Materialization optimizations for feature selection workloads

By Ce Zhang, Arun Kumar, Christopher R'e. 2014

147 citations recorded by the source corpus.

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Artificial Intelligence

Citations in the corpus

BoostClean: Automated Error Detection and Repair for Machine Learning (2017, background): “d machine learning (ML) models in prediction, recommendation, and automation. The database community has built systems to support almost every stage of the development process including featurization [1,53], distributed model training [2,17,19,24], and model deployment [16]. However, an under-served”
Socratic Learning (2016, background): “This observation suggests that the generative model can be improved if it incorporates such features.”
Automatic Database Management System Tuning Through Large-scale Machine Learning (2017, background): “There is also both practical and theoretical work backing its effectiveness as a consistent feature selection algorithm [56, 57, 64, 9].”
Fonduer: Knowledge Base Construction from Richly Formatted Data (2017, background): “This library eliminates features as an input a user must provide, which significantly reduces development time [33].”
Active Sampler: Light-weight Accelerator for Complex Data Analytics at Scale (2015, background): “Many optimizations have been proposed from a systems perspective for speciﬁc classes of models [31, 32, 13, 30, 4, 7].”
SystemML: Declarative Machine Learning on Spark (2016, methodology): “Frameworks like MLBase [20, 32] or Columbus [42] further enable declarative ML tasks of model or feature selection, where they optimize both model accuracy and performance of ML algorithms.”
Data programming with DDLite: putting humans in a different part of the loop (2016, methodology): “To enable faster development of these machine learning-based systems, we explored automated feature extraction libraries [2, 13] and accelerated inference methods [14, 9].”
ProvDB: A System for Lifecycle Management of Collaborative Analysis Workflows (2016, background): “Much of that work has focused on the “train-ing” aspect; several general-purpose systems like GraphLab, Ten-sorFlow, Parameter Server, etc., have been designed over the years, and there is also much work on speciﬁc aspects of ML pipeline (e.g., feature engineering [55].”
Visual exploration of machine learning results using data cube analysis (2016, background): “With this expertise, many researchers have studied on helping machine learning engineers perform feature engineering [2, 26], but only a few researchers have studied on model selection or interpretation [13, 6].”
The Missing Piece in Complex Analytics: Low Latency, Scalable Model Management and Serving with Velox (2014, background): “Similarly, Columbus [22] demonstrates the power of caching and model reuse in in-situ feature learning.”
KeystoneML: Optimizing Pipelines for Large-Scale Advanced Analytics (2016, background): “Other work [60, 5] has looked at optimizing caching strategies and operator selection in the regime of feature selection and feature generation workloads.”
Probabilistic Demand Forecasting at Scale (2017, background): “Examples of such optimization aims include the increase of task parallelism [5] as well as enabling intelligent materialization and re-use of intermediate results [31, 16].”