اشترك بالحزمة الذهبية واحصل على وصول غير محدود شمرا أكاديميا
تسجيل مستخدم جديدSmall Data, Big Decisions: Model Selection in the Small-Data Regime
86
0
0.0
(
0
)
اسأل ChatGPT حول البحث
ﻻ يوجد ملخص باللغة العربية
Highly overparametrized neural networks can display curiously strong generalization performance - a phenomenon that has recently garnered a wealth of theoretical and empirical research in order to better understand it. In contrast to most previous work, which typically considers the performance as a function of the model size, in this paper we empirically study the generalization performance as the size of the training set varies over multiple orders of magnitude. These systematic experiments lead to some interesting and potentially very useful observations; perhaps most notably that training on smaller subsets of the data can lead to more reliable model selection decisions whilst simultaneously enjoying smaller computational costs. Our experiments furthermore allow us to estimate Minimum Description Lengths for common datasets given modern neural network architectures, thereby paving the way for principled model selection taking into account Occams-razor.
قيم البحث
اقرأ أيضاً
Bin Packing problems have been widely studied because of their broad applications in different domains. Known as a set of NP-hard problems, they have different vari- ations and many heuristics have been proposed for obtaining approximate solutions. S
pecifically, for the 1D variable sized bin packing problem, the two key sets of optimization heuristics are the bin assignment and the bin allocation. Usually the performance of a single static optimization heuristic can not beat that of a dynamic one which is tailored for each bin packing instance. Building such an adaptive system requires modeling the relationship between bin features and packing perform profiles. The primary drawbacks of traditional AI machine learnings for this task are the natural limitations of feature engineering, such as the curse of dimensionality and feature selection quality. We introduce a deep learning approach to overcome the drawbacks by applying a large training data set, auto feature selection and fast, accurate labeling. We show in this paper how to build such a system by both theoretical formulation and engineering practices. Our prediction system achieves up to 89% training accuracy and 72% validation accuracy to select the best heuristic that can generate a better quality bin packing solution.
We are surrounded by huge amounts of large-scale high dimensional data. It is desirable to reduce the dimensionality of data for many learning tasks due to the curse of dimensionality. Feature selection has shown its effectiveness in many application
s by building simpler and more comprehensive model, improving learning performance, and preparing clean, understandable data. Recently, some unique characteristics of big data such as data velocity and data variety present challenges to the feature selection problem. In this paper, we envision these challenges of feature selection for big data analytics. In particular, we first give a brief introduction about feature selection and then detail the challenges of feature selection for structured, heterogeneous and streaming data as well as its scalability and stability issues. At last, to facilitate and promote the feature selection research, we present an open-source feature selection repository (scikit-feature), which consists of most of current popular feature selection algorithms.
Most of existing clustering algorithms are proposed without considering the selection bias in data. In many real applications, however, one cannot guarantee the data is unbiased. Selection bias might bring the unexpected correlation between features
and ignoring those unexpected correlations will hurt the performance of clustering algorithms. Therefore, how to remove those unexpected correlations induced by selection bias is extremely important yet largely unexplored for clustering. In this paper, we propose a novel Decorrelation regularized K-Means algorithm (DCKM) for clustering with data selection bias. Specifically, the decorrelation regularizer aims to learn the global sample weights which are capable of balancing the sample distribution, so as to remove unexpected correlations among features. Meanwhile, the learned weights are combined with k-means, which makes the reweighted k-means cluster on the inherent data distribution without unexpected correlation influence. Moreover, we derive the updating rules to effectively infer the parameters in DCKM. Extensive experiments results on real world datasets well demonstrate that our DCKM algorithm achieves significant performance gains, indicating the necessity of removing unexpected feature correlations induced by selection bias when clustering.
In Big data era, information integration often requires abundant data extracted from massive data sources. Due to a large number of data sources, data source selection plays a crucial role in information integration, since it is costly and even impos
sible to access all data sources. Data Source selection should consider both efficiency and effectiveness issues. For efficiency, the approach should achieve high performance and be scalability to fit large data source amount. From effectiveness aspect, data quality and overlapping of sources are to be considered, since data quality varies much from data sources, with significant differences in the accuracy and coverage of the data provided, and the overlapping of sources can even lower the quality of data integrated from selected data sources. In this paper, we study source selection problem in textit{Big Data Era} and propose methods which can scale to datasets with up to millions of data sources and guarantee the quality of results. Motivated by this, we propose a new object function taking the expected number of true values a source can provide as a criteria to evaluate the contribution of a data source. Based on our proposed index we present a scalable algorithm and two pruning strategies to improve the efficiency without sacrificing precision. Experimental results on both real world and synthetic data sets show that our methods can select sources providing a large proportion of true values efficiently and can scale to massive data sources.
When applying Value at Risk (VaR) procedures to specific positions or portfolios, we often focus on developing procedures only for the specific assets in the portfolio. However, since this small portfolio risk analysis ignores information from assets
outside the target portfolio, there may be significant information loss. In this paper, we develop a dynamic process to incorporate the ignored information. We also study how to overcome the curse of dimensionality and discuss where and when benefits occur from a large number of assets, which is called the blessing of dimensionality. We find empirical support for the proposed method.
الأسئلة المقترحة
سجل دخول لتتمكن من نشر تعليقات
التعليقات
جاري جلب التعليقات
سجل دخول لتتمكن من متابعة معايير البحث التي قمت باختيارها
