Materialization optimizations for feature selection workloads

Abstract

There is an arms race in the data management industry to support analytics, in which one critical step is feature selection, the process of selecting a feature set that will be used to build a statistical model. Analytics is one of the biggest topics in data management, and feature selection is widely regarded as the most critical step of analytics; thus, we argue that managing the feature selection process is a pressing data management challenge. We study this challenge by describing a feature-selection language and a supporting prototype system that builds on top of current industrial, R-integration layers. From our interactions with analysts, we learned that feature selection is an interactive, human-in-the-loop process, which means that feature selection workloads are rife with reuse opportunities. Thus, we study how to materialize portions of this computation using not only classical database materialization optimizations but also methods that have not previously been used in database optimization, including structural decomposition methods (like QR factorization) and warmstart. These new methods have no analog in traditional SQL systems, but they may be interesting for array and scientific database applications. On a diverse set of data sets and programs, we find that traditional database-style approaches that ignore these new opportunities are more than two orders of magnitude slower than an optimal plan in this new tradeoff space across multiple R-backends. Furthermore, we show that it is possible to build a simple cost-based optimizer to automatically select a near-optimal execution plan for feature selection.

Query Optimizations for Deep Learning Systems

“…We do so by formalizing partial CNN inference operations as first-class citizens for query processing. In doing so, our work expands a recent line of work on materialization optimizations for feature selection in linear models [168,291] and integrating ML with relational joins [174,91,248,175]. Finally, our work also expands the work in database systems on optimizing memory usage based on data access patterns of queries [98].…”

On optimizing machine learning workloads via kernel fusion

“…We note however that, in practice, it is not very common to encounter such dense data sets with large number of columns [35].…”

“…For a device with 48KB shared memory per SM , the limit on n is close to 6K. Such matrices with limited number of columns are indeed common in many enterprise workloads [35].…”

Regularizing conjunctive features for classification

“…Feature engineering is a critical and resource-consuming task in the development of machine-learning solutions in general, and classifiers in particular [16,18,34]. In the framework proposed by Kimelfeld and Ré [22], the general goal is to utilize the database's knowledge of the raw data structure to provide automated assistance in feature engineering.…”