2 Search Results for "Qi, Wei"

Document
DOI: 10.4230/LIPIcs.ITC.2023.15
Online Mergers and Applications to Registration-Based Encryption and Accumulators

Authors: Mohammad Mahmoody and Wei Qi

Published in: LIPIcs, Volume 267, 4th Conference on Information-Theoretic Cryptography (ITC 2023)

Abstract
In this work we study a new information theoretic problem, called online merging, that has direct applications for constructing public-state accumulators and registration-based encryption schemes. An {online merger} receives the sequence of sets {1}, {2}, … in an online way, and right after receiving {i}, it can re-partition the elements 1,…,i into T₁,…,T_{m_i} by merging some of these sets. The goal of the merger is to balance the trade-off between the maximum number of sets wid = max_{i ∈ [n]} m_i that co-exist at any moment, called the width of the scheme, with its depth dep = max_{i ∈ [n]} d_i, where d_i is the number of times that the sets that contain i get merged. An online merger can be used to maintain a set of Merkle trees that occasionally get merged. An online merger can be directly used to obtain public-state accumulators (using collision-resistant hashing) and registration-based encryptions (relying on more assumptions). Doing so, the width of an online merger translates into the size of the public-parameter of the constructed scheme, and the depth of the online algorithm corresponds to the number of times that parties need to update their "witness" (for accumulators) or their decryption key (for RBE). In this work, we construct online mergers with poly(log n) width and O(log n / log log n) depth, which can be shown to be optimal for all schemes with poly(log n) width. More generally, we show how to achieve optimal depth for a given fixed width and to achieve a 2-approximate optimal width for a given depth d that can possibly grow as a function of n (e.g., d = 2 or d = log n / log log n). As applications, we obtain accumulators with O(log n / log log n) number of updates for parties' witnesses (which can be shown to be optimal for accumulator digests of length poly(log n)) as well as registration based encryptions that again have an optimal O(log n / log log n) number of decryption updates, resolving the open question of Mahmoody, Rahimi, Qi [TCC'22] who proved that Ω(log n / log log n) number of decryption updates are necessary for any RBE (with public parameter of length poly(log n)). More generally, for any given number of decryption updates d = d(n) (under believable computational assumptions) our online merger implies RBE schemes with public parameters of length that is optimal, up to a constant factor that depends on the security parameter. For example, for any constant number of updates d, we get RBE schemes with public parameters of length O(n^{1/(d+1)}).
Cite as

Mohammad Mahmoody and Wei Qi. Online Mergers and Applications to Registration-Based Encryption and Accumulators. In 4th Conference on Information-Theoretic Cryptography (ITC 2023). Leibniz International Proceedings in Informatics (LIPIcs), Volume 267, pp. 15:1-15:23, Schloss Dagstuhl - Leibniz-Zentrum für Informatik (2023)

Copy BibTex To Clipboard

@InProceedings{mahmoody_et_al:LIPIcs.ITC.2023.15,
  author =	{Mahmoody, Mohammad and Qi, Wei},
  title =	{{Online Mergers and Applications to Registration-Based Encryption and Accumulators}},
  booktitle =	{4th Conference on Information-Theoretic Cryptography (ITC 2023)},
  pages =	{15:1--15:23},
  series =	{Leibniz International Proceedings in Informatics (LIPIcs)},
  ISBN =	{978-3-95977-271-6},
  ISSN =	{1868-8969},
  year =	{2023},
  volume =	{267},
  editor =	{Chung, Kai-Min},
  publisher =	{Schloss Dagstuhl -- Leibniz-Zentrum f{\"u}r Informatik},
  address =	{Dagstuhl, Germany},
  URL =		{https://drops.dagstuhl.de/entities/document/10.4230/LIPIcs.ITC.2023.15},
  URN =		{urn:nbn:de:0030-drops-183432},
  doi =		{10.4230/LIPIcs.ITC.2023.15},
  annote =	{Keywords: Registration-based encryption, Accumulators, Merkle Trees}
}
Document
DOI: 10.4230/LITES.8.1.3
HW-Flow: A Multi-Abstraction Level HW-CNN Codesign Pruning Methodology

Authors: Manoj-Rohit Vemparala, Nael Fasfous, Alexander Frickenstein, Emanuele Valpreda, Manfredi Camalleri, Qi Zhao, Christian Unger, Naveen-Shankar Nagaraja, Maurizio Martina, and Walter Stechele

Published in: LITES, Volume 8, Issue 1 (2022): Special Issue on Embedded Systems for Computer Vision. Leibniz Transactions on Embedded Systems, Volume 8, Issue 1

Abstract
Convolutional neural networks (CNNs) have produced unprecedented accuracy for many computer vision problems in the recent past. In power and compute-constrained embedded platforms, deploying modern CNNs can present many challenges. Most CNN architectures do not run in real-time due to the high number of computational operations involved during the inference phase. This emphasizes the role of CNN optimization techniques in early design space exploration. To estimate their efficacy in satisfying the target constraints, existing techniques are either hardware (HW) agnostic, pseudo-HW-aware by considering parameter and operation counts, or HW-aware through inflexible hardware-in-the-loop (HIL) setups. In this work, we introduce HW-Flow, a framework for optimizing and exploring CNN models based on three levels of hardware abstraction: Coarse, Mid and Fine. Through these levels, CNN design and optimization can be iteratively refined towards efficient execution on the target hardware platform. We present HW-Flow in the context of CNN pruning by augmenting a reinforcement learning agent with key metrics to understand the influence of its pruning actions on the inference hardware. With 2× reduction in energy and latency, we prune ResNet56, ResNet50, and DeepLabv3 with minimal accuracy degradation on the CIFAR-10, ImageNet, and CityScapes datasets, respectively.
Cite as

Manoj-Rohit Vemparala, Nael Fasfous, Alexander Frickenstein, Emanuele Valpreda, Manfredi Camalleri, Qi Zhao, Christian Unger, Naveen-Shankar Nagaraja, Maurizio Martina, and Walter Stechele. HW-Flow: A Multi-Abstraction Level HW-CNN Codesign Pruning Methodology. In LITES, Volume 8, Issue 1 (2022): Special Issue on Embedded Systems for Computer Vision. Leibniz Transactions on Embedded Systems, Volume 8, Issue 1, pp. 03:1-03:30, Schloss Dagstuhl - Leibniz-Zentrum für Informatik (2022)

Copy BibTex To Clipboard

@Article{vemparala_et_al:LITES.8.1.3,
  author =	{Vemparala, Manoj-Rohit and Fasfous, Nael and Frickenstein, Alexander and Valpreda, Emanuele and Camalleri, Manfredi and Zhao, Qi and Unger, Christian and Nagaraja, Naveen-Shankar and Martina, Maurizio and Stechele, Walter},
  title =	{{HW-Flow: A Multi-Abstraction Level HW-CNN Codesign Pruning Methodology}},
  booktitle =	{LITES, Volume 8, Issue 1 (2022): Special Issue on Embedded Systems for Computer Vision},
  pages =	{03:1--03:30},
  journal =	{Leibniz Transactions on Embedded Systems},
  ISSN =	{2199-2002},
  year =	{2022},
  volume =	{8},
  number =	{1},
  editor =	{Vemparala, Manoj-Rohit and Fasfous, Nael and Frickenstein, Alexander and Valpreda, Emanuele and Camalleri, Manfredi and Zhao, Qi and Unger, Christian and Nagaraja, Naveen-Shankar and Martina, Maurizio and Stechele, Walter},
  publisher =	{Schloss Dagstuhl -- Leibniz-Zentrum f{\"u}r Informatik},
  address =	{Dagstuhl, Germany},
  URL =		{https://drops.dagstuhl.de/entities/document/10.4230/LITES.8.1.3},
  doi =		{10.4230/LITES.8.1.3},
  annote =	{Keywords: Convolutional Neural Networks, Optimization, Hardware Modeling, Pruning}
}
  • Refine by Author
  • 1 Camalleri, Manfredi
  • 1 Fasfous, Nael
  • 1 Frickenstein, Alexander
  • 1 Mahmoody, Mohammad
  • 1 Martina, Maurizio
  • Show More...

  • Refine by Classification
  • 1 Computing methodologies → Artificial intelligence
  • 1 Theory of computation → Computational complexity and cryptography

  • Refine by Keyword
  • 1 Accumulators
  • 1 Convolutional Neural Networks
  • 1 Hardware Modeling
  • 1 Merkle Trees
  • 1 Optimization
  • Show More...

  • Refine by Type
  • 2 document

  • Refine by Publication Year
  • 1 2022
  • 1 2023

Questions / Remarks / Feedback
X

Feedback for Dagstuhl Publishing


Thanks for your feedback!

Feedback submitted

Could not send message

Please try again later or send an E-mail
© 2023 Schloss Dagstuhl - LZI GmbH Imprint Privacy Contact