ترغب بنشر مسار تعليمي؟ اضغط هنا

Sparse Training via Boosting Pruning Plasticity with Neuroregeneration

92   0   0.0 ( 0 )
 نشر من قبل Shiwei Liu
 تاريخ النشر 2021
  مجال البحث الهندسة المعلوماتية
والبحث باللغة English




اسأل ChatGPT حول البحث

Works on lottery ticket hypothesis (LTH) and single-shot network pruning (SNIP) have raised a lot of attention currently on post-training pruning (iterative magnitude pruning), and before-training pruning (pruning at initialization). The former method suffers from an extremely large computation cost and the latter category of methods usually struggles with insufficient performance. In comparison, during-training pruning, a class of pruning methods that simultaneously enjoys the training/inference efficiency and the comparable performance, temporarily, has been less explored. To better understand during-training pruning, we quantitatively study the effect of pruning throughout training from the perspective of pruning plasticity (the ability of the pruned networks to recover the original performance). Pruning plasticity can help explain several other empirical observations about neural network pruning in literature. We further find that pruning plasticity can be substantially improved by injecting a brain-inspired mechanism called neuroregeneration, i.e., to regenerate the same number of connections as pruned. Based on the insights from pruning plasticity, we design a novel gradual magnitude pruning (GMP) method, named gradual pruning with zero-cost neuroregeneration (GraNet), and its dynamic sparse training (DST) variant (GraNet-ST). Both of them advance state of the art. Perhaps most impressively, the latter for the first time boosts the sparse-to-sparse training performance over various dense-to-sparse methods by a large margin with ResNet-50 on ImageNet. We will release all codes.



قيم البحث

اقرأ أيضاً

Sparsification is an efficient approach to accelerate CNN inference, but it is challenging to take advantage of sparsity in training procedure because the involved gradients are dynamically changed. Actually, an important observation shows that most of the activation gradients in back-propagation are very close to zero and only have a tiny impact on weight-updating. Hence, we consider pruning these very small gradients randomly to accelerate CNN training according to the statistical distribution of activation gradients. Meanwhile, we theoretically analyze the impact of pruning algorithm on the convergence. The proposed approach is evaluated on AlexNet and ResNet-{18, 34, 50, 101, 152} with CIFAR-{10, 100} and ImageNet datasets. Experimental results show that our training approach could substantially achieve up to $5.92 times$ speedups at back-propagation stage with negligible accuracy loss.
Adversarial training (AT) is one of the most effective defenses against adversarial attacks for deep learning models. In this work, we advocate incorporating the hypersphere embedding (HE) mechanism into the AT procedure by regularizing the features onto compact manifolds, which constitutes a lightweight yet effective module to blend in the strength of representation learning. Our extensive analyses reveal that AT and HE are well coupled to benefit the robustness of the adversarially trained models from several aspects. We validate the effectiveness and adaptability of HE by embedding it into the popular AT frameworks including PGD-AT, ALP, and TRADES, as well as the FreeAT and FastAT strategies. In the experiments, we evaluate our methods under a wide range of adversarial attacks on the CIFAR-10 and ImageNet datasets, which verifies that integrating HE can consistently enhance the model robustness for each AT framework with little extra computation.
Learning with noisy labels is an important and challenging task for training accurate deep neural networks. Some commonly-used loss functions, such as Cross Entropy (CE), suffer from severe overfitting to noisy labels. Robust loss functions that sati sfy the symmetric condition were tailored to remedy this problem, which however encounter the underfitting effect. In this paper, we theoretically prove that textbf{any loss can be made robust to noisy labels} by restricting the network output to the set of permutations over a fixed vector. When the fixed vector is one-hot, we only need to constrain the output to be one-hot, which however produces zero gradients almost everywhere and thus makes gradient-based optimization difficult. In this work, we introduce the sparse regularization strategy to approximate the one-hot constraint, which is composed of network output sharpening operation that enforces the output distribution of a network to be sharp and the $ell_p$-norm ($ple 1$) regularization that promotes the network output to be sparse. This simple approach guarantees the robustness of arbitrary loss functions while not hindering the fitting ability. Experimental results demonstrate that our method can significantly improve the performance of commonly-used loss functions in the presence of noisy labels and class imbalance, and outperform the state-of-the-art methods. The code is available at https://github.com/hitcszx/lnl_sr.
Deep Neural Network (DNN) is powerful but computationally expensive and memory intensive, thus impeding its practical usage on resource-constrained front-end devices. DNN pruning is an approach for deep model compression, which aims at eliminating so me parameters with tolerable performance degradation. In this paper, we propose a novel momentum-SGD-based optimization method to reduce the network complexity by on-the-fly pruning. Concretely, given a global compression ratio, we categorize all the parameters into two parts at each training iteration which are updated using different rules. In this way, we gradually zero out the redundant parameters, as we update them using only the ordinary weight decay but no gradients derived from the objective function. As a departure from prior methods that require heavy human works to tune the layer-wise sparsity ratios, prune by solving complicated non-differentiable problems or finetune the model after pruning, our method is characterized by 1) global compression that automatically finds the appropriate per-layer sparsity ratios; 2) end-to-end training; 3) no need for a time-consuming re-training process after pruning; and 4) superior capability to find better winning tickets which have won the initialization lottery.
With leveraging the weight-sharing and continuous relaxation to enable gradient-descent to alternately optimize the supernet weights and the architecture parameters through a bi-level optimization paradigm, textit{Differentiable ARchiTecture Search} (DARTS) has become the mainstream method in Neural Architecture Search (NAS) due to its simplicity and efficiency. However, more recent works found that the performance of the searched architecture barely increases with the optimization proceeding in DARTS. In addition, several concurrent works show that the NAS could find more competitive architectures without labels. The above observations reveal that the supervision signal in DARTS may be a poor indicator for architecture optimization, inspiring a foundational question: instead of using the supervision signal to perform bi-level optimization, textit{can we find high-quality architectures textbf{without any training nor labels}}? We provide an affirmative answer by customizing the NAS as a network pruning at initialization problem. By leveraging recent techniques on the network pruning at initialization, we designed a FreeFlow proxy to score the importance of candidate operations in NAS without any training nor labels, and proposed a novel framework called textit{training and label free neural architecture search} (textbf{FreeNAS}) accordingly. We show that, without any training nor labels, FreeNAS with the proposed FreeFlow proxy can outperform most NAS baselines. More importantly, our framework is extremely efficient, which completes the architecture search within only textbf{3.6s} and textbf{79s} on a single GPU for the NAS-Bench-201 and DARTS search space, respectively. We hope our work inspires more attempts in solving NAS from the perspective of pruning at initialization.

الأسئلة المقترحة

التعليقات
جاري جلب التعليقات جاري جلب التعليقات
mircosoft-partner

هل ترغب بارسال اشعارات عن اخر التحديثات في شمرا-اكاديميا