Pith. sign in

REVIEW 5 major objections 5 minor 1 cited by

Unicorn: Unified Neural Image Compression with One Number Reconstruction

T0 review · 5 major / 5 minor · reviewed 2026-08-11 · deepseek-v4-flash

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

Pith's one-line read A shared diffusion decoder can reconstruct each image in a set from just an index number, making per-image transmission cost fall as the set grows.

desk verdict Clever set-wise compression idea, but the main claim rests on an unstated deterministic decoding assumption. read the letter →

arxiv 2412.08210 v1 pith:XS7CU4GC submitted 2024-12-11 cs.CV eess.IV

Qi Zheng , Haozhi Wang , Zihao Liu , Jiaming Liu , Peiye Liu , Zhijian Hao , Yanheng Lu , Dimin Niu
show 3 more authors
Jinjia Zhou Minge Jing Yibo Fan
This is my paper · ORCID
classification cs.CVeess.IV
keywords imagecompressionlatentdiffusionmodelsunifieddecoderindex-imagebijectionperceptualqualityrate-distortionneuralone-numberreconstruction
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

The paper claims that an entire set of images can be compressed by training one shared decoder to memorize a random one-to-one assignment between images and index numbers. After training, each image is transmitted as only its index, and the decoder produces a visually pleasing picture from that index plus random noise; the only remaining cost is the quantized decoder, shared by all images. The central quantitative claim is that the description length of $M$ images is $M\log_2 M + K(q)$, where $K(q)$ is the size of the compressed decoder, so the per-image online cost falls to $\log_2 M$ bits once the decoder is installed. The paper's prototype, Laduree, is a latent diffusion transformer, and experiments on 4000-image sets report better perceptual quality at comparable bitrates than the per-image codecs ELIC and HiFiC, with savings that grow as more images are compressed together.

What carries the argument

The load-bearing object is the index-image pair with a shared conditional diffusion decoder: a bijection $\{(I_i, Y_i)\}_{i=1}^{M}$ between images and uniformly sampled indexes in $\{1,\dots,M\}$, learned by a latent diffusion model that denoises random noise conditioned on $Y_i$. The identity that carries the argument is the description-length formula $L^{\mathrm{Unicorn}}_{DL}(\tilde S) = M\log_2 M + K(q)$, which turns compression into memorization of random labels: the first term is the cost of transmitting $M$ indexes and the second is the quantized size of the unified decoder. Within the prototype, the cross-attention-with-gating block and the Gaussian random frequency embedding are the tailored parts chosen to introduce few extra parameters, and weight quantization supplies the model cost term $K(q)$.

What would settle it

Run the trained decoder twice with the same index and two different random initial noises and compare the reconstructions; if they diverge noticeably, the index alone is not a reproducible representation. A complementary check is to compute the actual storage cost for one image and for 4000 images, and test whether the total follows $M\log_2 M + K(q)$ as claimed.

Watch

Extended reading notes

Core claim

On its own terms, the discovery is that a single conditional diffusion decoder can learn a bijection between $M$ arbitrary indexes and $M$ images well enough that the index alone, together with a random denoising seed, determines the reconstructed image. The paper derives the description length of the whole set as $-\log_2 P_q(I|Y) + K(q) = M\log_2 M + K(q)$, treating the decoder as a shared codebook and the index as the codeword. The prototype trains this decoder on index-conditioned latent diffusion, using a Gaussian random frequency embedding for the index and a gated cross-attention block for conditioning, then quantizes the weights to keep $K(q)$ small. In experiments at high perceptual quality (LPIPS around 0.10), Laduree reports 21.73% bitrate savings over ELIC and 17.87% over HiFiC on a 4000-cat-image set, and similar trends on a five-category hybrid set, with the advantage increasing as the number of images grows.

Load-bearing premise

The decoding procedure assumes that the same index together with the same initial random noise always yields the same image; the paper does not specify the initial noise or verify this determinism, yet every bitrate comparison depends on it.

Editorial extensions

If this is right

  • If the claim holds, online transmission of an image becomes a $\log_2 M$-bit index lookup once the shared decoder is installed, so the per-image bitrate shrinks as the collection grows.
  • Total compressed size grows sublinearly in the number of images, so the compression ratio improves with $M$ instead of staying flat as it does for per-image codecs.
  • The scheme is strongest for semantically similar image sets, because the diffusion model can reuse shared structure across the memorized images; the cat set shows larger savings than the five-category hybrid set.
  • Quantizing the decoder weights to roughly 14 bits keeps quality losses acceptable, giving a concrete handle on trading decoder cost against distortion.
  • At the same time, the paper only matches the GAN-based codec HiFiC in PSNR, consistent with a rate-distortion-perception trade-off, so the gains are perceptual rather than pixel-fidelity gains.

Reading between the lines

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

  • An implication the authors leave implicit is that the scheme is essentially a learned lookup table, so the real scaling question is how decoder cost $K(q)$ and reconstruction quality behave as $M$ and image diversity grow, not whether the bijection can be memorized.
  • The paper never specifies a fixed initial noise per index or demonstrates same-index, same-seed determinism; a testable extension would fix a seed for each index and report reconstruction variance across restarts, since all bitrate comparisons assume the decoder is a deterministic function.
  • If the approach scales, it suggests a general 'compression by memorization' strategy in which the decoder is a shared codebook and the index is the codeword, naturally combinable with stronger model-compression and pre-trained priors.
  • At small $M$, the decoder cost $K(q)$ is not amortized and likely dominates, so the claimed advantage over per-image codecs could reverse for very small collections; the paper's comparisons start at 1000 images.
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

5 major / 5 minor

Summary. The manuscript proposes Unicorn, a paradigm for image compression in which a set of M images is assigned random indices and a single conditional latent diffusion model is trained to map (index, noise) to each image. The rate for the whole set is given in Eq. (4) as M log2 M + K(q), where K(q) is the encoded model size; at decode time the receiver uses the transmitted index together with randomly generated noise. A prototype, Laduree, is evaluated on 4000-image cat and hybrid sets against JPEG/HEVC/VVC, HiFiC/ELIC, and COIN/Combiner, reporting lower bitrates at matched perceptual quality and an increasing compression ratio as M grows. The paper additionally studies index embeddings, conditioning blocks, latent normalization, and weight quantization.

Significance. If the paradigm works as claimed, it offers a genuinely different viewpoint on dataset-specific compression, and the prototype's design explorations (CAG conditioning, GRF embedding, latent standardization) are useful engineering contributions. The paper also makes a concrete, externally testable claim: a single small conditional diffusion model can memorize thousands of index-image pairs and beat general-purpose codecs on perceptual rate-distortion for that fixed set. However, the current manuscript does not establish that decoding is deterministic, and Eq. (4) is not a rate-distortion statement as written. These gaps are central rather than cosmetic, so the paper needs substantive revision before the claims can be accepted.

major comments (5)
  1. [§1, §4.2, Eq. (4)] The decoding procedure is underspecified with respect to initial noise. The text says the decoding process is started from 'any random generated noise' (Sec. 1) and later that 'the index controls the latent denoising diffusion process' (Sec. 4.2), but it never fixes the initial noise, a seed, or a deterministic sampler. If the receiver draws a different initial noise than the encoder used, the same index reconstructs a different image; then the index-image mapping is not the bijection required for Eq. (4), and the rate-distortion comparisons in Secs. 5.2 and 5.3 are not well-defined. Please specify the exact decoding protocol (e.g., fixed initial noise, DDIM with a fixed seed) and report the variance of the reported metrics over decoding seeds for at least one operating point.
  2. [§4.1, Eq. (4)] Eq. (4) counts only the index code length and the model cost; it omits any distortion term. In the LMDL framework of Eq. (1), the description length is meaningful only for a fixed distortion D, but here the model reconstructs each image only approximately and no term accounts for the reconstruction error. Therefore Eq. (4) cannot by itself support the rate-distortion claims in Sec. 5. The authors should either replace the derivation with a rate-distortion formulation that includes the distortion, or explicitly state that Eq. (4) is only the rate component at an empirically measured distortion.
  3. [§5.3, Eq. (4)] The 'unique bitrate superiority' scaling with M is in large part a restatement of Eq. (4). If K(q) is held constant as M grows, the per-image cost is log2 M + K(q)/M, which necessarily decreases with M. The empirical contribution should be framed as demonstrating that a trainable model can maintain a near-constant K(q) while preserving quality as M grows; as written, the paper presents a mathematical consequence of its own rate formula as a discovered property. Please separate the definitional scaling from the empirical comparison against ELIC and HiFiC.
  4. [§5.1, §5.2, §5.3] The paper does not specify exactly how K(q) is converted to bits-per-pixel in the RD curves and file-size plots. It must state whether the VAE decoder and any model-compression side information are included, how the quantized weights are counted (number of parameters times W bits, or an entropy-coded length), and whether the same accounting is applied to the INR baselines. Without this, the reported bpp values in Figures 4-6 are not auditable and the comparison with COIN and Combiner may not be fair.
  5. [§3, §4.2, §5.1, §5.2] The manuscript repeatedly refers to appendices that are not present in the submitted text: Appendix 1.3 for the quantitative comparison between Unicorn and ELIC, Appendix 2.2 for the proof that the conditional reverse process reduces to an unconditional one, and Appendices 3.1, 4.1, and 4.2 for model configurations and extra results. The proof in Appendix 2.2 is load-bearing for the 'efficiency through similarity' claim, and Appendix 1.3 is load-bearing for the claim that Unicorn maintains superiority at 'normal magnitude' image counts. Please include the appendices or remove the references.
minor comments (5)
  1. [Abstract and title] There are typos in the abstract: 'One Nnumber Reconstruction' and 'Quantitive'; the main text also has 'rate-distrotion' in Sec. 4.2.
  2. [§5.1] The model naming convention 'Data-XHX W X' (e.g., CAT-1500H120 W 14) is introduced without a definition; please define the notation clearly before first use.
  3. [Figure 6] The y-axis label 'Compression ratio' is not defined; specify the reference (e.g., ratio of uncompressed size to compressed size, or ratio relative to a baseline) so the reader can interpret the plot.
  4. [§4.2] The sentence 'the bitrate consumption in compressing images only comes from the encoded weights of the latent diffusion model' appears to ignore the M log2 M index cost in Eq. (4); please reconcile the wording with the formal rate expression.
  5. [§4.1] The notation Pq(I|Y) is not defined precisely; in particular, it is unclear whether I denotes a random image, a latent feature Z, or the reconstructed image, and how the approximate nature of the reconstruction is represented in the probability model.

Circularity Check

1 steps flagged · score 6.0 of 10

The 'unique bitrate superiority' scaling claim is Eq. (4) restated; the prototype's RD comparisons are externally anchored.

  1. self definitional [Sec. 4.1, Eq. (4); Sec. 5.3 'Unique bitrate superiority']
    "we initialize a set of fake/random index uniformly sampled in {1, ..., M} to construct a bijection function with the image set ... The probability distribution is identically equal to a uniform distribution P (I|Y ) = 1/M ... Therefore, the DL of compressing eS can be computed as: LUnicorn DL ( eS) = − log2 Pq(I|Y ) + K(q) = M log2 M + K(q). (4) ... This unique bitrate superiority related to the quantity of images is consistent with the HYBRID dataset."

    Eq. (4) defines the total description length as M log2 M + K(q). Dividing by M gives a per-image rate of log2 M + K(q)/M, which decreases with M as long as the shared decoder cost K(q) is charged once. Sec. 5.3 then presents exactly this monotone decrease as an empirically discovered 'unique bitrate superiority' and attributes it to inter-image redundancy. The trend is a direct algebraic consequence of the paper's own bitrate accounting, not an independent finding of the trained model. The prototype's absolute RD comparisons to ELIC/HiFiC/COIN etc. are external and non-circular, but the headline scaling claim reduces to the definition in Eq. (4).

full rationale

The paper is mostly a self-contained empirical prototype: a conditional latent diffusion model is trained on actual image-index pairs, quantized, and its measured file sizes are compared against external baselines (ELIC, HiFiC, COIN, Combiner, JPEG/HEVC/VVC). Those rate-distortion comparisons are independent support and are not circular. The circular element is the scaling claim. Eq. (4) sets the total bitrate to M log2 M + K(q), so the per-image rate log2 M + K(q)/M falls with M whenever the decoder is shared; Sec. 5.3 reports that arithmetic consequence as a 'unique bitrate superiority' rather than as a consequence of the definition. No load-bearing self-citation chain is present; the cited overfitting and diffusion results are external. Separately, the paper's assumption that one index plus unspecified random noise yields one deterministic reconstruction is a correctness risk for the codec claim, but it is not a circularity under the definitions used here.

Assumptions & free parameters 5 free parameters · 4 assumptions · 0 invented entities

The central claim rests on the decoder weights being the compressed representation and on the decoder being shareable without counting the VAE. The hand-set latent scaling and bitrate hyperparameters show that the reported RD curves are obtained with several degrees of freedom beyond the mathematical formulation.

free parameters (5)
  • Diffusion model weights theta = Quantized weights at 10-32 bits, hidden sizes 108-240, 12 transformer blocks
    The model weights are the compressed payload K(q). They are fitted to each image set via MSE x0-prediction, so the compression result lives entirely in these fitted values.
  • Latent feature standard deviation = 1/3
    Chosen in Sec. 4.2 so latent features concentrate in [-1,1]. Table 1 shows this choice shifts LPIPS from 0.16 to 0.08 on CAT-1000, so it is a hand-set parameter with large effect.
  • Number of denoising timesteps = 50
    Fixed in Sec. 5.1 for all models; trades quality against compute and affects the variance of reconstructions.
  • Transformer hidden size H = 108-240 depending on dataset and subset
    Varied per dataset to produce RD operating points; larger H for more images, so the model cost K(q) grows with M.
  • Quantization precision W = 10-32 bits
    Varied to control bitrate; robustness is shown in Fig. 9, but the selection of W for each RD point is empirical.
assumptions (4)
  • domain assumption Neural networks can memorize arbitrary random label-image pairings.
    Invoked in Sec. 4.1 via Zhang et al. (2021) to justify learning a bijection between random indices and images. If this failed for large M, Eq. 4 would count an index code for images the decoder cannot actually reproduce.
  • ad hoc to paper Each index maps to a fixed reconstruction regardless of the random noise used to start decoding.
    Used implicitly in Sec. 4.1 and Fig. 2. The paper never states whether the same index plus different random noise yields the same image, yet every rate-distortion measurement presumes a deterministic decode.
  • domain assumption The pretrained VAE is a shared module whose bitrate is not counted.
    Sec. 4.2 states bitrate comes only from the latent diffusion model weights because the VAE is off-the-shelf. This requires the receiver to have the VAE or the sender to transmit it at no counted cost.
  • ad hoc to paper The unified decoder can be transmitted once and pre-shared with the receiver for all images in the set.
    This is the scenario that makes Eq. 4's M log2 M plus K(q) meaningful. If the decoder cannot be shared, per-image cost is dominated by K(q) and the compression advantage over EIC vanishes for small sets.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Unicorn: Unified Neural Image Compression with One Number Reconstruction." pith.science (2026). https://pith.science/paper/XS7CU4GC

@misc{pith2026241208210,
  author       = {Pith},
  title        = {Pith review of: Unicorn: Unified Neural Image Compression with One Number Reconstruction},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/XS7CU4GC}},
  note         = {Machine review of arXiv:2412.08210}
}
read the original abstract

Prevalent lossy image compression schemes can be divided into: 1) explicit image compression (EIC), including traditional standards and neural end-to-end algorithms; 2) implicit image compression (IIC) based on implicit neural representations (INR). The former is encountering impasses of either leveling off bitrate reduction at a cost of tremendous complexity while the latter suffers from excessive smoothing quality as well as lengthy decoder models. In this paper, we propose an innovative paradigm, which we dub \textbf{Unicorn} (\textbf{U}nified \textbf{N}eural \textbf{I}mage \textbf{C}ompression with \textbf{O}ne \textbf{N}number \textbf{R}econstruction). By conceptualizing the images as index-image pairs and learning the inherent distribution of pairs in a subtle neural network model, Unicorn can reconstruct a visually pleasing image from a randomly generated noise with only one index number. The neural model serves as the unified decoder of images while the noises and indexes corresponds to explicit representations. As a proof of concept, we propose an effective and efficient prototype of Unicorn based on latent diffusion models with tailored model designs. Quantitive and qualitative experimental results demonstrate that our prototype achieves significant bitrates reduction compared with EIC and IIC algorithms. More impressively, benefitting from the unified decoder, our compression ratio escalates as the quantity of images increases. We envision that more advanced model designs will endow Unicorn with greater potential in image compression. We will release our codes in \url{https://github.com/uniqzheng/Unicorn-Laduree}.

Figures

Figures reproduced from arXiv: 2412.08210 by the authors.

Figure 1
Figure 1. Bitrates comparison among EIC, IIC, and Unicorn [PITH_FULL_IMAGE:figures/full_fig_p001_1.png] view at source ↗
Figure 2
Figure 2. Overall framework of the proposed paradigm Unicorn specified by the proposed prototype Laduree. [PITH_FULL_IMAGE:figures/full_fig_p003_2.png] view at source ↗
Figure 3
Figure 3. Design space explorations on various manners for index embedding and condition within Transformer-based denoising [PITH_FULL_IMAGE:figures/full_fig_p004_3.png] view at source ↗
Figures from the paper (5 more)
Figure 4
Figure 4. Figure 4: RD curves of evaluated image compression models on CAT (top row) and HYBRID (bottom row) when compressing [PITH_FULL_IMAGE:figures/full_fig_p006_4.png]
Figure 5
Figure 5. Figure 5: RD performance comparison in terms of PSNR. [PITH_FULL_IMAGE:figures/full_fig_p006_5.png]
Figure 7
Figure 7. Figure 7: Visual comparison among evaluated image compression models. Better zoom in. [PITH_FULL_IMAGE:figures/full_fig_p007_7.png]
Figure 9
Figure 9. Figure 9: Robustness evaluation on weight quantization. [PITH_FULL_IMAGE:figures/full_fig_p007_9.png]
Figure 8
Figure 8. Figure 8: RD performance comparison of index embedding [PITH_FULL_IMAGE:figures/full_fig_p007_8.png]

Discussion (0). Continue with ORCID to comment.

Forward citations

Cited by 1 Pith paper

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score. Full citation record

  1. M3-CVC: Controllable Video Compression with Multimodal Generative Models

    eess.IV 2024-11 conditional novelty 5.0 of 10

    M3-CVC compresses video at ultra-low bitrates by encoding keyframes as discrete latents plus text prompts, then regenerates clips with a video diffusion model, claiming large rate reductions over VVC.

Reference graph

Works this paper leans on

41 extracted references · 18 canonical work pages · cited by 1 Pith paper

  1. [1]

    Agustsson, E.; Tschannen, M.; Mentzer, F.; Timofte, R.; and Gool, L. V. 2019. Generative adversarial networks for extreme learned image compression. In Proceedings of the IEEE/CVF International Conference on Computer Vision, 221--231

    work page 2019

  2. [2]

    J.; and Johnston, N

    Ball \'e , J.; Minnen, D.; Singh, S.; Hwang, S. J.; and Johnston, N. 2018. Variational image compression with a scale hyperprior. arXiv preprint arXiv:1802.01436

    arXiv 2018

  3. [3]

    Blau, Y.; and Michaeli, T. 2019. Rethinking lossy compression: The rate-distortion-perception tradeoff. In International Conference on Machine Learning, 675--685. PMLR

    work page 2019

  4. [4]

    Blier, L.; and Ollivier, Y. 2018. The description length of deep learning models. Advances in Neural Information Processing Systems, 31

    work page 2018

  5. [5]

    Bossen, F.; S \"u hring, K.; Wieckowski, A.; and Liu, S. 2021. VVC complexity and software implementation analysis. IEEE Transactions on Circuits and Systems for Video Technology, 31(10): 3765--3778

    work page 2021

  6. [6]

    J.; and Ohm, J.-R

    Bross, B.; Wang, Y.-K.; Ye, Y.; Liu, S.; Chen, J.; Sullivan, G. J.; and Ohm, J.-R. 2021. Overview of the versatile video coding (VVC) standard and its applications. IEEE Transactions on Circuits and Systems for Video Technology, 31(10): 3736--3764

    work page 2021

  7. [7]

    J.; Verbeek, J.; and Lathuili \`e re, S

    Careil, M.; Muckley, M. J.; Verbeek, J.; and Lathuili \`e re, S. 2023. Towards image compression with perfect realism at ultra-low bitrates. In The Twelfth International Conference on Learning Representations

    work page 2023

  8. [8]

    K.; Aitchison, M.; Orseau, L.; et al

    Del \'e tang, G.; Ruoss, A.; Duquenne, P.-A.; Catt, E.; Genewein, T.; Mattern, C.; Grau-Moya, J.; Wenliang, L. K.; Aitchison, M.; Orseau, L.; et al. 2023. Language modeling is compression. arXiv preprint arXiv:2309.10668

    arXiv 2023

Show all 41 references
  1. [9]

    Ding, K.; Ma, K.; Wang, S.; and Simoncelli, E. P. 2020. Image quality assessment: Unifying structure and texture similarity. IEEE transactions on pattern analysis and machine intelligence, 44(5): 2567--2581

    2020

  2. [10]

    W.; and Doucet, A

    Dupont, E.; Goli \'n ski, A.; Alizadeh, M.; Teh, Y. W.; and Doucet, A. 2021. Coin: Compression with implicit neural representations. arXiv preprint arXiv:2103.03123

    2021 arXiv

  3. [11]

    Guo, Z.; Flamich, G.; He, J.; Chen, Z.; and Hern \'a ndez-Lobato, J. M. 2023. Compression with bayesian implicit neural representations. Advances in Neural Information Processing Systems, 36: 1938--1956

    2023

  4. [12]

    He, D.; Yang, Z.; Peng, W.; Ma, R.; Qin, H.; and Wang, Y. 2022. Elic: Efficient learned image compression with unevenly grouped space-channel contextual adaptive coding. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 5718--5727

    2022

  5. [13]

    He, D.; Zheng, Y.; Sun, B.; Wang, Y.; and Qin, H. 2021. Checkerboard context model for efficient learned image compression. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 14771--14780

    2021

  6. [14]

    Heusel, M.; Ramsauer, H.; Unterthiner, T.; Nessler, B.; and Hochreiter, S. 2017. Gans trained by a two time-scale update rule converge to a local nash equilibrium. Advances in neural information processing systems, 30

    2017

  7. [15]

    Ho, J.; Jain, A.; and Abbeel, P. 2020. Denoising diffusion probabilistic models. Advances in neural information processing systems, 33: 6840--6851

    2020

  8. [16]

    Hu, Y.; Yang, W.; Ma, Z.; and Liu, J. 2021. Learning end-to-end lossy image compression: A benchmark. IEEE Transactions on Pattern Analysis and Machine Intelligence, 44(8): 4194--4211

    2021

  9. [17]

    Kahan, W. 1996. IEEE standard 754 for binary floating-point arithmetic. Lecture Notes on the Status of IEEE, 754(94720-1776): 11

    1996

  10. [18]

    Kolmogorov, A. N. 1998. On tables of random numbers. Theoretical Computer Science, 207(2): 387--395

    1998

  11. [19]

    Kontoyiannis, I.; and Zhang, J. 2002. Arbitrary source models and Bayesian codebooks in rate-distortion theory. IEEE Transactions on information theory, 48(8): 2276--2290

    2002

  12. [20]

    Ladune, T.; Philippe, P.; Henry, F.; Clare, G.; and Leguay, T. 2023 a . COOL-CHIC: Coordinate-based Low Complexity Hierarchical Image Codec. In Proceedings of the IEEE/CVF International Conference on Computer Vision (ICCV), 13515--13522

    2023

  13. [21]

    Ladune, T.; Philippe, P.; Henry, F.; Clare, G.; and Leguay, T. 2023 b . Cool-chic: Coordinate-based low complexity hierarchical image codec. In Proceedings of the IEEE/CVF International Conference on Computer Vision, 13515--13522

    2023

  14. [22]

    Ma, Y.; Derksen, H.; Hong, W.; and Wright, J. 2007. Segmentation of multivariate mixed data via lossy data coding and compression. IEEE transactions on pattern analysis and machine intelligence, 29(9): 1546--1562

    2007

  15. [23]

    Madiman, M.; Harrison, M.; and Kontoyiannis, I. 2004. Minimum description length vs. maximum likelihood in lossy data compression. In IEEE International Symposium on Information Theory, 461--461

    2004

  16. [24]

    D.; Tschannen, M.; and Agustsson, E

    Mentzer, F.; Toderici, G. D.; Tschannen, M.; and Agustsson, E. 2020. High-fidelity generative image compression. Advances in Neural Information Processing Systems, 33: 11913--11924

    2020

  17. [25]

    K.; and Bovik, A

    Mittal, A.; Moorthy, A. K.; and Bovik, A. C. 2012. No-reference image quality assessment in the spatial domain. IEEE Trans. Image Process., 21(12): 4695--4708

    2012

  18. [26]

    Completely Blind

    Mittal, A.; Soundararajan, R.; and Bovik, A. C. 2013. Making a “Completely Blind” Image Quality Analyzer. IEEE Signal Processing Letters, 20(3): 209--212

    2013

  19. [27]

    Nikankin, Y.; Haim, N.; and Irani, M. 2022. Sinfusion: Training diffusion models on a single image or video. arXiv preprint arXiv:2211.11743

    2022 arXiv

  20. [28]

    Peebles, W.; and Xie, S. 2023. Scalable diffusion models with transformers. In Proceedings of the IEEE/CVF International Conference on Computer Vision, 4195--4205

    2023

  21. [29]

    Rombach, R.; Blattmann, A.; Lorenz, D.; Esser, P.; and Ommer, B. 2022. High-resolution image synthesis with latent diffusion models. In Proceedings of the IEEE/CVF conference on computer vision and pattern recognition, 10684--10695

    2022

  22. [30]

    Russakovsky, O.; Deng, J.; Su, H.; Krause, J.; Satheesh, S.; Ma, S.; Huang, Z.; Karpathy, A.; Khosla, A.; Bernstein, M.; et al. 2015. Imagenet large scale visual recognition challenge. International journal of computer vision, 115: 211--252

    2015

  23. [31]

    Skodras, A.; Christopoulos, C.; and Ebrahimi, T. 2001. The JPEG 2000 still image compression standard. IEEE Signal processing magazine, 18(5): 36--58

    2001

  24. [32]

    V.; and Tombari, F

    Str \"u mpler, Y.; Postels, J.; Yang, R.; Gool, L. V.; and Tombari, F. 2022. Implicit neural representations for image compression. In European Conference on Computer Vision, 74--91. Springer

    2022

  25. [33]

    J.; Ohm, J.-R.; Han, W.-J.; and Wiegand, T

    Sullivan, G. J.; Ohm, J.-R.; Han, W.-J.; and Wiegand, T. 2012. Overview of the high efficiency video coding (HEVC) standard. IEEE Transactions on circuits and systems for video technology, 22(12): 1649--1668

    2012

  26. [34]

    N.; Kaiser, .; and Polosukhin, I

    Vaswani, A.; Shazeer, N.; Parmar, N.; Uszkoreit, J.; Jones, L.; Gomez, A. N.; Kaiser, .; and Polosukhin, I. 2017. Attention is all you need. Advances in neural information processing systems, 30

    2017

  27. [35]

    Wallace, G. K. 1991. The JPEG still picture compression standard. Communications of the ACM, 34(4): 30--44

    1991

  28. [36]

    C.; and Loy, C

    Wang, J.; Chan, K. C.; and Loy, C. C. 2023. Exploring clip for assessing the look and feel of images. In Proceedings of the AAAI Conference on Artificial Intelligence, volume 37, 2555--2563

    2023

  29. [37]

    Yang, R.; and Mandt, S. 2024. Lossy image compression with conditional diffusion models. Advances in Neural Information Processing Systems, 36

    2024

  30. [38]

    Zhang, C.; Bengio, S.; Hardt, M.; Recht, B.; and Vinyals, O. 2021. Understanding deep learning (still) requires rethinking generalization. Communications of the ACM, 64(3): 107--115

    2021

  31. [39]

    A.; Shechtman, E.; and Wang, O

    Zhang, R.; Isola, P.; Efros, A. A.; Shechtman, E.; and Wang, O. 2018. The unreasonable effectiveness of deep features as a perceptual metric. In Proceedings of the IEEE conference on computer vision and pattern recognition, 586--595

    2018

  32. [40]

    , " * write output.state after.block = add.period write newline

    ENTRY address archivePrefix author booktitle chapter edition editor eid eprint howpublished institution isbn journal key month note number organization pages publisher school series title type volume year label extra.label sort.label short.list INTEGERS output.state before.all...

  33. [41]

    write newline

    " write newline "" before.all 'output.state := FUNCTION n.dashify 't := "" t empty not t #1 #1 substring "-" = t #1 #2 substring "--" = not "--" * t #2 global.max substring 't := t #1 #1 substring "-" = "-" * t #2 global.max substring 't := while if t #1 #1 substring * t #2 gl...

Pith tools

Reviewed August 11, 2026 · model on record in the stance chip above.