Corpus ID: 9384267

A Reinforcement Learning Approach to Online Learning of Decision Trees

  title={A Reinforcement Learning Approach to Online Learning of Decision Trees},
  author={Abhinav Garlapati and Aditi Raghunathan and Vaishnavh Nagarajan and Balaraman Ravindran},
Online decision tree learning algorithms typically examine all features of a new data point to update model parameters. We propose a novel alternative, Reinforcement Learning- based Decision Trees (RLDT), that uses Reinforcement Learning (RL) to actively examine a minimal number of features of a data point to classify it with high accuracy. Furthermore, RLDT optimizes a long term return, providing a better alternative to the traditional myopic greedy approach to growing decision trees. We… Expand
Reinforcement Learning Based Decision Tree Induction Over Data Streams with Concept Drifts
A reinforcement learning approach is used to train a Markov Decision Process (MDP), which enables the creation of a short and highly accurate decision tree and a reset operation is proposed that allows for local re-learning of outdated parts of the tree. Expand
Dyna: toward a self-optimizing declarative language for machine learning applications
It is hoped that reinforcement learning will identify good policies reasonably quickly-offloading the burden of writing efficient code from human programmers, and adaptively choosing among these strategies to maximize efficiency for a given workload. Expand
Automatic web content personalization through reinforcement learning
A system that employs Web intelligence to perform automatic adaptations on single elements composing a Web page and a reinforcement learning algorithm is utilized to manage user profiles to understand and predict users' behaviors and needs. Expand
Learning topic description from clustering of trusted user roles and event models characterizing distributed provenance networks: a reinforcement learning approach
A methodology for unifying similar provenance graph models for adapting and averaging the policy path classifiers associated with individual models to produce a reduced set of unified models derived during training is provided. Expand


Self Organizing Decision Tree Based on Reinforcement Learning and its Application on State Space Partition
A tree induction problem is regarded as a reinforcement learning problem and solved by the technique in that problem domain and the state spare partition in a critic actor model, adaptive heuristic critic, is replaced by a regression tree, which is constructed by the proposed method. Expand
Adaptive building of decision trees by reinforcement learning
Decision tree learning represents a well known family of inductive learning algorithms that are able to extract, from the presented training sets, classification rules whose preconditions can beExpand
Efficient determination of dynamic split points in a decision tree
A number of alternative approaches that generate a small number of candidate split points dynamically with little overhead are described, and it is argued that these approaches are preferable to pre-discretization, and experimental evidence that they yield probabilistic decision trees with the same prediction accuracy as the traditional dynamic approach. Expand
Reinforcement Learning with a Gaussian mixture model
A new approach for RL in continuous domains based on Probability Density Estimations is proposed, which is non-parametric and provides an estimation of the variance of the approximated function at any point of the domain. Expand
Incremental Induction of Decision Trees
  • P. Utgoff
  • Computer Science
  • Machine Learning
  • 2004
An incremental algorithm for inducing decision trees equivalent to those formed by Quinlan's nonincremental ID3 algorithm, given the same training instances is presented, named ID5R. Expand
Learning Sorting and Decision Trees with POMDPs
The main lessons are that the use of suitable heuris-tics and representations allows for the solution of sorting and classiication pomdps of non-trivial sizes, and the quality of the resulting solutions are competitive with the best algorithms, and problematic aspects in decision tree learning are naturally accommodated. Expand
Online feature selection for mining big data
This work investigates the problem of Online Feature Selection (OFS) in which the online learner is only allowed to maintain a classifier involved a small and fixed number of features, and presents an effective algorithm to solve the problem, and gives the theoretical analysis and empirical performance of the proposed algorithms. Expand
Efficient Online and Batch Learning Using Forward Backward Splitting
The two phase approach enables sparse solutions when used in conjunction with regularization functions that promote sparsity, such as l1, l2, l22, and l∞ regularization, and is extended and given efficient implementations for very high-dimensional data with sparsity. Expand
Sparse Online Learning via Truncated Gradient
This work proposes a general method called truncated gradient to induce sparsity in the weights of online-learning algorithms with convex loss and finds for datasets with large numbers of features, substantial sparsity is discoverable. Expand
Mining high-speed data streams
This paper describes and evaluates VFDT, an anytime system that builds decision trees using constant memory and constant time per example, and applies it to mining the continuous stream of Web access data from the whole University of Washington main campus. Expand