A Survey of Methods for Scaling Up Inductive Algorithms

By Foster J. Provost, Venkateswarlu Kolluri. 1999

How researchers see this paper

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

Citations in the corpus

Multiagent Learning: Basics, Challenges, and Prospects (2012, background): “…is, an approach in which two or more learning strategies are combined into one learning system; and parallel inductive learning, that is, the domain that studies how to exploit the parallelism present in many learning algorithms in order to scale to more complex problems (Provost and Kolluri 1999).”
SimPoint 3.0: Faster and More Flexible Program Phase Analysis (2005, background): “Sampling is common in clustering for datasets which are too large to ﬁt in main memory [25, 26].”
Stochastic gradient boosted distributed decision trees (2009, background): “There are already several existing approaches to distributed tree construction that aimed to improve scalability in terms of memory usage and improving training performance [8].”
Data mining in soft computing framework: a survey (2002, background): “The three main approaches to scaling up include [54] the following:”
Recent advances and emerging challenges of feature selection in the context of big data (2015, background): “The crucial point is seldom how fast you can run on a particular problem, but rather how large a problem you can deal with [100].”
Big Data Investment, Skills, and Firm Value (2014, unknown): “Although distributed data processing has a long history (e.g., see Provost and Kolluri 1999 for a survey of the literature over a decade ago), the scale and rate of data collection in recent years has raised the returns to innovation in data processing technologies.”
Tree Induction for Probability-Based Ranking (2003, background): “Tree induction also works comparatively well with very large data sets ( Provost & Kolluri, 1999 ), with large numbers of variables, and with mixed-type data (continuous, nominal, Boolean, etc.).”
PLANET: Massively Parallel Learning of Tree Ensembles with MapReduce (2009, background): “[27] contains an excellent survey of existing approaches, along with the motivations for large scale tree learning.”
Tree Induction vs. Logistic Regression: A Learning-Curve Analysis (2003, unknown): “If the resultant AUR is high (0.85 or greater) continue to explore tree-based (or other nonparametric) options (for example, BPET, or methods that can deal with more data than can fit in main memory (Provost and Kolluri, 1999)).”
Predictive Modeling With Big Data: Is Bigger Really Better? (2013, unknown): “Authors have argued for decades for the need to scale up predictive modeling algorithms to massive data.(1,2) However, there is surprisingly scant empirical evidence supporting continued”
Machine learning strategies for systems with invariance properties (2016, unknown): “For RFs with no cap on the maximum tree depth, the number of tree leaves tends to scale linearly with the training data set size (if each data point adds new information to the data set), and the tree depth scales roughly logarithmically with the number of tree leaves [49].”
Interpretable models from distributed data via merging of decision trees (2013, background): “Many authors have designed decision tree algorithms working on distributed data [9], [4], [10], [11], [12].”