. Deep variational information bottleneck. In International Conference on Learning Representations (ICLR) , Apr 2017

    work page 2017

  7. [6]

    David Barber and Felix V . Agakov. The IM algorithm: A variational approach to information maximization. In NeurIPS, 2003

    work page 2003

  8. [7]

    Higher-order, adversary-aware, double JPEG-detection via selected training on attacked samples

    Mauro Barni, Ehsan Nowroozi, and Benedetta Tondi. Higher-order, adversary-aware, double JPEG-detection via selected training on attacked samples. In 25th European Signal Processing Conference (EUSIPCO) , pages 281 – 285, 08 2017. 8 To Beta or Not To Beta: Information Bottleneck for Digital Image Forensics A PREPRINT

    work page 2017

Show all 39 references
  1. [8]

    Mutual information neural estimation

    Mohamed Ishmael Belghazi, Aristide Baratin, Sai Rajeshwar, Sherjil Ozair, Yoshua Bengio, Aaron Courville, and Devon Hjelm. Mutual information neural estimation. In Proceedings of the 35th International Conference on Machine Learning, (ICML), pages 531–540, 10–15 Jul 2018

    2018

  2. [9]

    Tampering detection and localization through clustering of camera-based CNN features

    Luca Bondi, Silvia Lameri, David Güera, Paolo Bestagini, Edward Delp, and Stefano Tubaro. Tampering detection and localization through clustering of camera-based CNN features. In The IEEE Conference on Computer Vision and Pattern Recognition (CVPR), pages 1855–1864, 07 2017

    2017

  3. [10]

    Burgess, Irina Higgins, Arka Pal, Loic Matthey, Nick Watters, Guillaume Desjardins, and Alexander Lerchner

    Christopher P. Burgess, Irina Higgins, Arka Pal, Loic Matthey, Nick Watters, Guillaume Desjardins, and Alexander Lerchner. Understanding disentangling in β-V AE.arXiv e-prints, page arXiv:1804.03599, Apr 2018

    2018 arXiv

  4. [11]

    Determining image origin and integrity using sensor noise

    Mo Chen, Jessica Fridrich, Miroslav Goljan, and Jan Lukás. Determining image origin and integrity using sensor noise. Information F orensics and Security, IEEE Transactions on, 3:74 – 90, 04 2008

    2008

  5. [12]

    InfoGAN: Interpretable representation learning by information maximizing generative adversarial nets

    Xi Chen, Yan Duan, Rein Houthooft, John Schulman, Ilya Sutskever, and Pieter Abbeel. InfoGAN: Interpretable representation learning by information maximizing generative adversarial nets. In NeurIPS, 2016

    2016

  6. [13]

    Cozzolino, G

    D. Cozzolino, G. Poggi, and L. Verdoliva. Splicebuster: A new blind image splicing detector. In 2015 IEEE International Workshop on Information F orensics and Security (WIFS), pages 1–6, 11 2015

    2015

  7. [15]

    T. J. d. Carvalho, C. Riess, E. Angelopoulou, H. Pedrini, and A. d. R. Rocha. Exposing digital image forgeries by illumination color classification. IEEE Transactions on Information F orensics and Security, 8(7):1182–1194, 07 2013

    2013

  8. [16]

    Learning deep representations by mutual information estimation and maximization.arXiv e-prints, page arXiv:1808.06670, Aug 2018

    R Devon Hjelm, Alex Fedorov, Samuel Lavoie-Marchildon, Karan Grewal, Phil Bachman, Adam Trischler, and Yoshua Bengio. Learning deep representations by mutual information estimation and maximization.arXiv e-prints, page arXiv:1808.06670, Aug 2018

    2018 arXiv

  9. [17]

    Image tamper detection based on demosaicing artifacts

    Ahmet Emir Dirik and Nasir Memon. Image tamper detection based on demosaicing artifacts. In in Proceedings of the 2009 IEEE International Conference on Image Processing (ICIP) , 2009

    2009

  10. [18]

    The 2017 Nimble Challenge Evaluation: Results and Future Directions, 2017

    Jonathan Fiscus, Haiying Guan, Yooyoung Lee, Amy Yates, Andrew Delgado, Daniel Zhou, David Joy, and August Pereira. The 2017 Nimble Challenge Evaluation: Results and Future Directions, 2017

    2017

  11. [19]

    Fridrich and J

    J. Fridrich and J. Kodovsky. Rich models for steganalysis of digital images. IEEE Transactions on Information F orensics and Security, 7(3):868–882, 06 2012

    2012

  12. [20]

    The ‘Dresden Image Database’ for benchmarking digital image forensics

    Thomas Gloe and Rainer Böhme. The ‘Dresden Image Database’ for benchmarking digital image forensics. In Proceedings of the 25th Symposium On Applied Computing (ACM SAC) , volume 2, pages 1585–1591, 2010

    2010

  13. [21]

    Goodfellow, Jean Pouget-Abadie, Mehdi Mirza, Bing Xu, David Warde-Farley, Sherjil Ozair, Aaron Courville, and Yoshua Bengio

    Ian J. Goodfellow, Jean Pouget-Abadie, Mehdi Mirza, Bing Xu, David Warde-Farley, Sherjil Ozair, Aaron Courville, and Yoshua Bengio. Generative adversarial nets. In NeurIPS, 2014

    2014

  14. [22]

    K. He, X. Zhang, S. Ren, and J. Sun. Deep residual learning for image recognition. In The IEEE Conference on Computer Vision and Pattern Recognition (CVPR), pages 770–778, 06 2016

    2016

  15. [23]

    In International Conference on Learning Representations (ICLR) , Apr 2017

    Irina Higgins, Loic Matthey, Arka Pal, Christopher Burgess, Xavier Glorot, Matthew Botvinick, Shakir Mohamed, and Alexander Lerchner.β-V AE: Learning basic visual concepts with a constrained variational framework. In International Conference on Learning Representations (ICLR) ...

    2017

  16. [24]

    Minyoung Huh, Andrew Liu, Andrew Owens, and Alexei A. Efros. Fighting fake news: Image splice detection via learned self-consistency. In Vittorio Ferrari, Martial Hebert, Cristian Sminchisescu, and Yair Weiss, editors, Proceedings of the European Conference on Computer Vision ...

    2018

  17. [25]

    Kingma and Max Welling

    Diederik P. Kingma and Max Welling. Auto-encoding variational bayes. In 2nd International Conference on Learning Representations, (ICLR) 2014, Banff, AB, Canada, April 14-16, 2014, Conference Track Proceedings , 2014

    2014

  18. [26]

    Image inpainting for irregular holes using partial convolutions

    Guilin Liu, Fitsum A Reda, Kevin J Shih, Ting-Chun Wang, Andrew Tao, and Bryan Catanzaro. Image inpainting for irregular holes using partial convolutions. In Proceedings of the European Conference on Computer Vision (ECCV), pages 85–100, 2018

    2018

  19. [27]

    Lukas, J

    J. Lukas, J. Fridrich, and M. Goljan. Detecting digital image forgeries using sensor pattern noise - art. no. 60720y. Proceedings of SPIE - The International Society for Optical Engineering , 6072:362–372, 02 2006

    2006

  20. [28]

    Using noise inconsistencies for blind image forensics

    Babak Mahdian and Stanislav Saic. Using noise inconsistencies for blind image forensics. Image and Vision Computing, 27(10):1497 – 1503, 2009. Special Section: Computer Vision Methods for Ambient Intelligence

    2009

  21. [29]

    Owen Mayer and Mathew C. Stamm. Learned forensic source similarity for unknown camera models. In IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP) . IEEE SigPort, 2018. 9 To Beta or Not To Beta: Information Bottleneck for Digital Image Forensics...

    2018

  22. [30]

    Edgeconnect: Generative image inpainting with adversarial edge learning

    Kamyar Nazeri, Eric Ng, Tony Joseph, Faisal Qureshi, and Mehran Ebrahimi. Edgeconnect: Generative image inpainting with adversarial edge learning. arXiv preprint arXiv:1901.00212, 2019

    1901 arXiv

  23. [31]

    N. Otsu. A threshold selection method from gray-level histograms. IEEE Transactions on Systems, Man, and Cybernetics, 9(1):62–66, Jan 1979

    1979

  24. [32]

    Variational Discriminator Bottleneck: Improving Imitation Learning, Inverse RL, and GANs by Constraining Information Flow

    Xue Bin Peng, Angjoo Kanazawa, Sam Toyer, Pieter Abbeel, and Sergey Levine. Variational Discriminator Bottleneck: Improving Imitation Learning, Inverse RL, and GANs by Constraining Information Flow. In International Conference on Learning Representations (ICLR) , May 2019

    2019

  25. [33]

    A. C. Popescu and H. Farid. Exposing digital forgeries in color filter array interpolated images. IEEE Transactions on Signal Processing, 53(10):3948–3959, 10 2005

    2005

  26. [34]

    On the information bottleneck theory of deep learning

    Andrew Michael Saxe, Yamini Bansal, Joel Dapello, Madhu Advani, Artemy Kolchinsky, Brendan Daniel Tracey, and David Daniel Cox. On the information bottleneck theory of deep learning. In International Conference on Learning Representations (ICLR), May 2018

    2018

  27. [35]

    Opening the Black Box of Deep Neural Networks via Information

    Ravid Shwartz-Ziv and Naftali Tishby. Opening the Black Box of Deep Neural Networks via Information. arXiv e-prints, page arXiv:1703.00810, Mar 2017

    2017 arXiv

  28. [36]

    Pereira, and William Bialek

    Naftali Tishby, Fernando C. Pereira, and William Bialek. The information bottleneck method. In Proc. of the 37-th Annual Allerton Conference on Communication, Control and Computing , pages 368–377, 1999

    1999

  29. [37]

    Generative image inpainting with contextual attention

    Jiahui Yu, Zhe Lin, Jimei Yang, Xiaohui Shen, Xin Lu, and Thomas S Huang. Generative image inpainting with contextual attention. In Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition (CVPR), pages 5505–5514, 2018

    2018

  30. [38]

    Large-scale evaluation of splicing localization algorithms for web images

    Markos Zampoglou, Symeon Papadopoulos, and Ioannis Kompatsiaris. Large-scale evaluation of splicing localization algorithms for web images. Multimedia Tools and Applications, 09 2016

    2016

  31. [39]

    Morariu, and Larry S

    Peng Zhou, Xintong Han, Vlad I. Morariu, and Larry S. Davis. Learning rich features for image manipulation detection. In The IEEE Conference on Computer Vision and Pattern Recognition (CVPR) , pages 1053–1061, 06 2018. 10

    2018

Pith tools

Reviewed August 14, 2026 · model on record in the stance chip above.