Incremental Induction of Decision Trees

By Paul E. Utgoff. 1989

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

Towards Making Systems Forget with Machine Unlearning (2015, methodology): “[74] proposed an incremental algorithm to induce decision trees equivalent to the trees formed by Quinlan’s ID3 algorithm.”
Extremely Fast Decision Tree (2018, methodology): “Methods proposed prior to HT are either significantly less tight compared to HT in their approximation of a batch tree [14] or unsuitable for noisy streams and prohibitively computationally expensive [26].”
Hybrid decision tree and naïve Bayes classifiers for multi-class classification tasks (2014, methodology): “DT is a classiﬁcation tool commonly used in data mining tasks such as ID3 (Quinlan, 1986), ID4 (Utgoff, 1989), ID5 (Utgoff, 1988), C4.5 (Quinlan, 1993), C5.”
Hybrid decision tree (2002, methodology): “Utgoff (1989) presented ID5R which incrementally generated a tree identical to that could be generated by ID3 (Quinlan, 1986) through using a pull-up process to move expected splits to the roots of subtrees when new examples were provided.”
Decision Trees: An Overview and Their Use in Medicine (2002, methodology): “In this manner, different hybrid approaches from both fields have been introduced [Utgoff, 1989b; Craven, 1996; Zorman, 1999].”
Monotonicity Maintenance in Information-Theoretic Machine Learning Algorithms (1995, methodology): “TDIDT algorithms that were not studied here, such as C4.5 (Quinlan, 1987), ID4 (Schlimmer & Fisher, 1986), and ID5R ( Utgoff, 1989 ), may also be adapted to deal with MOC problems using a similar approach.”
Unifying Instance-Based and Rule-Based Induction (1996, methodology): “The score is a measure that compares all 5 algorithms simultaneously (default included), looking not only at which one is the most accurate one, but also taking into consideration that being the second is better than being the worst, etc. Speci(cid:12)cally, for each domain the most accurate…”
Instance-Based Learning Algorithms (1991, background): “Several researchers have shown how to decrease these costs significantly (Schlimmer & Fisher, 1986; Utgoff, 1989; Van de Velde, 1989).”
Mining concept-drifting data streams using ensemble classifiers (2003, background): “Incremental or online data mining methods [29, 15] are another option for mining data streams.”
Transfer learning for speech and language processing (2015, background): “Y+ Y− M(x)+ M(x) X+ P(X)+ Conventional ML Model transfer[10] Multitask learning[11] P(X)- Model Adaptation[12], [13], incremental learning[14] X− Co-training[15] Heterogeneous transfer learning[16], [17] Analogy learning [18]”
A Survey of Methods for Scaling Up Inductive Algorithms (1999, background): “They show that for eight out of ﬁfteen data sets, T2 produces two-level trees which rival or surpass the de facto standard C4.5 (see below) (Quinlan, 1993).”