3,359 Hits in 5.5 sec

New Bounds For Distributed Mean Estimation and Variance Reduction [article]

Peter Davies, Vijaykrishna Gurunathan, Niusha Moshrefi, Saleh Ashkboos, Dan Alistarh
2021 arXiv   pre-print
We also provide lower bounds showing that the communication to error trade-off of our algorithms is asymptotically optimal.  ...  Previous work typically assumes an upper bound on the norm of the input vectors, and achieves an error bound in terms of this norm.  ...  We will show a trade-off between bits of communication and output variance for both problems; in the case of VarianceReduction, though, there is an 'upper limit' to this trade-off, since we cannot go below  ... 
arXiv:2002.09268v4 fatcat:vldzamh7tba6ladtgkh6o6ckx4

CAE-ADMM: Implicit Bitrate Optimization via ADMM-based Pruning in Compressive Autoencoders [article]

Haimeng Zhao, Peiyuan Liao
2019 arXiv   pre-print
We introduce ADMM-pruned Compressive AutoEncoder (CAE-ADMM) that uses Alternative Direction Method of Multipliers (ADMM) to optimize the trade-off between distortion and efficiency of lossy image compression  ...  Specifically, ADMM in our method is to promote sparsity to implicitly optimize the bitrate, different from entropy estimators used in the previous research.  ...  The goal of such network is to optimize the trade-off between the amount of distortion and the efficiency of the compression, usually expressed by the bitrate or bits per pixel (bpp).  ... 
arXiv:1901.07196v4 fatcat:l4me45vydjfgrpx7tl3p73plsa

Deep Learning for the Gaussian Wiretap Channel [article]

Rick Fritschek, Rafael F. Schaefer, Gerhard Wunder
2019 arXiv   pre-print
This secure loss function approach is applied in a Gaussian wiretap channel setup, for which it is shown that the neural network learns a trade-off between reliable communication and information secrecy  ...  In this approach, neural networks learn to simultaneously optimize encoding and decoding functions to establish reliable message transmission.  ...  For that we introduced a modified loss function which results in a trade-off between security and communication rate.  ... 
arXiv:1810.12655v2 fatcat:6unu4li36feifidh62swqpg2ra

Simple and Effective VAE Training with Calibrated Decoders [article]

Oleh Rybkin, Kostas Daniilidis, Sergey Levine
2021 arXiv   pre-print
We study the impact of calibrated decoders, which learn the uncertainty of the decoding distribution and can determine this amount of information automatically, on the VAE performance.  ...  We further propose a simple but novel modification to the commonly used Gaussian decoder, which computes the prediction variance analytically.  ...  Here, we show that the crucial trade-off also controlled by β is the trade-off between two components of the rate itself, which control expressivity of representations and the match between the variational  ... 
arXiv:2006.13202v3 fatcat:xcvj5qhmprfjllupbtcvo5nevu

Bayesian Distributional Policy Gradients [article]

Luchen Li, A. Aldo Faisal
2021 arXiv   pre-print
off exploration and exploitation and policy learning in general.  ...  We demonstrate in a suite of Atari 2600 games and MuJoCo tasks, including well known hard-exploration challenges, how BDPG learns generally faster and with higher asymptotic performance than reference  ...  In our method W p (p X , G θ# Q) acts as an upper bound when Q = p Z , whereas in EM the surrogate objective is a lower bound.  ... 
arXiv:2103.11265v2 fatcat:f5ykig2iqvbixblazqbsamo3nu

UOLO - Automatic Object Detection and Segmentation in Biomedical Images [chapter]

Teresa Araújo, Guilherme Aresta, Adrian Galdran, Pedro Costa, Ana Maria Mendonça, Aurélio Campilho
2018 Lecture Notes in Computer Science  
UOLO is trained on a set of bounding boxes enclosing the objects to detect, as well as pixel-wise segmentation information, when available.  ...  We propose UOLO, a novel framework for the simultaneous detection and segmentation of structures of interest in medical images.  ...  However, one has to consider the trade-off between computational burden and performance, since UOLO network has 23 347 063 parameters, whereas U-Net has 15 063 985 and YOLOv2 has 21 831 470, being that  ... 
doi:10.1007/978-3-030-00889-5_19 fatcat:syiervmqgzak5josstien4qgaa

Fast Interleaved Bidirectional Sequence Generation [article]

Biao Zhang, Ivan Titov, Rico Sennrich
2020 arXiv   pre-print
To achieve even higher speedups, we explore hybrid models where we either simultaneously predict multiple neighbouring tokens per direction, or perform multi-directional decoding by partitioning the target  ...  Notably, it outperforms left-to-right SA because the independence assumptions in IBDecoder are more felicitous.  ...  Effect of c Results in Figure 4 show that c controls the trade-off between translation quality and speedup.  ... 
arXiv:2010.14481v1 fatcat:niqyamzo3rhpbmsxohx2wkjovu

Few-Shot Keypoint Detection as Task Adaptation via Latent Embeddings [article]

Mel Vecerik and Jackie Kay and Raia Hadsell and Lourdes Agapito and Jon Scholz
2021 arXiv   pre-print
We present results illustrating this capacity vs. accuracy trade-off, and demonstrate the ability to zero-shot transfer to new object instances (within-class) using a real-robot pick-and-place task.  ...  In this paper we explore a middle ground based on the observation that the number of relevant points at a given time are typically relatively few, e.g. grasp points on a target object.  ...  To E XAMPLES FROM S IMULATED DATASET explore this we chose saliency maps as in [23] to explore In Fig. 18 we show examples from the evaluation dataset both the encoder and decoder  ... 
arXiv:2112.04910v2 fatcat:nzxyvwp2mjch7odt2bwxjoccf4

Scalable Neural Decoder for Topological Surface Codes [article]

Kai Meinerz, Chae-Yeun Park, Simon Trebst
2021 arXiv   pre-print
to other recent machine learning inspired decoders) and exhibits faster decoding times than the state-of-the-art union find decoder for a wide range of error rates (down to 1%).  ...  Here we present a neural network based decoder that, for a family of stabilizer codes subject to depolarizing noise and syndrome measurement errors, is scalable to tens of thousands of qubits (in contrast  ...  -In the days prior to submission of this manuscript a tensor network decoder has been introduced [43] , for which an error threshold for depolarizing noise in the toric code of p th = 0.1881(3) is reported  ... 
arXiv:2101.07285v2 fatcat:7f2pxwj7h5gopjpqhd5oxskqbq

To Learn or Not to Learn: Deep Learning Assisted Wireless Modem Design [article]

S. Xue, A. Li, J. Wang, N. Yi, Y. Ma, R. Tafazolli, T. Dodgson
2019 arXiv   pre-print
Through several physical-layer case studies, we argue for a significant role that machine learning could play, for instance in parallel error-control coding and decoding, channel equalization, interference  ...  In addition, we will also discuss the fundamental bottlenecks of machine learning as well as their potential solutions in this paper.  ...  trade-off between performances and complexities.  ... 
arXiv:1909.07791v1 fatcat:ggschowl2fc4xemexn2i6pcz5q

Controllable Guarantees for Fair Outcomes via Contrastive Information Estimation [article]

Umang Gupta and Aaron M Ferber and Bistra Dilkina and Greg Ver Steeg
2021 arXiv   pre-print
Controlling bias in training datasets is vital for ensuring equal treatment, or parity, between different groups in downstream applications.  ...  We demonstrate an effective method for controlling parity through mutual information based on contrastive information estimators and show that they outperform approaches that rely on variational bounds  ...  We would like to thank Shobhit Jain for spotting an error in our Thm. 2.  ... 
arXiv:2101.04108v3 fatcat:adqnfpjhxjfxfnbdcv62agtyxa

On the Transferability of VAE Embeddings using Relational Knowledge with Semi-Supervision [article]

Harald Strömfelt, Luke Dickens, Artur d'Avila Garcez, Alessandra Russo
2020 arXiv   pre-print
We compare the relative benefits of relation-decoder complexity and latent space structure on both inductive and transductive transfer learning.  ...  We explore this trade-off by investigating the inductive and transductive transfer performance of two relation-decoders: the "Neural Tensor Network" (NTN), a powerful latent factor model (LFM) [12, 13  ...  We argue that this presents a trade-off between decoder complexity accuracy achievable via highly complex decoders and the value of a latent representation carries over to new data or tasks.  ... 
arXiv:2011.07137v1 fatcat:6sx2zpcnmvcvda2syoojj5v6ce

Navigating with Graph Representations for Fast and Scalable Decoding of Neural Language Models [article]

Minjia Zhang, Xiaodong Liu, Wenhan Wang, Jianfeng Gao, Yuxiong He
2018 arXiv   pre-print
We observe that, in decoding of many NLP tasks, only the probabilities of the top-K hypotheses need to be calculated preciously and K is often much smaller than the vocabulary size.  ...  We demonstrate that FGD reduces the decoding time by an order of magnitude while attaining close to the full softmax baseline accuracy on neural machine translation and language modeling tasks.  ...  The length of the queue is bounded by ef Search, a hyperparameter that controls the trade-off between search time and accuracy.  ... 
arXiv:1806.04189v1 fatcat:syrup2twcvf7dl32hduoafpub4

Burst Erasure Correction Codes With Low Decoding Delay

E. Martinian, C.-E.W. Sundberg
2004 IEEE Transactions on Information Theory  
Specifically, we show that the new encoders have the shortest possible decoding delay required to correct all bursts of a given length with a fixed redundancy.  ...  By comparing the new encoders to Maximum Distance Separable (MDS) codes, we show that the latter generally require either more redundancy or more delay to correct bursts of a given length.  ...  Note that the decoding delay versus burst length trade-off in Theorem 2 is optimal in the sense that it meets the bound in (1) with equality.  ... 
doi:10.1109/tit.2004.834844 fatcat:6chvfte5rfhpzpzab42cpn26pi

Center3D: Center-based Monocular 3D Object Detection with Joint Depth Understanding [article]

Yunlei Tang, Sebastian Dorn, Chiragkumar Savani
2020 arXiv   pre-print
Compared with state-of-the-art detectors, Center3D has achieved the best speed-accuracy trade-off in realtime monocular object detection.  ...  Evaluating on KITTI dataset for moderate objects, Center3D improved the AP in BEV from 29.7% to 42.8%, and the AP in 3D from 18.6% to 39.1%.  ...  Additionally, it achieved the best trade-off between speed and performances not only in 2D but also in 3D.  ... 
arXiv:2005.13423v1 fatcat:se3akh6fqvhd5iz6ks5jh262c4
« Previous Showing results 1 — 15 out of 3,359 results