Deep neural networks (DNNs) often have to be compressed, via pruning and/or quantization, before they can be deployed in practical settings. In this work we propose a new compression-aware minimizer dubbed CrAM that modifies the optimization step in a principled way, in order to produce models whose local loss behavior is stable under compression operations such as pruning. Thus, dense models trained via CrAM should be compressible post-training, in a single step, without significant accuracy loss. Experimental results on standard benchmarks, such as residual networks for ImageNet classification and BERT models for language modelling, show that CrAM produces dense models that can be more accurate than the standard SGD/Adam-based baselines, but which are stable under weight pruning: specifically, we can prune models in one-shot to 70-80% sparsity with almost no accuracy loss, and to 90% with reasonable ($\sim 1\%$) accuracy loss, which is competitive with gradual compression methods. Additionally, CrAM can produce sparse models which perform well for transfer learning, and it also works for semi-structured 2:4 pruning patterns supported by GPU hardware. The code for reproducing the results is available at //github.com/IST-DASLab/CrAM .
相關內容
In response to the prevalent challenge of overfitting in deep neural networks, this paper introduces Simultaneous Learning, a regularization approach drawing on principles of Transfer Learning and Multi-task Learning. We leverage auxiliary datasets with the target dataset, the UFOP-HVD, to facilitate simultaneous classification guided by a customized loss function featuring an inter-group penalty. This experimental configuration allows for a detailed examination of model performance across similar (PlantNet) and dissimilar (ImageNet) domains, thereby enriching the generalizability of Convolutional Neural Network models. Remarkably, our approach demonstrates superior performance over models without regularization and those applying dropout regularization exclusively, enhancing accuracy by 5 to 22 percentage points. Moreover, when combined with dropout, the proposed approach improves generalization, securing state-of-the-art results for the UFOP-HVD challenge. The method also showcases efficiency with significantly smaller sample sizes, suggesting its broad applicability across a spectrum of related tasks. In addition, an interpretability approach is deployed to evaluate feature quality by analyzing class feature correlations within the network's convolutional layers. The findings of this study provide deeper insights into the efficacy of Simultaneous Learning, particularly concerning its interaction with the auxiliary and target datasets.
LoSparse: Structured Compression of Large Language Models based on Low-Rank and Sparse Approximation
Transformer models have achieved remarkable results in various natural language tasks, but they are often prohibitively large, requiring massive memories and computational resources. To reduce the size and complexity of these models, we propose LoSparse (Low-Rank and Sparse approximation), a novel model compression technique that approximates a weight matrix by the sum of a low-rank matrix and a sparse matrix. Our method combines the advantages of both low-rank approximations and pruning, while avoiding their limitations. Low-rank approximation compresses the coherent and expressive parts in neurons, while pruning removes the incoherent and non-expressive parts in neurons. Pruning enhances the diversity of low-rank approximations, and low-rank approximation prevents pruning from losing too many expressive neurons. We evaluate our method on natural language understanding, question answering, and natural language generation tasks. We show that it significantly outperforms existing compression methods.
The term "metaverse", a three-dimensional virtual universe similar to the real realm, has always been full of imagination since it was put forward in the 1990s. Recently, it is possible to realize the metaverse with the continuous emergence and progress of various technologies, and thus it has attracted extensive attention again. It may bring a lot of benefits to human society such as reducing discrimination, eliminating individual differences, and socializing. However, everything has security and privacy concerns, which is no exception for the metaverse. In this article, we firstly analyze the concept of the metaverse and propose that it is a super virtual-reality (VR) ecosystem compared with other VR technologies. Then, we carefully analyze and elaborate on possible security and privacy concerns from four perspectives: user information, communication, scenario, and goods, and immediately, the potential solutions are correspondingly put forward. Meanwhile, we propose the need to take advantage of the new buckets effect to comprehensively address security and privacy concerns from a philosophical perspective, which hopefully will bring some progress to the metaverse community.
In this study, we examine numerical approximations for 2nd-order linear-nonlinear differential equations with diverse boundary conditions, followed by the residual corrections of the first approximations. We first obtain numerical results using the Galerkin weighted residual approach with Bernstein polynomials. The generation of residuals is brought on by the fact that our first approximation is computed using numerical methods. To minimize these residuals, we use the compact finite difference scheme of 4th-order convergence to solve the error differential equations in accordance with the error boundary conditions. We also introduce the formulation of the compact finite difference method of fourth-order convergence for the nonlinear BVPs. The improved approximations are produced by adding the error values derived from the approximations of the error differential equation to the weighted residual values. Numerical results are compared to the exact solutions and to the solutions available in the published literature to validate the proposed scheme, and high accuracy is achieved in all cases
In this paper we introduce InDistill, a model compression approach that combines knowledge distillation and channel pruning in a unified framework for the transfer of the critical information flow paths from a heavyweight teacher to a lightweight student. Such information is typically collapsed in previous methods due to an encoding stage prior to distillation. By contrast, InDistill leverages a pruning operation applied to the teacher's intermediate layers reducing their width to the corresponding student layers' width. In that way, we force architectural alignment enabling the intermediate layers to be directly distilled without the need of an encoding stage. Additionally, a curriculum learning-based training scheme is adopted considering the distillation difficulty of each layer and the critical learning periods in which the information flow paths are created. The proposed method surpasses state-of-the-art performance on three standard benchmarks, i.e. CIFAR-10, CUB-200, and FashionMNIST by 3.08%, 14.27%, and 1% mAP, respectively, as well as on more challenging evaluation settings, i.e. ImageNet and CIFAR-100 by 1.97% and 5.65% mAP, respectively.
Leveraging second-order information at the scale of deep networks is one of the main lines of approach for improving the performance of current optimizers for deep learning. Yet, existing approaches for accurate full-matrix preconditioning, such as Full-Matrix Adagrad (GGT) or Matrix-Free Approximate Curvature (M-FAC) suffer from massive storage costs when applied even to medium-scale models, as they must store a sliding window of gradients, whose memory requirements are multiplicative in the model dimension. In this paper, we address this issue via an efficient and simple-to-implement error-feedback technique that can be applied to compress preconditioners by up to two orders of magnitude in practice, without loss of convergence. Specifically, our approach compresses the gradient information via sparsification or low-rank compression \emph{before} it is fed into the preconditioner, feeding the compression error back into future iterations. Extensive experiments on deep neural networks for vision show that this approach can compress full-matrix preconditioners by up to two orders of magnitude without impact on accuracy, effectively removing the memory overhead of full-matrix preconditioning for implementations of full-matrix Adagrad (GGT) and natural gradient (M-FAC). Our code is available at //github.com/IST-DASLab/EFCP.
Few-shot learning (FSL) is one of the significant and hard problems in the field of image classification. However, in contrast to the rapid development of the visible light dataset, the progress in SAR target image classification is much slower. The lack of unified benchmark is a key reason for this phenomenon, which may be severely overlooked by the current literature. The researchers of SAR target image classification always report their new results on their own datasets and experimental setup. It leads to inefficiency in result comparison and impedes the further progress of this area. Motivated by this observation, we propose a novel few-shot SAR image classification benchmark (FewSAR) to address this issue. FewSAR consists of an open-source Python code library of 15 classic methods in three categories for few-shot SAR image classification. It provides an accessible and customizable testbed for different few-shot SAR image classification task. To further understanding the performance of different few-shot methods, we establish evaluation protocols and conduct extensive experiments within the benchmark. By analyzing the quantitative results and runtime under the same setting, we observe that the accuracy of metric learning methods can achieve the best results. Meta-learning methods and fine-tuning methods perform poorly on few-shot SAR images, which is primarily due to the bias of existing datasets. We believe that FewSAR will open up a new avenue for future research and development, on real-world challenges at the intersection of SAR image classification and few-shot deep learning. We will provide our code for the proposed FewSAR at //github.com/solarlee/FewSAR.
With the rise of powerful pre-trained vision-language models like CLIP, it becomes essential to investigate ways to adapt these models to downstream datasets. A recently proposed method named Context Optimization (CoOp) introduces the concept of prompt learning -- a recent trend in NLP -- to the vision domain for adapting pre-trained vision-language models. Specifically, CoOp turns context words in a prompt into a set of learnable vectors and, with only a few labeled images for learning, can achieve huge improvements over intensively-tuned manual prompts. In our study we identify a critical problem of CoOp: the learned context is not generalizable to wider unseen classes within the same dataset, suggesting that CoOp overfits base classes observed during training. To address the problem, we propose Conditional Context Optimization (CoCoOp), which extends CoOp by further learning a lightweight neural network to generate for each image an input-conditional token (vector). Compared to CoOp's static prompts, our dynamic prompts adapt to each instance and are thus less sensitive to class shift. Extensive experiments show that CoCoOp generalizes much better than CoOp to unseen classes, even showing promising transferability beyond a single dataset; and yields stronger domain generalization performance as well. Code is available at //github.com/KaiyangZhou/CoOp.
Reasoning with knowledge expressed in natural language and Knowledge Bases (KBs) is a major challenge for Artificial Intelligence, with applications in machine reading, dialogue, and question answering. General neural architectures that jointly learn representations and transformations of text are very data-inefficient, and it is hard to analyse their reasoning process. These issues are addressed by end-to-end differentiable reasoning systems such as Neural Theorem Provers (NTPs), although they can only be used with small-scale symbolic KBs. In this paper we first propose Greedy NTPs (GNTPs), an extension to NTPs addressing their complexity and scalability limitations, thus making them applicable to real-world datasets. This result is achieved by dynamically constructing the computation graph of NTPs and including only the most promising proof paths during inference, thus obtaining orders of magnitude more efficient models. Then, we propose a novel approach for jointly reasoning over KBs and textual mentions, by embedding logic facts and natural language sentences in a shared embedding space. We show that GNTPs perform on par with NTPs at a fraction of their cost while achieving competitive link prediction results on large datasets, providing explanations for predictions, and inducing interpretable models. Source code, datasets, and supplementary material are available online at //github.com/uclnlp/gntp.
Pre-trained deep neural network language models such as ELMo, GPT, BERT and XLNet have recently achieved state-of-the-art performance on a variety of language understanding tasks. However, their size makes them impractical for a number of scenarios, especially on mobile and edge devices. In particular, the input word embedding matrix accounts for a significant proportion of the model's memory footprint, due to the large input vocabulary and embedding dimensions. Knowledge distillation techniques have had success at compressing large neural network models, but they are ineffective at yielding student models with vocabularies different from the original teacher models. We introduce a novel knowledge distillation technique for training a student model with a significantly smaller vocabulary as well as lower embedding and hidden state dimensions. Specifically, we employ a dual-training mechanism that trains the teacher and student models simultaneously to obtain optimal word embeddings for the student vocabulary. We combine this approach with learning shared projection matrices that transfer layer-wise knowledge from the teacher model to the student model. Our method is able to compress the BERT_BASE model by more than 60x, with only a minor drop in downstream task metrics, resulting in a language model with a footprint of under 7MB. Experimental results also demonstrate higher compression efficiency and accuracy when compared with other state-of-the-art compression techniques.
小貼士
登錄享
注冊地址： 北京市海淀區羊坊店路18號2幢3層301-191