Filters








4,799 Hits in 5.1 sec

Evaluating Lossy Compression Rates of Deep Generative Models [article]

Sicong Huang, Alireza Makhzani, Yanshuai Cao, Roger Grosse
2020 arXiv   pre-print
We evaluate lossy compression rates of VAEs, GANs, and adversarial autoencoders (AAEs) on the MNIST and CIFAR10 datasets.  ...  In this work, we propose to use rate distortion (RD) curves to evaluate and compare deep generative models.  ...  We aim to achieve the best of both worlds by measuring lossy compression rates of deep generative models.  ... 
arXiv:2008.06653v1 fatcat:o4doah7oszatrbd4et4r75mcjq

End-to-end lossless compression of high precision depth maps guided by pseudo-residual [article]

Yuyang Wu, Wei Gao
2022 arXiv   pre-print
The deep lossless compression network consists of two sub-networks, named lossy compression network and lossless compression network.  ...  We leverage the concept of pseudo-residual to guide the generation of distribution for residual and avoid introducing context models.  ...  It first uses a deep lossy compression model to compress the input image and then use a context model to estimate the likelihood of the residual.  ... 
arXiv:2201.03195v1 fatcat:haaknkzb2bgujplpb3vqk6v26y

On the Impact of Lossy Image and Video Compression on the Performance of Deep Convolutional Neural Network Architectures

Matt Poyser, Amir Atapour-Abarghouei, Toby P. Breckon
2021 2020 25th International Conference on Pattern Recognition (ICPR)  
Recent advances in generalized image understanding have seen a surge in the use of deep convolutional neural networks (CNN) across a broad range of image-based detection, classification and prediction  ...  performance of such deep learning architectures.  ...  CONCLUSION This study has investigated the impact of lossy image compression on a multitude of existing deep CNN architectures.  ... 
doi:10.1109/icpr48806.2021.9412455 fatcat:weoq2fm6tjffplnwqwtuztr3sq

On the Impact of Lossy Image and Video Compression on the Performance of Deep Convolutional Neural Network Architectures [article]

Matt Poyser, Amir Atapour-Abarghouei, Toby P. Breckon
2020 arXiv   pre-print
Recent advances in generalized image understanding have seen a surge in the use of deep convolutional neural networks (CNN) across a broad range of image-based detection, classification and prediction  ...  performance of such deep learning architectures.  ...  CONCLUSION This study has investigated the impact of lossy image compression on a multitude of existing deep CNN architectures.  ... 
arXiv:2007.14314v1 fatcat:pn2sfavdkbdqzhevcsi75xymzy

Learned Lossless Image Compression with a HyperPrior and Discretized Gaussian Mixture Likelihoods [article]

Zhengxue Cheng, Heming Sun, Masaru Takeuchi, Jiro Katto
2020 arXiv   pre-print
This paper generalizes the hyperprior from lossy model to lossless compression, and proposes a L2-norm term into the loss function to speed up training procedure.  ...  Recently, deep learning based approaches have started to show the potential at this point. HyperPrior is an effective technique proposed for lossy image compression.  ...  ACKNOWLEDGEMENT The authors would like to thank Fabian Mentzer (the first author of L3C) for fruitful discussion and insightful feedback on the evaluation methods and datasets.  ... 
arXiv:2002.01657v1 fatcat:h6xxus4njnbtxfw6rxdfmr4vs4

Dynamic-Deep: Tune ECG Task Performance and Optimize Compression in IoT Architectures [article]

Eli Brosh, Elad Wasserstein, Anat Bremler-Barr
2022 arXiv   pre-print
We conduct an extensive evaluation of our approach on common ECG datasets using two popular ECG applications, which includes heart rate (HR) arrhythmia classification.  ...  A lossy compression provides high compression gain (CG), but may reduce the performance of an ECG application (downstream task) due to information loss.  ...  Acknowledgment -We thank Guy Vinograd from bio-T for his comments, and deeply grateful to Elad Levy for his valuable feedback on model design and analysis.  ... 
arXiv:2106.00606v2 fatcat:6nuyalbfi5cw7oskiuwe6brlx4

Deep Perceptual Compression [article]

Yash Patel, Srikar Appalaraju, R. Manmatha
2019 arXiv   pre-print
We then propose Deep Perceptual Compression (DPC) which makes use of an encoder-decoder based image compression model to jointly optimize on the deep perceptual metric and MS-SSIM.  ...  Several deep learned lossy compression techniques have been proposed in the recent literature.  ...  We now discuss the internals of Perceptual loss and its effects on lossy image compression. Deep Perceptual Loss Zhang et al. [48] show the utility of deep CNNs to measure perceptual similarity.  ... 
arXiv:1907.08310v2 fatcat:6t7fvosy4bcf5gfbnrxvvqo3re

A Novel Memory-Efficient Deep Learning Training Framework via Error-Bounded Lossy Compression [article]

Sian Jin, Guanpeng Li, Shuaiwen Leon Song, Dingwen Tao
2020 arXiv   pre-print
Different from the state-of-the-art solutions that adopt image-based lossy compressors such as JPEG to compress the activation data, our framework purposely designs error-bounded lossy compression with  ...  We evaluate our design against state-of-the-art solutions with four popular DNNs and the ImageNet dataset.  ...  Different lossy compressors can provide different compression modes, such as error-bounded mode and fixed-rate mode.  ... 
arXiv:2011.09017v1 fatcat:qcsstdu7ozblpo6k5tr3x77edy

Lossy Compression with Distortion Constrained Optimization

Ties van Rozendaal, Guillaume Sautiere, Taco S. Cohen
2020 2020 IEEE/CVF Conference on Computer Vision and Pattern Recognition Workshops (CVPRW)  
When training end-to-end learned models for lossy compression, one has to balance the rate and distortion losses.  ...  We argue that the constrained optimization method of Rezende and Viola, 2018 [29] is a lot more appropriate for training lossy compression models because it allows us to obtain the best possible rate subject  ...  In this paper we focus on the implementation and evaluation of constrained optimization for practical lossy image compression.  ... 
doi:10.1109/cvprw50498.2020.00091 dblp:conf/cvpr/RozendaalSC20 fatcat:alteuvh5abbqjhc2ohdqpsywpm

Exploring Autoencoder-based Error-bounded Compression for Scientific Data [article]

Jinyang Liu, Sheng Di, Kai Zhao, Sian Jin, Dingwen Tao, Xin Liang, Zizhong Chen, Franck Cappello
2022 arXiv   pre-print
investigation of the characteristics of various autoencoder models and develop an error-bounded autoencoder-based framework in terms of the SZ model. (2) We optimize the compression quality for main stages  ...  Error-bounded lossy compression is becoming an indispensable technique for the success of today's scientific projects with vast volumes of data produced during the simulations or instrument data acquisitions  ...  Department of Energy, Office of Science, under contract DE-AC02-06CH11357, and supported by the National Science Foundation under Grant OAC-2003709 and OAC-2003624/2042084.  ... 
arXiv:2105.11730v6 fatcat:vqt2evw6unbeppdrwcidnjirj4

The Bearable Lightness of Big Data: Towards Massive Public Datasets in Scientific Machine Learning [article]

Wai Tong Chung and Ki Sung Jung and Jacqueline H. Chen and Matthias Ihme
2022 arXiv   pre-print
In general, large datasets enable deep learning models to perform with good accuracy and generalizability.  ...  To illustrate this point, we demonstrate that deep learning models, trained and tested on data from a petascale CFD simulation, are robust to errors introduced during lossy compression in a semantic segmentation  ...  Acknowledgments The authors acknowledge funding support from the Department of Energy (DoE) Office of Basic Energy Sciences under award DE-SC002222.  ... 
arXiv:2207.12546v1 fatcat:crnxcw7zcbg5xmhyaxgd7hfglq

Restoring degraded speech via a modified diffusion model [article]

Jianwei Zhang, Suren Jayasuriya, Visar Berisha
2021 arXiv   pre-print
Post-training, only the degraded mel-spectrum is used as input and the model generates an estimate of the original speech.  ...  Improvements over baseline are further amplified in a out-of-corpus evaluation setting.  ...  Lossy Operations: In this paper, we conduct three experiments to evaluate our model: 1) Restoring speech compressed by the LPC-10 algorithm [16] , 2) Restoring speech compressed by the AMR-NB algorithm  ... 
arXiv:2104.11347v1 fatcat:emamqehflnbufljzmznwuigwyu

Real-time Pedestrian Detection Approach with an Efficient Data Communication Bandwidth Strategy [article]

Mizanur Rahman, Mhafuzul Islam, Jon Calhoun, Mashrur Chowdhury
2019 arXiv   pre-print
We utilize a lossy compression technique on traffic camera data to determine the tradeoff between the reduction of the communication bandwidth requirements and a defined object detection accuracy.  ...  Because of this limitation, multiple traffic cameras at the signalized intersection can be used to accurately detect and locate pedestrians using deep learning and broadcast safety alerts related to pedestrians  ...  Figure 1 presents the general framework for the real-time pedestrian detection using the YOLOv3 (YOLO Model -version 3) deep learning model combined with lossy data compression.  ... 
arXiv:1808.09023v3 fatcat:ntfv74v4ind7tgtexkpqupzpwa

Communication-oriented Model Fine-tuning for Packet-loss Resilient Distributed Inference under Highly Lossy IoT Networks [article]

Sohei Itahara, Takayuki Nishio, Yusuke Koda, Koji Yamamoto
2021 arXiv   pre-print
In COMtune, the key idea is to fine-tune the ML model by emulating the effect of unreliable communication links through the application of the dropout technique.  ...  The distributed inference (DI) framework has gained traction as a technique for real-time applications empowered by cutting-edge deep machine learning (ML) on resource-constrained Internet of things (IoT  ...  Generally, the lossy compression method in DNN literature implies the compression of both, the activation and Dimensional reduction.  ... 
arXiv:2112.09407v1 fatcat:4fxyo5uqbbbepg2truxpikf3u4

Image Compression Using Deep Learning: Methods and Techniques

Arwa Sahib Abd-Alzhra, Mohammed S. H. Al- Tamimi
2022 Iraqi Journal of Science  
This article survey most common techniques and methods of image compression focusing on auto-encoder of deep learning.  ...  Image compression generally represents the significant image information compactly with a smaller size of bytes while insignificant image information (redundancy) already been removed for this reason image  ...  coding.Inspired by Dense Net the tightly linked AutoEncoder architecture for lossy compression.The compression model of the Deep Densely AutoEncoder is been tested on a subset of Challenge on Learned  ... 
doi:10.24996/ijs.2022.63.3.34 fatcat:43hnfu33krahrarvnab4ul3rq4
« Previous Showing results 1 — 15 out of 4,799 results