Towards Mixed-Precision Quantization of Neural Networks via Constrained Optimization

• URL: http://arxiv.org/abs/2110.06554v1
• Date: Wed, 13 Oct 2021 08:09:26 GMT
• Title: Towards Mixed-Precision Quantization of Neural Networks via Constrained Optimization
• Authors: Weihan Chen, Peisong Wang, Jian Cheng
• Abstract summary: We present a principled framework to solve the mixed-precision quantization problem. We show that our method is derived in a principled way and much more computationally efficient.
• Score: 28.76708310896311
• License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/
• Abstract: Quantization is a widely used technique to compress and accelerate deep neural networks. However, conventional quantization methods use the same bit-width for all (or most of) the layers, which often suffer significant accuracy degradation in the ultra-low precision regime and ignore the fact that emergent hardware accelerators begin to support mixed-precision computation. Consequently, we present a novel and principled framework to solve the mixed-precision quantization problem in this paper. Briefly speaking, we first formulate the mixed-precision quantization as a discrete constrained optimization problem. Then, to make the optimization tractable, we approximate the objective function with second-order Taylor expansion and propose an efficient approach to compute its Hessian matrix. Finally, based on the above simplification, we show that the original problem can be reformulated as a Multiple-Choice Knapsack Problem (MCKP) and propose a greedy search algorithm to solve it efficiently. Compared with existing mixed-precision quantization works, our method is derived in a principled way and much more computationally efficient. Moreover, extensive experiments conducted on the ImageNet dataset and various kinds of network architectures also demonstrate its superiority over existing uniform and mixed-precision quantization approaches.

Related papers

• A Primal-dual algorithm for image reconstruction with ICNNs [3.4797100095791706]
  We address the optimization problem in a data-driven variational framework, where the regularizer is parameterized by an input- neural network (ICNN) While gradient-based methods are commonly used to solve such problems, they struggle to effectively handle nonsmoothness. We show that a proposed approach outperforms subgradient methods in terms of both speed and stability.
  arXiv  Detail & Related papers  (2024-10-16T10:36:29Z)
• A coherent approach to quantum-classical optimization [0.0]
  Hybrid quantum-classical optimization techniques have been shown to allow for the reduction of quantum computational resources. We identify the coherence entropy as a crucial metric in determining the suitability of quantum states. We propose a quantum-classical optimization protocol that significantly improves on previous approaches for such tasks.
  arXiv  Detail & Related papers  (2024-09-20T22:22:53Z)
• An Optimization-based Deep Equilibrium Model for Hyperspectral Image Deconvolution with Convergence Guarantees [71.57324258813675]
  We propose a novel methodology for addressing the hyperspectral image deconvolution problem. A new optimization problem is formulated, leveraging a learnable regularizer in the form of a neural network. The derived iterative solver is then expressed as a fixed-point calculation problem within the Deep Equilibrium framework.
  arXiv  Detail & Related papers  (2023-06-10T08:25:16Z)
• Linearization Algorithms for Fully Composite Optimization [61.20539085730636]
  This paper studies first-order algorithms for solving fully composite optimization problems convex compact sets. We leverage the structure of the objective by handling differentiable and non-differentiable separately, linearizing only the smooth parts.
  arXiv  Detail & Related papers  (2023-02-24T18:41:48Z)
• AskewSGD : An Annealed interval-constrained Optimisation method to train Quantized Neural Networks [12.229154524476405]
  We develop a new algorithm, Annealed Skewed SGD - AskewSGD - for training deep neural networks (DNNs) with quantized weights. Unlike algorithms with active sets and feasible directions, AskewSGD avoids projections or optimization under the entire feasible set. Experimental results show that the AskewSGD algorithm performs better than or on par with state of the art methods in classical benchmarks.
  arXiv  Detail & Related papers  (2022-11-07T18:13:44Z)
• Finite-Bit Quantization For Distributed Algorithms With Linear Convergence [6.293059137498172]
  We study distributed algorithms for (strongly convex) composite optimization problems over mesh networks subject to quantized communications. A new quantizer coupled with a communication-efficient encoding scheme is proposed, which efficiently implements the Biased Compression (BC-)rule. Numerical results validate our theoretical findings and show that distributed algorithms equipped with the proposed quantizer have more favorable communication complexity than algorithms using existing quantization rules.
  arXiv  Detail & Related papers  (2021-07-23T15:31:31Z)
• Variational Quantum Optimization with Multi-Basis Encodings [62.72309460291971]
  We introduce a new variational quantum algorithm that benefits from two innovations: multi-basis graph complexity and nonlinear activation functions. Our results in increased optimization performance, two increase in effective landscapes and a reduction in measurement progress.
  arXiv  Detail & Related papers  (2021-06-24T20:16:02Z)
• Quantized Proximal Averaging Network for Analysis Sparse Coding [23.080395291046408]
  We unfold an iterative algorithm into a trainable network that facilitates learning sparsity prior to quantization. We demonstrate applications to compressed image recovery and magnetic resonance image reconstruction.
  arXiv  Detail & Related papers  (2021-05-13T12:05:35Z)
• Adaptive pruning-based optimization of parameterized quantum circuits [62.997667081978825]
  Variisy hybrid quantum-classical algorithms are powerful tools to maximize the use of Noisy Intermediate Scale Quantum devices. We propose a strategy for such ansatze used in variational quantum algorithms, which we call "Efficient Circuit Training" (PECT) Instead of optimizing all of the ansatz parameters at once, PECT launches a sequence of variational algorithms.
  arXiv  Detail & Related papers  (2020-10-01T18:14:11Z)
• Cross Entropy Hyperparameter Optimization for Constrained Problem Hamiltonians Applied to QAOA [68.11912614360878]
  Hybrid quantum-classical algorithms such as Quantum Approximate Optimization Algorithm (QAOA) are considered as one of the most encouraging approaches for taking advantage of near-term quantum computers in practical applications. Such algorithms are usually implemented in a variational form, combining a classical optimization method with a quantum machine to find good solutions to an optimization problem. In this study we apply a Cross-Entropy method to shape this landscape, which allows the classical parameter to find better parameters more easily and hence results in an improved performance.
  arXiv  Detail & Related papers  (2020-03-11T13:52:41Z)
• Optimal Randomized First-Order Methods for Least-Squares Problems [56.05635751529922]
  This class of algorithms encompasses several randomized methods among the fastest solvers for least-squares problems. We focus on two classical embeddings, namely, Gaussian projections and subsampled Hadamard transforms. Our resulting algorithm yields the best complexity known for solving least-squares problems with no condition number dependence.
  arXiv  Detail & Related papers  (2020-02-21T17:45:32Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.