skip to main content


Title: Get More at Once: Alternating Sparse Training with Gradient Correction
Recently, a new trend of exploring training sparsity has emerged, which remove parameters during training, leading to both training and inference efficiency improvement. This line of works primarily aims to obtain a single sparse model under a pre-defined large sparsity ratio. It leads to a static/fixed sparse inference model that is not capable of adjusting or re-configuring its computation complexity (i.e., inference structure, latency) after training for real-world varying and dynamic hardware resource availability. To enable such run-time or post-training network morphing, the concept of `dynamic inference' or `training-once-for-all' has been proposed to train a single network consisting of multiple sub-nets once, but each sub-net could perform the same inference function with different computing complexity. However, the traditional dynamic inference training method requires a joint training scheme with multi-objective optimization, which suffers from very large training overhead. In this work, for the first time, we propose a novel alternating sparse training (AST) scheme to train multiple sparse sub-nets for dynamic inference without extra training cost compared to the case of training a single sparse model from scratch. Furthermore, to mitigate the interference of weight update among sub-nets, we propose gradient correction within the inner-group iterations to reduce their weight update interference. We validate the proposed AST on multiple datasets against state-of-the-art sparse training method, which shows that AST achieves similar or better accuracy, but only needs to train once to get multiple sparse sub-nets with different sparsity ratios. More importantly, compared with the traditional joint training based dynamic inference training methodology, the large training overhead is completely eliminated without affecting the accuracy of each sub-net.  more » « less
Award ID(s):
1931871 2144751
PAR ID:
10389140
Author(s) / Creator(s):
; ; ;
Date Published:
Journal Name:
Thirty-Sixth Conference on Neural Information Processing Systems
Format(s):
Medium: X
Sponsoring Org:
National Science Foundation
More Like this
  1. null (Ed.)
    To deploy powerful deep neural network (DNN) into smart, but resource limited IoT devices, many prior works have been proposed to compress DNN to reduce the network size and computation complexity with negligible accuracy degradation, such as weight quantization, network pruning, convolution decomposition, etc. However, by utilizing conventional DNN compression methods, a smaller, but fixed, network is generated from a relative large background model to achieve resource limited hardware acceleration. However, such optimization lacks the ability to adjust its structure in real-time to adapt for a dynamic computing hardware resource allocation and workloads. In this paper, we mainly review our two prior works [13], [15] to tackle this challenge, discussing how to construct a dynamic DNN by means of either uniform or non-uniform sub-nets generation methods. Moreover, to generate multiple non-uniform sub-nets, [15] needs to fully retrain the background model for each sub-net individually, named as multi-path method. To reduce the training cost, in this work, we further propose a single-path sub-nets generation method that can sample multiple sub-nets in different epochs within one training round. The constructed dynamic DNN, consisting of multiple sub-nets, provides the ability to run-time trade-off the inference accuracy and latency according to hardware resources and environment requirements. In the end, we study the the dynamic DNNs with different sub-nets generation methods on both CIFAR-10 and ImageNet dataset. We also present the run-time tuning of accuracy and latency on both GPU and CPU. 
    more » « less
  2. This article investigates a robust receiver scheme for a single carrier, multiple-input–multiple-output (MIMO) underwater acoustic (UWA) communications, which uses the sparse Bayesian learning algorithm for iterative channel estimation embedded in Turbo equalization (TEQ). We derive a block-wise sparse Bayesian learning framework modeling the spatial correlation of the MIMO UWA channels, where a more robust expectation–maximization algorithm is proposed for updating the joint estimates of channel impulse response, residual noise, and channel covariance matrix. By exploiting the spatially correlated sparsity of MIMO UWA channels and the second-order a priori channel statistics from the training sequence, the proposed Bayesian channel estimator enjoys not only relatively low complexity but also more stable control of the hyperparameters that determine the channel sparsity and recovery accuracy. Moreover, this article proposes a low complexity space-time soft decision feedback equalizer (ST-SDFE) with successive soft interference cancellation. Evaluated by the undersea 2008 Surface Processes and Acoustic Communications Experiment, the improved sparse Bayesian learning channel estimation algorithm outperforms the conventional Bayesian algorithms in terms of the robustness and complexity, while enjoying better estimation accuracy than the orthogonal matching pursuit and the improved proportionate normalized least mean squares algorithms. We have also verified that the proposed ST-SDFE TEQ significantly outperforms the low-complexity minimum mean square error TEQ in terms of the bit error rate and error propagation. 
    more » « less
  3. It is appealing but challenging to achieve real-time deep neural network (DNN) inference on mobile devices because even the powerful modern mobile devices are considered “resource-constrained” when executing large-scale DNNs. It necessitates the sparse model inference via weight pruning, i.e., DNN weight sparsity, and it is desirable to design a new DNN weight sparsity scheme that can facilitate real-time inference on mobile devices while preserving a high sparse model accuracy. This paper designs a novel mobile inference acceleration framework GRIM that is General to both convolutional neural networks (CNNs) and recurrent neural networks (RNNs) and that achieves Real-time execution and high accuracy, leveraging fine-grained structured sparse model Inference and compiler optimizations for Mobiles. We start by proposing a new fine-grained structured sparsity scheme through the Block-based Column-Row (BCR) pruning. Based on this new fine-grained structured sparsity, our GRIM framework consists of two parts: (a) the compiler optimization and code generation for real-time mobile inference; and (b) the BCR pruning optimizations for determining pruning hyperparameters and performing weight pruning. We compare GRIM with Alibaba MNN, TVM, TensorFlow-Lite, a sparse implementation based on CSR, PatDNN, and ESE (a representative FPGA inference acceleration framework for RNNs), and achieve up to 14.08× speedup. 
    more » « less
  4. Massive multiple-input multiple-output (MIMO) enjoys great advantage in 5G wireless communication systems owing to its spectrum and energy efficiency. However, hundreds of antennas require large volumes of pilot overhead to guarantee reliable channel estimation in FDD massive MIMO system. Compressive sensing (CS) has been applied for channel estimation by exploiting the inherent sparse structure of massive MIMO channel but suffer from high complexity. To overcome this challenge, this paper develops a hybrid channel estimation scheme by integrating the model-driven CS and data-driven deep unrolling technique. The proposed scheme consists of a coarse estimation part and a fine correction part to respectively exploit the inter- and intraframe sparsities of channels to greatly reduce the pilot overhead. Theoretical result is provided to indicate the convergence of the fine correction and coarse estimation net. Simulation results are provided to verify that our scheme can estimate MIMO channels with low pilot overhead while guaranteeing estimation accuracy with relatively low complexity. 
    more » « less
  5. Neural networks tend to achieve better accuracy with training if they are larger — even if the resulting models are overparameterized. Nevertheless, carefully removing such excess of parameters before, during, or after training may also produce models with similar or even improved accuracy. In many cases, that can be curiously achieved by heuristics as simple as removing a percentage of the weights with the smallest absolute value — even though absolute value is not a perfect proxy for weight relevance. With the premise that obtaining significantly better performance from pruning depends on accounting for the combined effect of removing multiple weights, we revisit one of the classic approaches for impact-based pruning: the Optimal Brain Surgeon (OBS). We propose a tractable heuristic for solving the combinatorial extension of OBS, in which we select weights for simultaneous removal, and we combine it with a single-pass systematic update of unpruned weights. Our selection method outperforms other methods for high sparsity, and the single-pass weight update is also advantageous if applied after those methods. 
    more » « less