Lossy Compression with Distortion Constrained Optimization

Rozendaal, Ties van; Sautiere, Guillaume; Cohen, Taco

doi:10.1109/cvprw50498.2020.00091

Cited by 10 publications

(6 citation statements)

References 11 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Remark 1. In the previous techniques [4,[6][7][8][9][10][12][13][14][15][16], a loss function of the form Ref. [4] JPEG Ref.…”

Section: Our Methods 21 Loss Functionmentioning

confidence: 99%

“…They then transmitted the latent representation of a single target image after quantization and entropy coding. This technique has been improved by reformulating the loss function in [6][7][8][9]. Toderici et al [10] proposed another technique based on a recurrent NN [11] in which they optimized the network parameters for many images (as in [4,[6][7][8][9]) and transmitted residual information in addition to the latent representation.…”

Section: Motivationmentioning

confidence: 99%

“…This technique has been improved by reformulating the loss function in [6][7][8][9]. Toderici et al [10] proposed another technique based on a recurrent NN [11] in which they optimized the network parameters for many images (as in [4,[6][7][8][9]) and transmitted residual information in addition to the latent representation. A successor of this technique has been proposed by the same research group [12].…”

Section: Motivationmentioning

confidence: 99%

“…Various quantization and entropy coding techniques for a latent representation y have been studied extensively in previous works [4,[6][7][8][9][10][12][13][14][15][16]. Following [4], we encode y in (2) by using ordinary scalar quantization and context-based adaptive binary arithmetic coding [20].…”

Section: Coding Of Y and ψmentioning

confidence: 99%

See 3 more Smart Citations

An Efficient Image Compression Method Based On Neural Network: An Overfitting Approach

Mikami

Tsutake

Takahashi

et al. 2021

2021 IEEE International Conference on Image Processing (ICIP)

View full text Add to dashboard Cite

Over the past decade, nonlinear image compression techniques based on neural networks have been rapidly developed to achieve more efficient storage and transmission of images compared with conventional linear techniques. A typical nonlinear technique is implemented as a neural network trained on a vast set of images, and the latent representation of a target image is transmitted. In contrast to the previous nonlinear techniques, we propose a new image compression method in which a neural network model is trained exclusively on a single target image, rather than a set of images. Such an overfitting strategy enables us to embed fine image features in not only the latent representation but also the network parameters, which helps reduce the reconstruction error against the target image. The effectiveness of our method is validated through a comparison with conventional image compression techniques in terms of a rate-distortion criterion.

show abstract

“…Remark 1. In the previous techniques [4,[6][7][8][9][10][12][13][14][15][16], a loss function of the form Ref. [4] JPEG Ref.…”

Section: Our Methods 21 Loss Functionmentioning

confidence: 99%

Section: Motivationmentioning

confidence: 99%

Section: Motivationmentioning

confidence: 99%

Section: Coding Of Y and ψmentioning

confidence: 99%

See 2 more Smart Citations

An Efficient Image Compression Method Based On Neural Network: An Overfitting Approach

Mikami

Tsutake

Takahashi

et al. 2021

2021 IEEE International Conference on Image Processing (ICIP)

View full text Add to dashboard Cite

show abstract

“…Most current endto-end learned lossy compression methods to be discussed in Section 3.2 follow this approach, training one codec for each λ. We note that theoretically, it is not always possible to attain every point on the operational R-D curve with this approach [82], and alternative approaches exist [83], [84].…”

Section: Rate-distortion Theorymentioning

confidence: 99%

An Introduction to Neural Data Compression

Yang¹,

Mandt²,

Theis³

2022

Preprint

View full text Add to dashboard Cite

Neural compression is the application of neural networks and other machine learning methods to data compression. While machine learning deals with many concepts closely related to compression, entering the field of neural compression can be difficult due to its reliance on information theory, perceptual metrics, and other knowledge specific to the field. This introduction hopes to fill in the necessary background by reviewing basic coding topics such as entropy coding and rate-distortion theory, related machine learning ideas such as bits-back coding and perceptual metrics, and providing a guide through the representative works in the literature so far.

show abstract