Results 1 
8 of
8
A Continuous Approach to Inductive Inference
 Mathematical Programming
, 1992
"... In this paper we describe an interior point mathematical programming approach to inductive inference. We list several versions of this problem and study in detail the formulation based on hidden Boolean logic. We consider the problem of identifying a hidden Boolean function F : f0; 1g n ! f0; 1g ..."
Abstract

Cited by 42 (2 self)
 Add to MetaCart
(Show Context)
In this paper we describe an interior point mathematical programming approach to inductive inference. We list several versions of this problem and study in detail the formulation based on hidden Boolean logic. We consider the problem of identifying a hidden Boolean function F : f0; 1g n ! f0; 1g using outputs obtained by applying a limited number of random inputs to the hidden function. Given this inputoutput sample, we give a method to synthesize a Boolean function that describes the sample. We pose the Boolean Function Synthesis Problem as a particular type of Satisfiability Problem. The Satisfiability Problem is translated into an integer programming feasibility problem, that is solved with an interior point algorithm for integer programming. A similar integer programming implementation has been used in a previous study to solve randomly generated instances of the Satisfiability Problem. In this paper we introduce a new variant of this algorithm, where the Riemannian metric used...
A Minsat approach for learning in logic domains
 INFORMS Journal on computing
, 2002
"... This paper describes a method for learning logic relationships that correctlyclassifya given data set. The method derives from given logic data certain minimum cost satisfiabilityproblems, solves these problems, and deduces from the solutions the desired logic relationships. Uses of the method inclu ..."
Abstract

Cited by 18 (15 self)
 Add to MetaCart
(Show Context)
This paper describes a method for learning logic relationships that correctlyclassifya given data set. The method derives from given logic data certain minimum cost satisfiabilityproblems, solves these problems, and deduces from the solutions the desired logic relationships. Uses of the method include data mining, learning logic in expert systems, and identification of critical characteristics for recognition systems. Computational tests have proved that the method is fast and effective.
A Method for Controlling Errors in TwoClass Classification
, 1998
"... Two types of errors can be associated with classifying data into two categories A and B: misclassifying an item in A as one of B and, conversely, misclassifying an item in B as one of A. Tight control over these two errors is needed in many practical situations. For example, an error of one type may ..."
Abstract

Cited by 7 (5 self)
 Add to MetaCart
Two types of errors can be associated with classifying data into two categories A and B: misclassifying an item in A as one of B and, conversely, misclassifying an item in B as one of A. Tight control over these two errors is needed in many practical situations. For example, an error of one type may have far more serious consequences than one of the other type. Most previous work on twoclass classifiers does not allow for such control. In this paper, we describe a general approach that supports tight error control for twoclass classification and that can utilize almost any twoclass classification method as part of the decision mechanism. The main idea is to construct from the given training data a family of classifiers and then, using the training data once more, to estimate two distributions of certain vote totals. The error control is achieved via the two estimated distributions. The control is effective if the estimates of the two distributions are close to the true distributions...
Application of a New Logic Domain Method for the Diagnosis of Hepatocellular Carcinoma
, 2001
"... In this paper we describe the application of a new learning tool for the diagnosis of hepatocellular carcinoma. The method adopted operates in the logic domain and presents several interesting features for the development of medical diagnostic systems. We consider a database of 128 patients, 64 of w ..."
Abstract
 Add to MetaCart
In this paper we describe the application of a new learning tool for the diagnosis of hepatocellular carcinoma. The method adopted operates in the logic domain and presents several interesting features for the development of medical diagnostic systems. We consider a database of 128 patients, 64 of which suffered from hepatocellular carcinoma, while the others suffered from cirrhosis but not from hepatocellular carcinoma. Each patient is described by a number of attributes measured in noninvasive way. The system, after the training, is able to correctly separate the 64 patients affected by cirrhosis from the others 64 affected by hepatocellular carcinoma and is now ready to produce automatic diagnosis for new patients. The hepatocellular carcinoma is one of the most widely spread malignant tumors in the world. The ability to detect the tumor in its early stages in a minimally invasive way is crucial to the treatment of patients with this disease.
1ON THE MINIMUM NUMBER OF LOGICAL CLAUSES INFERRED FROM EXAMPLES
, 1995
"... Scope and Purpose. One of the critical challenges in learning a set of rules (logical clauses) is to derive a very small number of rules, which still satisfy the pertinent requirements. These requirements are derived from positive (successes) and negative examples (failures). The present paper uses ..."
Abstract
 Add to MetaCart
(Show Context)
Scope and Purpose. One of the critical challenges in learning a set of rules (logical clauses) is to derive a very small number of rules, which still satisfy the pertinent requirements. These requirements are derived from positive (successes) and negative examples (failures). The present paper uses some graph theoretic approaches in establishing ways for partitioning large learning problems and also determining bounds on the minimum number of rules derivable from given sets of positive and negative examples. Abstract. Given two sets of positive and negative examples, the inductive inference problem is to infer a small set of logical clauses which appropriately classify the examples. A graph theoretic approach is used to establish a lower limit on the minimum number of required clauses. Furthermore, the findings of this paper reveal methods for partitioning the original data and thus solving efficiently large scale problems.
A feature mining based approach for the classification of text
, 2000
"... documents into disjoint classes ..."
(Show Context)
1 A Heuristic for Mining Association Rules In Polynomial Time*
"... Abstract: Mining association rules from databases has attracted great interest because of its potentially very practical applications. Given a database, the problem of interest is how to mine association rules (which could describe patterns of consumers ’ behaviors) in an efficient and effective wa ..."
Abstract
 Add to MetaCart
Abstract: Mining association rules from databases has attracted great interest because of its potentially very practical applications. Given a database, the problem of interest is how to mine association rules (which could describe patterns of consumers ’ behaviors) in an efficient and effective way. The databases involved in today’s business environment can be very large. Thus, fast and effective algorithms are needed to mine association rules out of large databases. Previous approaches may cause an exponential computing resource consumption. A combinatorial explosion occurs because existing approaches exhaustively mine all the rules. The proposed algorithm takes a previously developed approach, called the Randomized Algorithm 1 (or RA1), and adapts it to mine association rules out of a database in an efficient way. The original RA1 approach was primarily developed for inferring logical clauses (i.e., a Boolean function) from examples. Numerous computational results suggest that the new approach is very promising. Key words: Data mining, association rules, algorithm analysis, the One Clause At a Time (OCAT)
Construction of Deterministic, Consistent, and Stable Explanations from Numerical Data and Prior Domain Knowledge
"... In the explanation problem, training records randomly taken from two populations and, possibly, partial prior domain knowledge are given. Two deterministic explanations for the differences between the two populations are to be constructed, or it must be declared that such explanations most likely ar ..."
Abstract
 Add to MetaCart
(Show Context)
In the explanation problem, training records randomly taken from two populations and, possibly, partial prior domain knowledge are given. Two deterministic explanations for the differences between the two populations are to be constructed, or it must be declared that such explanations most likely are not contained in the data. The explanations must be accurate and in conformance with the given prior domain knowledge. This paper presents a multistep solution algorithm called EXARP for the explanation problem. A key feature is the use of several alternate random processes (ARPs) which attempt to distort data or otherwise disrupt the solution process. Appropriate counteractions prevent or at least mitigate the negative effects of the ARPs.