Showing papers on "Decision tree model published in 1998"

Method for improvement accuracy of decision tree based text categorization

Abstract: A text categorization method automatically classifies electronic documents by developing a single pooled dictionary of words for a sample set of documents, and then generating a decision tree model, based on the pooled dictionary, for classifying new documents. Adaptive resampling techniques are applied to improve the accuracy of the decision tree model.

A Fast, Bottom-Up Decision Tree Pruning Algorithm with Near-Optimal Generalization

Abstract: In this work, we present a new bottom-up algorithm for decision tree pruning that is very e cient (requiring only a single pass through the given tree), and prove a strong performance guarantee for the generalization error of the resulting pruned tree. We work in the typical setting in which the given tree T may have been derived from the given training sample S, and thus may badly over t S. In this setting, we give bounds on the amount of additional generalization error that our pruning su ers compared to the optimal pruning of T . More generally, our results show that if there is a pruning of T with small error, and whose size is small compared to jSj, then our algorithm will nd a pruning whose error is not much larger. This style of result has been called an index of resolvability result by Barron and Cover in the context of density estimation. A novel feature of our algorithm is its locality | the decision to prune a subtree is based entirely on properties of that subtree and the sample reaching it. To analyze our algorithm, we develop tools of local uniform convergence, a generalization of the standard notion that may prove useful in other settings.

From optimal hyperplanes to optimal decision trees

Abstract: We present an optimal hyperplane searching method for decision tables using Genetic Algorithms. This method can be used to construct a decision tree for a given decision table. We also present some properties of the set of hyperplanes determined by our methods and evaluate an upper bound on the depth of the constructed decision tree.

Decision Trees for Geometric Models

Abstract: A fundamental problem in model-based computer vision is that of identifying which of a given set of geometric models is present in an image. Considering a "probe" to be an oracle that tells us whether or not a model is present at a given point, we study the problem of computing efficient strategies ("decision trees") for probing an image, with the goal to minimize the number of probes necessary (in the worst case) to determine which single model is present. We show that a ⌈lg k⌉ height binary decision tree always exists for k polygonal models (in fixed position), provided (1) they are non-degenerate (do not share boundaries) and (2) they share a common point of intersection. Further, we give an efficient algorithm for constructing such decision tress when the models are given as a set of polygons in the plane. We show that constructing a minimum height tree is NP-complete if either of the two assumptions is omitted. We provide an efficient greedy heuristic strategy and show that, in the general case, it yields a decision tree whose height is at most ⌈lg k⌉ times that of an optimal tree. Finally, we discuss some restricted cases whose special structure allows for improved results.

Complexity issues in Markov decision processes

Abstract: We survey the complexity of computational problems about Markov decision processes: evaluating policies, finding good and best policies, approximating best policies, and related decision problems.

An Optimized Algorithm of Decision Tree

Abstract: Optimization of decision tree is a significant branch in decision tree learning algorithm. An optimized learning algorithm of ID3, a typical decision tree learning algorithm is presented in this paper. When the algorithm selects a new attribute, not only the information gain of the current attribute, but also the information gain of succeeding attributes of this attribute is taken into consideration. In other words, the information gain of attributes in two levels of the decision tree is used. The computational complexity of the modified ID3 (MID3) is the same as that of the ID3. When the two algorithms are applied to learning logic expressions, the performance of MID3 is better than that of ID3.

The Hardness of Speeding-up Knapsack

Abstract: We show that it is not possible to speed-up the Knapsack problemeciently in the parallel algebraic decision tree model More speci -cally, we prove that any parallel algorithm in the xed degree algebraicdecision tree model that solves the decision version of the Knapsackproblem requires Ω(pn) rounds even by using 2 pn processors Weextend the result to the PRAM model without bit-operations Theseresults are consistent with Mulmuley’s [6] recent result on the separa-tion of the strongly-polynomial class and the corresponding NCclassin the arithmetic PRAM model Keywords lower-bounds, parallel algorithms, algebraic decision tree 1 Introduction The primary objective of designing parallel algorithms is to obtain fasteralgorithms Nonetheless, the pursuit of higher speed has to be weightedagainst the concerns of eciency, namely, if we are getting our money’s(processor’s) worth It has been an open theoretical problem whether all theproblems in the class Pcan be made to run in polylogarithmic running time

One help-bit doesn't help

Abstract: Nisan, Rudich, and Saks [25] a&cd whether one help bit can reduce the complexity of computing a boolean function on two inputs in the decision tree model. We prove that one help bit doesn’t help. In general, we prove that [log kJ help bits don’t help for computing k instances off. This result is the best possible, since we exhibit functions for which [log kJ + 1 help bits reduce complexity. This shows a gap between 0 and [log kJ + 1 help bits: either f can be evaluated in depth d (the complexity off) without any help bits, or else [log kJ + 1 help bits are required in order to evaluate k instances in depth d.

A new generalized software complexity metric for distributed programs

Abstract: With increasing complication in distributed programs we must do our best to establish the mechanism of maintaining distributed programs. This paper introduces the notion of the communication complexity unit based on the smallest event communication group, and presents a generalized software complexity metric for measuring the maintenance complexity of distributed programs. We affirm that the smallest event communication group is a unifying mechanism for event and process abstraction whatever distributed programming languages are used and advocate it as a basically generalized communication complexity unit. We have applied the proposed distributed software complexity metric to a moderately complex example. Experience indicates that the proposed metric is indeed very useful.

On Decision Trees with Minimal Average Depth

Abstract: Decision trees are studied in rough set theory [6],[7] and test theory [1], [2], [3] and are used in different areas of applications. The complexity of optimal decision tree (a decision tree with minimal average depth) construction is very high. In the paper some conditions reducing the search are formulated. If these conditions are satisfied, an optimal decision tree for the problem is a result of simple transformation of optimal decision trees for some problems, obtained by decomposition of the initial problem. The decompostion properties are used to show that bounds given in [4] are unimprovable bounds on minimal average depth of decision tree.

A Theoretical Model for Routing Complexity.

Abstract: This paper introduces a formal model for studying the complexity of routing in networks. The aim of this model is to capture both time complexity and space complexity. In particular, the model takes into account the input and output facilities of routers. A routing program is a RAM-program with five additional instructions that allow to handle incoming and outgoing headers, and input and output ports. One of these five additional instructions, called release, captures the possible use of hardware facilities to speed up routing. Using our model, we show that there are routing functions which, if compacted, would require an arbitrarily large computation time to be decoded. The latency is the sum of the time (in bit-operation) required at every intermediate node to establish the route. We also show that, in any n-node network of diameter D, the latency is bounded by O(D+ n1=k logn), for every constant k 2. This latter result has to be compared with the latency of the routing tables which is Θ(D logn).

Arthur-Merlin games in Boolean decision trees

Abstract: It is well known that probabilistic boolean decision trees cannot be much more powerful than deterministic ones. Motivated by a question if randomization can significantly speed up a nondeterministic computation via a boolean decision tree, we address structural properties of Arthur-Merlin games in this model and prove some lower bounds. We consider two cases of interest, the first when the length of communication between the players is limited and the second if it is not. While in the first case we can carry over the relations between the corresponding Turing complexity classes, in the second case we observe in contrast with Turing complexity that a one round Merlin-Arthur protocol is as powerful as a general interactive proof system and, in particular, can simulate a one-round Arthur-Merlin protocol. Moreover, we show that sometimes a Merlin-Arthur protocol can be more efficient than an Arthur-Merlin protocol, and than a Merlin-Arthur protocol with limited communication. This is the case for a boolean function whose set of zeroes is a code with high minimum distance and a natural uniformity condition. Such functions provide an example when the Merlin-Arthur complexity is 1 with one-sided error /spl epsiv//spl isin/(2/3, 1), but at the same time the nondeterministic decision tree complexity is /spl Omega/(n). The latter should be contrasted with another fact we prove. Namely, if a function has Merlin-Arthur complexity 1 with one-sided error probability /spl epsiv//spl isin/(0, 2/3], then its nondeterministic complexity is bounded by a constant.

Hierarchical noise in large systems of independent agents

Abstract: A generalization of the coherent-noise models [M. E. J. Newman and K. Sneppen, Phys. Rev. E{\bf54}, 6226 (1996)] is presented where the agents in the model are subjected to a multitude of stresses, generated in a hierarchy of different contexts. The hierarchy is realized as a Cayley-tree. Two different ways of stress propagation in the tree are considered. In both cases, coherence arises in large subsystems of the tree. Clear similarities between the behavior of the tree model and of the coherent-noise model can be observed. For one of the two methods of stress propagation, the behavior of the tree model can be approximated very well by an ensemble of coherent-noise models, where the sizes $k$ of the systems in the ensemble scale as $k^{-2}$. The results are found to be independent of the tree's structure for a large class of reasonable choices. Additionally, it is found that power-law distributed lifetimes of agents arise even under the complete absence of correlations between the stresses the agents feel.

A Tree Matching Algorithm and VLSI Architecture for Real-Time 2D Object Classification

Abstract: This paper presents a real-time classification algorithm for two-dimensional (2D) object contours using a tree model which is implemented in a modular very large scale integration (VLSI) architecture. The hardware implementation takes advantage of pipelining, parallelism, and the speed of VLSI technology to perform real-time object classification. Using the multiresolution tree model, the classification algorithm is invariant under 2D similarity transformations and recognizes the visible portions of occluded objects. The VLSI classification system is implemented in 0.8 mm CMOS and is capable of performing 34000 matchings per second.

Time Complexity of the Incremental Pruning Algorithm for Markov Decision Networks

Abstract: A new representation technique for the planning of decisions under uncertainty is presented. The technique, called Markov decision networks (mdns), solves planning tasks that cannot or can hardly be solved by any other known representation technique, such as decision trees, in‡uence diagrams, and pomdps. We show that linear mdns are pomdps. Hence, the incrementalpruning algorithm for pomdps is applicable to this subclass of mdns. It turns out that the time complexity of the algorithm is lower for mdns than for general pomdps, due to the division of the state space into state groups. An example shows the practical bene...t of our ...ndings: an optimal strategy can be computed in a few seconds.