3 Search Results for "Fu, Hu"

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}
}
Document
Track A: Algorithms, Complexity and Games
DOI: 10.4230/LIPIcs.ICALP.2021.68
Random Order Vertex Arrival Contention Resolution Schemes for Matching, with Applications

Authors: Hu Fu, Zhihao Gavin Tang, Hongxun Wu, Jinzhao Wu, and Qianfan Zhang

Published in: LIPIcs, Volume 198, 48th International Colloquium on Automata, Languages, and Programming (ICALP 2021)

Abstract
With a wide range of applications, stochastic matching problems have been studied in different models, including prophet inequality, Query-Commit, and Price-of-Information. While there have been recent breakthroughs in all these settings for bipartite graphs, few non-trivial results are known for general graphs. In this paper, we study the random order vertex arrival contention resolution scheme for matching in general graphs, which is inspired by the recent work of Ezra et al. (EC 2020). We design an 8/15-selectable batched RCRS for matching and apply it to achieve 8/15-competitive/approximate algorithms for all the three models. Our results are the first non-trivial results for random order prophet matching and Price-of-Information matching in general graphs. For the Query-Commit model, our result substantially improves upon the 0.501 approximation ratio by Tang et al. (STOC 2020). We also show that no batched RCRS for matching can be better than 1/2+1/(2e²) ≈ 0.567-selectable.
Cite as

Hu Fu, Zhihao Gavin Tang, Hongxun Wu, Jinzhao Wu, and Qianfan Zhang. Random Order Vertex Arrival Contention Resolution Schemes for Matching, with Applications. In 48th International Colloquium on Automata, Languages, and Programming (ICALP 2021). Leibniz International Proceedings in Informatics (LIPIcs), Volume 198, pp. 68:1-68:20, Schloss Dagstuhl - Leibniz-Zentrum für Informatik (2021)

Copy BibTex To Clipboard

@InProceedings{fu_et_al:LIPIcs.ICALP.2021.68,
  author =	{Fu, Hu and Tang, Zhihao Gavin and Wu, Hongxun and Wu, Jinzhao and Zhang, Qianfan},
  title =	{{Random Order Vertex Arrival Contention Resolution Schemes for Matching, with Applications}},
  booktitle =	{48th International Colloquium on Automata, Languages, and Programming (ICALP 2021)},
  pages =	{68:1--68:20},
  series =	{Leibniz International Proceedings in Informatics (LIPIcs)},
  ISBN =	{978-3-95977-195-5},
  ISSN =	{1868-8969},
  year =	{2021},
  volume =	{198},
  editor =	{Bansal, Nikhil and Merelli, Emanuela and Worrell, James},
  publisher =	{Schloss Dagstuhl -- Leibniz-Zentrum f{\"u}r Informatik},
  address =	{Dagstuhl, Germany},
  URL =		{https://drops.dagstuhl.de/entities/document/10.4230/LIPIcs.ICALP.2021.68},
  URN =		{urn:nbn:de:0030-drops-141376},
  doi =		{10.4230/LIPIcs.ICALP.2021.68},
  annote =	{Keywords: Matching, Contention Resolution Scheme, Price of Information, Query-Commit}
}
Document
DOI: 10.4230/LIPIcs.APPROX-RANDOM.2014.669
Improved Lower Bounds for Testing Triangle-freeness in Boolean Functions via Fast Matrix Multiplication

Authors: Hu Fu and Robert Kleinberg

Published in: LIPIcs, Volume 28, Approximation, Randomization, and Combinatorial Optimization. Algorithms and Techniques (APPROX/RANDOM 2014)

Abstract
Understanding the query complexity for testing linear-invariant properties has been a central open problem in the study of algebraic property testing. Triangle-freeness in Boolean functions is a simple property whose testing complexity is unknown. Three Boolean functions f1, f2 and f3, mapping {0,1}^k to {0,1}, are said to be triangle free if there is no x, y in {0,1}^k such that f1(x) = f2(y) = f3(x + y) = 1. This property is known to be strongly testable (Green 2005), but the number of queries needed is upper-bounded only by a tower of twos whose height is polynomial in 1 / epsislon, where epsislon is the distance between the tested function triple and triangle-freeness, i.e., the minimum fraction of function values that need to be modified to make the triple triangle free. A lower bound of (1 / epsilon)^2.423 for any one-sided tester was given by Bhattacharyya and Xie (2010). In this work we improve this bound to (1 / epsilon)^6.619. Interestingly, we prove this by way of a combinatorial construction called uniquely solvable puzzles that was at the heart of Coppersmith and Winograd's renowned matrix multiplication algorithm.
Cite as

Hu Fu and Robert Kleinberg. Improved Lower Bounds for Testing Triangle-freeness in Boolean Functions via Fast Matrix Multiplication. In Approximation, Randomization, and Combinatorial Optimization. Algorithms and Techniques (APPROX/RANDOM 2014). Leibniz International Proceedings in Informatics (LIPIcs), Volume 28, pp. 669-676, Schloss Dagstuhl - Leibniz-Zentrum für Informatik (2014)

Copy BibTex To Clipboard

@InProceedings{fu_et_al:LIPIcs.APPROX-RANDOM.2014.669,
  author =	{Fu, Hu and Kleinberg, Robert},
  title =	{{Improved Lower Bounds for Testing Triangle-freeness in Boolean Functions via Fast Matrix Multiplication}},
  booktitle =	{Approximation, Randomization, and Combinatorial Optimization. Algorithms and Techniques (APPROX/RANDOM 2014)},
  pages =	{669--676},
  series =	{Leibniz International Proceedings in Informatics (LIPIcs)},
  ISBN =	{978-3-939897-74-3},
  ISSN =	{1868-8969},
  year =	{2014},
  volume =	{28},
  editor =	{Jansen, Klaus and Rolim, Jos\'{e} and Devanur, Nikhil R. and Moore, Cristopher},
  publisher =	{Schloss Dagstuhl -- Leibniz-Zentrum f{\"u}r Informatik},
  address =	{Dagstuhl, Germany},
  URL =		{https://drops.dagstuhl.de/entities/document/10.4230/LIPIcs.APPROX-RANDOM.2014.669},
  URN =		{urn:nbn:de:0030-drops-47304},
  doi =		{10.4230/LIPIcs.APPROX-RANDOM.2014.669},
  annote =	{Keywords: Property testing, linear invariance, fast matrix multiplication, uniquely solvable puzzles}
}
  • Refine by Author
  • 2 Fu, Hu
  • 1 Camalleri, Manfredi
  • 1 Fasfous, Nael
  • 1 Frickenstein, Alexander
  • 1 Kleinberg, Robert
  • Show More...

  • Refine by Classification
  • 1 Computing methodologies → Artificial intelligence
  • 1 Theory of computation → Online algorithms

  • Refine by Keyword
  • 1 Contention Resolution Scheme
  • 1 Convolutional Neural Networks
  • 1 Hardware Modeling
  • 1 Matching
  • 1 Optimization
  • Show More...

  • Refine by Type
  • 3 document

  • Refine by Publication Year
  • 1 2014
  • 1 2021
  • 1 2022

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