Filters








29,993 Hits in 3.2 sec

Specifications and Proofs for Ensemble Layers [chapter]

Jason Hickey, Nancy Lynch, Robbert van Renesse
1999 Lecture Notes in Computer Science  
In this paper we use I/O automata for formalizing, specifying, and verifying the Ensemble implementation.  ...  Ensemble is a widely used group communication system that supports distributed programming by providing precise guarantees for synchronization, message ordering, and message delivery.  ...  This document gives the specifications and summarizes the proofs for the total order case study.  ... 
doi:10.1007/3-540-49059-0_9 fatcat:tzfppcmwsngjbbgkacawblojja

A proof environment for the development of group communication systems [chapter]

Christoph Kreitz, Mark Hayden, Jason Hickey
1998 Lecture Notes in Computer Science  
We discuss techniques for reasoning about critical properties of Ensemble as well as verified strategies for reconfiguring the Ensemble system in order to improve its performance in concrete applications  ...  Our approach makes methods of automated deduction applicable to the implementation of real-world systems by linking the Ensemble group communication toolkit to the NuPRL proof development system.  ...  In Section 4 we discuss techniques for verifying system properties and in Section 5 we describe proof and rewrite tactics for a verified reconfiguration of Ensemble in a given application-specific context  ... 
doi:10.1007/bfb0054269 fatcat:zep3db24kneinptry57bfor5v4

Building reliable, high-performance communication systems from components

Xiaoming Liu, Christoph Kreitz, Robbert van Renesse, Jason Hickey, Mark Hayden, Kenneth Birman, Robert Constable
2000 ACM SIGOPS Operating Systems Review  
Our paper answers these questions for the Ensemble communication architecture by showing how, with help of the Nuprl formal system, configurations may be checked against specifications, and how optimized  ...  The performance results show that we can substantially reduce end-to-end latency in the already optimized Ensemble system.  ...  Acknowledgments We would like to thank Mark Bickford, Alan Fekete, and Nancy Lynch for very helpful discussions.  ... 
doi:10.1145/346152.346174 fatcat:gpommc4wvzhb3p5zetnpdzt7zu

Building reliable, high-performance communication systems from components

Xiaoming Liu, Christoph Kreitz, Robbert van Renesse, Jason Hickey, Mark Hayden, Kenneth Birman, Robert Constable
1999 ACM SIGOPS Operating Systems Review  
Our paper answers these questions for the Ensemble communication architecture by showing how, with help of the Nuprl formal system, configurations may be checked against specifications, and how optimized  ...  The performance results show that we can substantially reduce end-to-end latency in the already optimized Ensemble system.  ...  Acknowledgments We would like to thank Mark Bickford, Alan Fekete, and Nancy Lynch for very helpful discussions.  ... 
doi:10.1145/319344.319157 fatcat:bkrsciavmvfwtdavdkaojuyxny

Building reliable, high-performance communication systems from components

Xiaoming Liu, Christoph Kreitz, Robbert van Renesse, Jason Hickey, Mark Hayden, Kenneth Birman, Robert Constable
1999 Proceedings of the seventeenth ACM symposium on Operating systems principles - SOSP '99  
Our paper answers these questions for the Ensemble communication architecture by showing how, with help of the Nuprl formal system, configurations may be checked against specifications, and how optimized  ...  The performance results show that we can substantially reduce end-to-end latency in the already optimized Ensemble system.  ...  Acknowledgments We would like to thank Mark Bickford, Alan Fekete, and Nancy Lynch for very helpful discussions.  ... 
doi:10.1145/319151.319157 dblp:conf/sosp/LiuKRHHBC99 fatcat:rca6mecrqvhtrk5cbgm7nmq5ea

Ensemble nonequivalence in random graphs with modular structure

Diego Garlaschelli, Frank den Hollander, Andrea Roccaverde
2016 Journal of Physics A: Mathematical and Theoretical  
In addition, we derive a formula for the specific relative entropy and provide an interpretation of this formula in terms of Poissonisation of the degrees.  ...  In the present paper we consider an arbitrary number of intra-connected and inter-connected layers, thus allowing for modular graphs with a multi-partite, multiplex, block-model or community structure.  ...  DG and AR are supported by EU-project 317532-MULTIPLEX. FdH and AR are supported by NWO Gravitation Grant 024.002.003-NETWORKS.  ... 
doi:10.1088/1751-8113/50/1/015001 fatcat:uxcc3puhiba6zpg2mdmpg44hey

Asymmetric Sigmoidal Activation Function for Feed-Forward Artificial Neural Networks

2019 VOLUME-8 ISSUE-10, AUGUST 2019, REGULAR ISSUE  
The networks using the proposed activation function are compared against those using the generally used logistic and the hyperbolic tangent activation function for the solution of 12 function approximation  ...  The requirement imposed on these networks is to have non-linear functions of a specific kind at the hidden nodes of the network.  ...  And, the tests are conducted for the ensemble errors over both the training data set and the test data set. IV.  ... 
doi:10.35940/ijitee.l3310.1081219 fatcat:h6nxl22x6jezjhtjol6y7ij544

The Loss Surface of Residual Networks: Ensembles and the Role of Batch Normalization [article]

Etai Littwin, Lior Wolf
2016 arXiv   pre-print
We show that these ensembles are dynamic: while initially the virtual ensemble is mostly at depths lower than half the network's depth, as training progresses, it becomes deeper and deeper.  ...  We explain this behavior and demonstrate the driving force behind it.  ...  Proof of Lemma 4. For simplicity, we ignore the constants in the binomial coefficient, and assume r = 1 z p r β r .  ... 
arXiv:1611.02525v1 fatcat:7w5mm5qybzalnekdu4d5cmyo6i

One-Shot Neural Ensemble Architecture Search by Diversity-Guided Search Space Shrinking [article]

Minghao Chen, Houwen Peng, Jianlong Fu, Haibin Ling
2021 arXiv   pre-print
Searching for ensembles is non-trivial and has two key challenges: enlarged search space and potentially more complexity for the searched model.  ...  For the second challenge, we enable a new search dimension to learn layer sharing among different models for efficiency purposes.  ...  We refer to Appendix A for a proof. Diversity-Guided Search Space Shrinking.  ... 
arXiv:2104.00597v2 fatcat:4jlb7fxan5ba3dt2dfsvoysib4

Design of Bilayer and Multi-layer LDPC Ensembles from Individual Degree Distributions [article]

Eshed Ram, Yuval Cassuto
2020 arXiv   pre-print
The ensembles are defined through individual uni-variate degree distributions, one for each layer.  ...  A new approach for designing bilayer and multi-layer LDPC codes is proposed and studied in the asymptotic regime.  ...  This completes the proof for the P 0 > 0 and λ (2) (0) = 0 case.  ... 
arXiv:2010.14758v1 fatcat:4qdo7s23n5e2je3ast3qb4qfme

On Deep Ensemble Learning from a Function Approximation Perspective [article]

Jiawei Zhang and Limeng Cui and Fisher B. Gouza
2018 arXiv   pre-print
According to the theoretic proof provided in this paper, given the input feature space of dimension d, the required unit model number will be 2d, if the ensemble model involves one single layer.  ...  In the case when the unit model mathematical mappings are bounded, sigmoidal and discriminatory, we demonstrate that the deep ensemble learning framework can achieve a universal approximation of any functions  ...  The specific number of layers involved is highly dependent on the shape of binary tree about the deep ensemble learning model.  ... 
arXiv:1805.07502v1 fatcat:5favykt7xjajtgj537xnngpchq

Exploiting Partial Order of Keys to Verify Security of a Vehicular Group Protocol [article]

Felipe Boeira, Mikael Asplund
2021 arXiv   pre-print
As for other cyber-physical systems, security is essential to the deployment of these applications and standardisation efforts are ongoing.  ...  We tackle the complexity of the resulting model with a proof strategy based on a relation on keys.  ...  Figure 1 represents the protocol stack and the associated standards for each layer.  ... 
arXiv:2105.02664v2 fatcat:st2ykgpkanctvp2rcvwiec6yum

Automated Fast-Track Reconfiguration of Group Communication Systems [chapter]

Christoph Kreitz
1999 Lecture Notes in Computer Science  
Our techniques are implemented by tactics and theorems of the NuPRL proof development system and have been used successfully for the reconfiguration of application systems built with the Ensemble group  ...  For common sequences of operations we identify a fast-path through a stack of communication protocols and reconfigure the system's code accordingly.  ...  Figure 5 presents the compression and expansion theorems for Ensemble.  ... 
doi:10.1007/3-540-49059-0_8 fatcat:xycqnuks2bfnpjupxvzaajbejy

Ensemble Robustness and Generalization of Stochastic Deep Learning Algorithms [article]

Tom Zahavy, Bingyi Kang, Alex Sivak, Jiashi Feng, Huan Xu, Shie Mannor
2017 arXiv   pre-print
To support our claims, we provide extensive simulations for different deep learning algorithms and different network architectures exhibiting a strong correlation between ensemble robustness and the ability  ...  ., Stochastic Gradient Descent, Dropout, and Bayes-by-backprop), we revisit the robustness arguments of Xu & Mannor, and introduce a new approach, ensemble robustness, that concerns the robustness of a  ...  For a deep neural network consisting of L layers, the random variable r l is the dropout randomness for the l-th layer.  ... 
arXiv:1602.02389v4 fatcat:tjc56pku4zfrjltd6lakr6g4x4

REx: Data-Free Residual Quantization Error Expansion [article]

Edouard Yvinec and Arnaud Dapgony and Matthieu Cord and Kevin Bailly
2022 arXiv   pre-print
In this paper, we propose REx, a data-free quantization algorithm for pre-trained models that is compliant with data protection regulations, convenient and fast to execute.  ...  Third, we show that this sparse expansion can be approximated by an ensemble of quantized neural networks to dramatically improve the evaluation speed through more efficient parallelization.  ...  B.4 Proof of Lemma 4: Ensemble Expansion Ensemble of two Layers DNNs Let f be a feed-forward DNN with two layers f 1 ,f 2 and σ a piece-wise affine activation function (e.g. ReLU).  ... 
arXiv:2203.14645v1 fatcat:zmaube2rqfckfh5vpyb3xpzxhi
« Previous Showing results 1 — 15 out of 29,993 results