Deep transfer learning has been widely used for knowledge transmission in recent years. The standard approach of pre-training and subsequently fine-tuning, or linear probing, has shown itself to be effective in many down-stream tasks. Therefore, a challenging and ongoing question arises: how to quantify cross-task transferability that is compatible with transferred results while keeping self-consistency? Existing transferability metrics are estimated on the particular model by conversing source and target tasks. They must be recalculated with all existing source tasks whenever a novel unknown target task is encountered, which is extremely computationally expensive. In this work, we highlight what properties should be satisfied and evaluate existing metrics in light of these characteristics. Building upon this, we propose Principal Gradient Expectation (PGE), a simple yet effective method for assessing transferability across tasks. Specifically, we use a restart scheme to calculate every batch gradient over each weight unit more than once, and then we take the average of all the gradients to get the expectation. Thus, the transferability between the source and target task is estimated by computing the distance of normalized principal gradients. Extensive experiments show that the proposed transferability metric is more stable, reliable and efficient than SOTA methods.
translated by 谷歌翻译
转移学习已成为利用计算机视觉中预先训练模型的流行方法。然而,在不执行计算上昂贵的微调的情况下,难以量化哪个预先训练的源模型适用于特定目标任务,或者相反地,可以容易地适应预先训练的源模型的任务。在这项工作中,我们提出了高斯Bhattacharyya系数(GBC),一种用于量化源模型和目标数据集之间的可转换性的新方法。在第一步中,我们在由源模型定义的特征空间中嵌入所有目标图像,并表示使用每类高斯。然后,我们使用Bhattacharyya系数估计它们的成对类可分离性,从而产生了一种简单有效的源模型转移到目标任务的程度。我们在数据集和架构选择的上下文中评估GBC在图像分类任务上。此外,我们还对更复杂的语义分割转移性估算任务进行实验。我们证明GBC在语义分割设置中大多数评估标准上的最先进的可转移性度量,匹配图像分类中的数据集转移性的最高方法的性能,并且在图像分类中执行最佳的架构选择问题。
translated by 谷歌翻译
Deep transfer learning (DTL) has formed a long-term quest toward enabling deep neural networks (DNNs) to reuse historical experiences as efficiently as humans. This ability is named knowledge transferability. A commonly used paradigm for DTL is firstly learning general knowledge (pre-training) and then reusing (fine-tuning) them for a specific target task. There are two consensuses of transferability of pre-trained DNNs: (1) a larger domain gap between pre-training and downstream data brings lower transferability; (2) the transferability gradually decreases from lower layers (near input) to higher layers (near output). However, these consensuses were basically drawn from the experiments based on natural images, which limits their scope of application. This work aims to study and complement them from a broader perspective by proposing a method to measure the transferability of pre-trained DNN parameters. Our experiments on twelve diverse image classification datasets get similar conclusions to the previous consensuses. More importantly, two new findings are presented, i.e., (1) in addition to the domain gap, a larger data amount and huge dataset diversity of downstream target task also prohibit the transferability; (2) although the lower layers learn basic image features, they are usually not the most transferable layers due to their domain sensitivity.
translated by 谷歌翻译
With the ever-growing model size and the limited availability of labeled training data, transfer learning has become an increasingly popular approach in many science and engineering domains. For classification problems, this work delves into the mystery of transfer learning through an intriguing phenomenon termed neural collapse (NC), where the last-layer features and classifiers of learned deep networks satisfy: (i) the within-class variability of the features collapses to zero, and (ii) the between-class feature means are maximally and equally separated. Through the lens of NC, our findings for transfer learning are the following: (i) when pre-training models, preventing intra-class variability collapse (to a certain extent) better preserves the intrinsic structures of the input data, so that it leads to better model transferability; (ii) when fine-tuning models on downstream tasks, obtaining features with more NC on downstream data results in better test accuracy on the given task. The above results not only demystify many widely used heuristics in model pre-training (e.g., data augmentation, projection head, self-supervised learning), but also leads to more efficient and principled fine-tuning method on downstream tasks that we demonstrate through extensive experimental results.
translated by 谷歌翻译
Does the dominant approach to learn representations (as a side effect of optimizing an expected cost for a single training distribution) remain a good approach when we are dealing with multiple distributions. Our thesis is that such scenarios are better served by representations that are "richer" than those obtained with a single optimization episode. This is supported by a collection of empirical results obtained with an apparently na\"ive ensembling technique: concatenating the representations obtained with multiple training episodes using the same data, model, algorithm, and hyper-parameters, but different random seeds. These independently trained networks perform similarly. Yet, in a number of scenarios involving new distributions, the concatenated representation performs substantially better than an equivalently sized network trained from scratch. This proves that the representations constructed by multiple training episodes are in fact different. Although their concatenation carries little additional information about the training task under the training distribution, it becomes substantially more informative when tasks or distributions change. Meanwhile, a single training episode is unlikely to yield such a redundant representation because the optimization process has no reason to accumulate features that do not incrementally improve the training performance.
translated by 谷歌翻译
自我监督的学习是一个有希望的无监督学习框架,实现了大型浮点网络取得成功。但这种网络不易部署到边缘设备。为了加速模型部署模型,在为各种下游任务中学习这种资源有限的设备的益处,我们向使用移动目标网络的二进制网络提出了一种自我监督的学习方法。特别是,我们建议共同列车,随机初始化的分类器,附加到预用浮点特征提取器,具有二进制网络。此外,我们提出了一种特征相似性损失,动态丢失平衡和改进的多级训练,以进一步提高准确性,并呼叫我们的方法燃烧。我们使用七个数据集的五个下游任务的经验验证显示,烧伤优于二进制网络的自我监督基线,有时优于预测预测。
translated by 谷歌翻译
具有许多预训练模型(PTM)的模型中心已经是深度学习的基石。尽管以高成本建造,但它们仍然保持\ emph {探索}:从业人员通常会通过普及从提供的模型中心中选择一个PTM,然后对PTM进行微调以解决目标任务。这种na \“我的但共同的实践构成了两个障碍,以充分利用预训练的模型中心:(1)通过受欢迎程度选择的PTM选择没有最佳保证;(2)仅使用一个PTM,而其余的PTM则被忽略。理想情况下。理想情况下。 ,为了最大程度地利用预训练的模型枢纽,需要尝试所有PTM的所有组合和广泛的微调每个PTM组合,这会产生指数组合和不可偿还的计算预算。在本文中,我们提出了一种新的范围排名和调整预训练的模型:(1)我们的会议论文〜\ citep {you_logme:_2021}提出的logMe,以估算预先训练模型提取的标签证据的最大值,该标签证据可以在模型中排名所有PTMS用于各种类型的PTM和任务的枢纽\ Emph {微调之前}。(2)如果我们不偏爱模型的体系结构,则可以对排名最佳的PTM进行微调和部署,或者可以通过TOPE调整目标PTM -k通过t排名PTM他提出了b-tuning算法。排名部分基于会议论文,我们在本文中完成了其理论分析,包括启发式证据最大化程序的收敛证明和特征维度的影响。调整零件引入了一种用于调整多个PTM的新型贝叶斯调整(B-Tuning)方法,该方法超过了专门的方法,该方法旨在调整均匀的PTMS,并为调整异质PTMS设置了一种新的技术。利用PTM枢纽的新范式对于整个机器学习社区的大量受众来说可能会很有趣。
translated by 谷歌翻译
基础模型不是模型生产管道的最后一章。以少数数据以少数数据传输到数千个下游任务正在成为基础模型的应用的趋势。在本文中,我们提出了一个通用转移框架:一个传输所有(OTA),将任何视觉基础模型(VFM)转移到具有少数下游数据的下游任务。我们首先通过图像重新表示微调(IRF)将VFM传输到特定于任务特定模型,然后将知识从特定于任务的模型蒸馏到部署的模型,其中包含由下游图像引导的生成(DIGG)产生的数据。OTA在传输时没有对上游数据,VFM和下游任务的依赖性。它还为VFM研究人员提供了一种方法,以释放其上游信息,以便更好地转移,但由于隐私要求而没有泄漏数据。大规模实验在少数数据设置中验证我们方法的有效性和优越性。我们的代码将被释放。
translated by 谷歌翻译
深层模型必须学习强大而可转移的表示形式,以便在新领域上表现良好。尽管已经提出了域转移方法(例如,域的适应性,域的概括)来学习跨域的可转移表示,但通常将它们应用于在Imagenet上预先训练的重置骨架。因此,现有作品很少关注预训练对域转移任务的影响。在本文中,我们对领域适应和泛化的预训练进行了广泛的研究和深入分析,即:网络体系结构,大小,训练损失和数据集。我们观察到,仅使用最先进的主链优于现有的最先进的域适应基线,并将新的基本线设置为Office-Home和Domainnet在10.7 \%和5.5 \%上提高。我们希望这项工作可以为未来的领域转移研究提供更多见解。
translated by 谷歌翻译
从自然语言监督中学习视觉表示,最近在许多开创性的作品中表现出了巨大的希望。通常,这些具有语言的视觉模型表现出对各种数据集和任务的强大可传递性。但是,由于缺乏易于使用的评估工具包和公共基准,评估这些模型的可转让性仍然很具有挑战性。为了解决这个问题,我们构建了高级版(评估语言的视觉任务级传输),这是用于评估(预训练)语言增强视觉模型的第一个基准和工具包。升华由三个组成部分组成。 (i)数据集。作为下游评估套件,它由20个图像分类数据集和35个对象检测数据集组成,每个数据集都用外部知识来增强。 (ii)工具包。开发了自动高参数调谐工具包,以促进下游任务的模型评估。 (iii)指标。多种评估指标用于测量样品效率(零射击和少量)和参数效率(线性探测和完整模型微调)。我们在https://computer-vision-in-the-wild.github.io/elevater/上公开发布leverater
translated by 谷歌翻译
可传递性估计是选择预训练模型和其中的层来转移学习,转移,以最大程度地提高目标任务上的性能并防止负转移的必不可少的工具。现有的估计算法要么需要对目标任务进行深入培训,要么在评估层之间的可传递性方面遇到困难。为此,我们提出了一种简单,高效且有效的可传递性度量,称为“超越”。通过单一传递目标任务的示例,越过可转移性作为在预训练模型及其标签提取的目标示例的特征之间的相互信息。我们通过诉诸于熵的有效替代方案来克服有效的共同信息估计的挑战。从特征表示的角度来看,所得的越来越多地评估了完整性(功能是否包含目标任务的足够信息)和紧凑性(每个类的特征是否足够紧凑,以实现良好的概括)。从理论上讲,我们已经分析了转移学习后的跨度与性能的紧密联系。尽管在10行代码中具有非凡的简单性,但在对32个预训练模型和16个下游任务的广泛评估中,越来越多地表现出色。
translated by 谷歌翻译
在Imagenet或其他大规模数据数据上的预培训模型导致计算机愿景的主要进步,尽管伴随着与策划成本,隐私,使用权和道德问题相关的缺点。在本文中,我们首次研究了基于由图形模拟器生成的合成数据到来自非常不同的域的下游任务的培训模型的可转换性。在使用此类合成数据进行预培训时,我们发现不同任务的下游性能受到不同配置的不同配置(例如,照明,对象姿势,背景等),并且没有单尺寸适合 - 所有解决方案。因此,更好地将合成的预训练数据量身定制到特定的下游任务,以获得最佳性能。我们介绍Task2SIM,一个统一的模型将下游任务表示映射到最佳模拟参数,以为它们生成合成的预训练数据。 Task2SIM通过培训学习此映射,以查找一组“看到”任务上的最佳参数集。曾经训练过,它可以用于预测一个新颖的“看不见”任务的最佳仿真参数,而无需额外的培训。鉴于每级图像数量的预算,我们具有20个不同的下游任务的广泛实验,显示了Task2SIM的任务 - 自适应预训练数据导致明显更好的下游性能,而不是在看见和看不见的任务上的非自适应选择模拟参数。它甚至是竞争对手的真实图像的竞争力。
translated by 谷歌翻译
ImageNet pre-training has enabled state-of-the-art results on many tasks. In spite of its recognized contribution to generalization, we observed in this study that ImageNet pre-training also transfers adversarial non-robustness from pre-trained model into fine-tuned model in the downstream classification tasks. We first conducted experiments on various datasets and network backbones to uncover the adversarial non-robustness in fine-tuned model. Further analysis was conducted on examining the learned knowledge of fine-tuned model and standard model, and revealed that the reason leading to the non-robustness is the non-robust features transferred from ImageNet pre-trained model. Finally, we analyzed the preference for feature learning of the pre-trained model, explored the factors influencing robustness, and introduced a simple robust ImageNet pre-training solution. Our code is available at \url{https://github.com/jiamingzhang94/ImageNet-Pretraining-transfers-non-robustness}.
translated by 谷歌翻译
我们提出了两个新颖的可传递性指标F-OTCE(基于快速最佳运输的条件熵)和JC-otce(联合通信OTCE),以评估源模型(任务)可以使目标任务的学习受益多少,并学习更可转移的表示形式。用于跨域交叉任务转移学习。与需要评估辅助任务的经验可转让性的现有指标不同,我们的指标是无辅助的,以便可以更有效地计算它们。具体而言,F-otce通过首先求解源和目标分布之间的最佳传输(OT)问题来估计可转移性,然后使用最佳耦合来计算源和目标标签之间的负条件熵。它还可以用作损失函数,以最大化目标任务填充源模型的可传递性。同时,JC-OTCE通过在OT问题中包含标签距离来提高F-otce的可转移性鲁棒性,尽管它可能会产生额外的计算成本。广泛的实验表明,F-otce和JC-otce优于最先进的无辅助指标,分别为18.85%和28.88%,与基础真相转移精度相关系数。通过消除辅助任务的训练成本,两个指标将前一个方法的总计算时间从43分钟减少到9.32s和10.78,用于一对任务。当用作损失函数时,F-otce在几个射击分类实验中显示出源模型的传输精度的一致性提高,精度增益高达4.41%。
translated by 谷歌翻译
本文解决了对预先训练的深神经网络进行排名并筛选最下游任务的重要问题。这是具有挑战性的,因为每个任务的基本模型排名只能通过微调目标数据集中的预训练模型来生成,该模型是蛮力且计算昂贵的。最近的高级方法提出了几个轻巧的可转移性指标来预测微调结果。但是,这些方法仅捕获静态表示,但忽略了微调动态。为此,本文提出了一个新的可传递性度量,称为\ textbf {s} elf-challenging \ textbf {f} isher \ textbf {d} is Criminant \ textbf {a} nalisy(\ textbf {\ textbf {sfda})现有作品没有的有吸引力的好处。首先,SFDA可以将静态特征嵌入渔民空间中,并完善它们,以在类之间更好地分离性。其次,SFDA使用一种自我挑战的机制来鼓励不同的预训练模型来区分硬性示例。第三,SFDA可以轻松地为模型集合选择多个预训练的模型。 $ 33 $预培训的$ 11 $下游任务的$ 33 $预培训模型的广泛实验表明,在测量预训练模型的可传递性时,SFDA具有高效,有效和健壮。例如,与最先进的方法NLEEP相比,SFDA平均显示了59.1美元的增益,同时带来了$ 22.5 $ x的墙壁速度速度。该代码将在\ url {https://github.com/tencentarc/sfda}上提供。
translated by 谷歌翻译
作为主导范式,微调目标数据的预先训练模型广泛用于许多深度学习应用,特别是对于小数据集。然而,最近的研究已经明确表明,一旦培训迭代的数量增加,划痕训练都没有比这一训练前策略更糟糕的最终表现。在这项工作中,我们从学习理论中流行的泛化分析的角度重新审视这种现象。我们的结果表明,最终预测精度可能具有对预训练模型的弱依赖性,特别是在大训练迭代的情况下。观察激励我们利用预训练预调整的数据,因为此数据也可用于微调。使用预训练数据的泛化结果表明,当适当的预训练数据包含在微调中时,可以提高目标任务的最终性能。随着理论发现的洞察力,我们提出了一种新颖的选择策略来选择从预训练数据中的子集,以帮助改善目标任务的概括。 8个基准数据集上的图像分类任务的广泛实验结果验证了基于数据选择的微调管道的有效性。
translated by 谷歌翻译
本文关注的是将许多预训练的深神经网络(DNN)(称为检查点)排名,以将学习转移到下游任务。由于广泛使用了DNN,我们可能很容易从各种来源收集数百个检查站。他们中的哪个将最好的人转移到我们感兴趣的下游任务?为了彻底回答这个问题,我们建立了一个神经检查点排名基准(Neucrab),并研究一些直观的排名措施。这些措施是通用的,适用于不同输出类型的检查点,而无需知道如何对哪个数据集进行检查。它们还产生了低计算成本,使它们实际上有意义。我们的结果表明,检查点提取的特征的线性可分离性是可传递性的强烈指标。我们还达到了一种新的排名NLEEP,这在实验中带来了最佳性能。
translated by 谷歌翻译
最近的几种少数学习算法中的大多数都是基于转移学习,其中模型是使用大量源数据进行预训练的,并且随后使用少量目标数据更新了预训练的模型。在基于转移的几次学习中,已经广泛研究了复杂的预训练方法,以进行通用和改进的表示。但是,几乎没有关于更新预训练模型以进行几次学习的研究。在本文中,我们比较了两种流行的更新方法,即微调(即更新整个网络)和线性探测(即仅更新线性分类器),考虑了源数据和目标数据之间的分布变化。我们发现,随着样品数量的增加,无论分布变化如何,微型调整都比线性探测更好。接下来,我们研究了对预训练模型进行微调时,数据增强的有效性和无效性。我们的基本分析表明,需要仔细考虑有关更新预训练模型的详细信息,才能获得更好的射击性能。
translated by 谷歌翻译
域的概括(DG)旨在仅使用有限的源域学习一个通用模型。先前的DG尝试仅由于训练和测试域之间的显着域移动而无法从源域中学习域不变表示。取而代之的是,我们使用Oracle模型使用共同信息重新构建了DG目标,该模型将概括为任何可能的域。我们通过通过预训练的模型近似oracle模型来得出一个可拖动的变化下限,称为使用Oracle(Miro)的相互信息正则化。我们的广泛实验表明,Miro可显着提高分布性能。此外,我们的缩放实验表明,预训练模型的尺度越大,miro的性能提高就越大。源代码可在https://github.com/kakaobrain/miro中获得。
translated by 谷歌翻译
Few-shot learning aims to fast adapt a deep model from a few examples. While pre-training and meta-training can create deep models powerful for few-shot generalization, we find that pre-training and meta-training focuses respectively on cross-domain transferability and cross-task transferability, which restricts their data efficiency in the entangled settings of domain shift and task shift. We thus propose the Omni-Training framework to seamlessly bridge pre-training and meta-training for data-efficient few-shot learning. Our first contribution is a tri-flow Omni-Net architecture. Besides the joint representation flow, Omni-Net introduces two parallel flows for pre-training and meta-training, responsible for improving domain transferability and task transferability respectively. Omni-Net further coordinates the parallel flows by routing their representations via the joint-flow, enabling knowledge transfer across flows. Our second contribution is the Omni-Loss, which introduces a self-distillation strategy separately on the pre-training and meta-training objectives for boosting knowledge transfer throughout different training stages. Omni-Training is a general framework to accommodate many existing algorithms. Evaluations justify that our single framework consistently and clearly outperforms the individual state-of-the-art methods on both cross-task and cross-domain settings in a variety of classification, regression and reinforcement learning problems.
translated by 谷歌翻译