本文解决了缺少嘈杂和非高斯数据数据的数据的问题。与其他流行的方法相比,一种经典的插补方法,即高斯混合模型的期望最大化(EM)算法,它显示出有趣的特性,例如基于K-Neartivt邻居或通过链式方程式进行多个归纳的方法。然而,已知高斯混合模型对异质数据不舒适,当数据被异常值污染或遵循非高斯分布时,这可能导致估计性能差。为了克服这个问题,研究了一种新的EM算法,用于椭圆形分布的混合物与处理潜在丢失数据的特性。本文表明,此问题减少了在通用假设下的角度高斯分布的混合物的估计(即,每个样品都是从椭圆形分布的混合物中绘制的,对于一个样品而言,这可能是不同的)。在这种情况下,与椭圆形分布的混合物相关的完整数据可能非常适合EM框架,由于其条件分布而缺少数据,这被证明是多元$ t $分布。合成数据的实验结果表明,所提出的算法对异常值是可靠的,可以与非高斯数据一起使用。此外,在现实世界数据集上进行的实验表明,与其他经典插补方法相比,该算法非常有竞争力。
translated by 谷歌翻译
线性和二次判别分析是众所周知的经典方法,但可以严重遭受非高斯分布和/或受污染的数据集,主要是因为潜在的高斯假设并不稳健。为了填补这个差距,本文提出了一种新的强大判别分析,其中每个数据点由其自身的任意椭圆对称(ES)分布和其自身的任意比例参数绘制。这种模型允许可能非常异构,独立但非相同的分布式样本。在推导出新的决策规则之后,显示与最先进的方法相比,最大似然参数估计和分类非常简单,快速且坚固。
translated by 谷歌翻译
文章提出了和理论上分析了\ emph {计算有效}多任务学习(MTL)扩展的流行主成分分析(PCA)的监督学习方案\ Cite {Barshan2011Supervised,Bair2006Prediction}。该分析显示(i)默认学习可以通过遭受\ {负转移}来急剧地失败,但(ii)对数据标签的简单反措施避免了负转移,并必须导致改进的性能。支持合成和实际数据基准测试的实验表明,该方法通过最先进的MTL方法实现了可比性的性能,但是在\ EMPH {显着降低计算成本}。
translated by 谷歌翻译
We present Azimuth, an open-source and easy-to-use tool to perform error analysis for text classification. Compared to other stages of the ML development cycle, such as model training and hyper-parameter tuning, the process and tooling for the error analysis stage are less mature. However, this stage is critical for the development of reliable and trustworthy AI systems. To make error analysis more systematic, we propose an approach comprising dataset analysis and model quality assessment, which Azimuth facilitates. We aim to help AI practitioners discover and address areas where the model does not generalize by leveraging and integrating a range of ML techniques, such as saliency maps, similarity, uncertainty, and behavioral analyses, all in one tool. Our code and documentation are available at github.com/servicenow/azimuth.
translated by 谷歌翻译
The proliferation of deep learning techniques led to a wide range of advanced analytics applications in important business areas such as predictive maintenance or product recommendation. However, as the effectiveness of advanced analytics naturally depends on the availability of sufficient data, an organization's ability to exploit the benefits might be restricted by limited data or likewise data access. These challenges could force organizations to spend substantial amounts of money on data, accept constrained analytics capacities, or even turn into a showstopper for analytics projects. Against this backdrop, recent advances in deep learning to generate synthetic data may help to overcome these barriers. Despite its great potential, however, synthetic data are rarely employed. Therefore, we present a taxonomy highlighting the various facets of deploying synthetic data for advanced analytics systems. Furthermore, we identify typical application scenarios for synthetic data to assess the current state of adoption and thereby unveil missed opportunities to pave the way for further research.
translated by 谷歌翻译
Privacy-preserving machine learning has become a key conundrum for multi-party artificial intelligence. Federated learning (FL) and Split Learning (SL) are two frameworks that enable collaborative learning while keeping the data private (on device). In FL, each data holder trains a model locally and releases it to a central server for aggregation. In SL, the clients must release individual cut-layer activations (smashed data) to the server and wait for its response (during both inference and back propagation). While relevant in several settings, both of these schemes have a high communication cost, rely on server-level computation algorithms and do not allow for tunable levels of collaboration. In this work, we present a novel approach for privacy-preserving machine learning, where the clients collaborate via online knowledge distillation using a contrastive loss (contrastive w.r.t. the labels). The goal is to ensure that the participants learn similar features on similar classes without sharing their input data. To do so, each client releases averaged last hidden layer activations of similar labels to a central server that only acts as a relay (i.e., is not involved in the training or aggregation of the models). Then, the clients download these last layer activations (feature representations) of the ensemble of users and distill their knowledge in their personal model using a contrastive objective. For cross-device applications (i.e., small local datasets and limited computational capacity), this approach increases the utility of the models compared to independent learning and other federated knowledge distillation (FD) schemes, is communication efficient and is scalable with the number of clients. We prove theoretically that our framework is well-posed, and we benchmark its performance against standard FD and FL on various datasets using different model architectures.
translated by 谷歌翻译
自适应多机构系统(AMAS)将机器学习问题转变为代理之间的本地合作问题。我们提出了Smapy,这是一种基于合奏的AMA用于移动性预测的实施,除合作规则外,还为其代理提供了机器学习模型。通过详细的方法,我们表明,如果将线性模型集成到合作多代理结构中,则可以在基准传输模式检测数据集上使用线性模型进行非线性分类。获得的结果表明,由于多代理方法,在非线性环境中线性模型的性能有了显着改善。
translated by 谷歌翻译
最近,引入了卷积自动编码器(CAE)进行图像编码。他们对最新的JPEG2000方法实现了性能改进。但是,这些表演是使用具有大量参数的大型CAE获得的,并且其训练需要大量的计算能力。\\在本文中,我们使用具有较小的内存足迹和低计算功率使用的CAE解决了有损图像压缩的问题。为了克服计算成本问题,大多数文献都使用拉格朗日近端正则化方法,这些方法很耗时。\\在这项工作中,我们提出了一种约束的方法和一种新的结构化稀疏学习方法。我们设计了一个算法并在三个约束上进行测试:经典$ \ ell_1 $约束,$ \ ell_ {1,\ infty} $和新的$ \ ell_ {1,1} $约束。实验结果表明,$ \ ell_ {1,1} $约束提供了最佳的结构性稀疏性,从而导致内存和计算成本的高度降低,并且与密集网络相似的速率延伸性能。
translated by 谷歌翻译
我们为级别集方法提出了一个数据驱动的均值曲线求解器。这项工作是我们在[arxiv:2201.12342] [1]和[doi:10.1016/j.jcp.2022.1111291] [arxiv:2201.12342] [1]中的二维策略的$ \ mathbb {r}^3 $的自然扩展。 ]。但是,与[1,2]建立了依赖分辨率的神经网络词典相比,在这里,我们在$ \ mathbb {r}^3 $中开发了两对模型,而不管网格大小如何。我们的前馈网络摄入的水平集,梯度和曲率数据转换为固定接口节点的数值均值曲率近似值。为了降低问题的复杂性,我们使用高斯曲率对模板进行了分类,并将模型分别适合于非堆肥和鞍模式。非插图模板更容易处理,因为它们表现出以单调性和对称性为特征的曲率误差分布。尽管后者允许我们仅在平均曲面频谱的一半上进行训练,但前者帮助我们将数据驱动的融合并在平坦区域附近无缝地融合了基线估计。另一方面,鞍形图案误差结构不太清楚。因此,我们没有利用超出已知信息的潜在信息。在这方面,我们不仅在球形和正弦和双曲线抛物面斑块上训练了我们的模型。我们构建他们的数据集的方法是系统的,但是随机收集样品,同时确保均衡度。我们还诉诸于标准化和降低尺寸,作为预处理步骤和集成正则化以最大程度地减少异常值。此外,我们利用曲率旋转/反射不变性在推理时提高精度。几项实验证实,与现代粒子的界面重建和水平设定方案相比,我们提出的系统可以产生更准确的均值曲线估计。
translated by 谷歌翻译
数据最初是由Peter Hammer引入的,对数据的逻辑分析是一种方法,旨在计算逻辑上的理由,以将一组数据划分为两组观测值,通常称为正和负基。将此分区视为对部分定义的布尔函数的描述;然后处理数据以识别属性的子集,其值可用于表征正组对负基组的观测值。 LAD构成了经典统计学习技术的一种有趣的基于规则的学习替代方案,并具有许多实际应用。然而,根据数据实例的属性,组表征的计算可能是昂贵的。我们工作的一个主要目的是通过计算一些给定属性确实表征正组和负面组来提供一些\ emph {先验}的概率来提供有效的工具来加速计算。为此,我们根据我们对其上的信息提出了几种代表观测数据集的模型。这些模型及其允许我们计算的概率也有助于快速评估当前实际数据的某些属性;此外,它们可以帮助我们更好地分析和理解解决方法所遇到的计算困难。一旦建立了模型,计算概率的数学工具就会来自分析组合。它们使我们能够将所需的概率表示为生成函数系数的比率,然后提供其数值的快速计算。本文的另一个远程目标是表明,分析组合学的方法可以帮助分析LAD和相关领域中各种算法的性能。
translated by 谷歌翻译