In this paper, we present a new method for propositionalization that works in a bottom-up, data-driven manner. It is tailored for biochemical databases, where the examples are 2-D descriptions of chemical compounds. The method generates all frequent fragments (i.e., linearly connected atoms) up to a user-specified length. A preliminary experiment in the domain of carcinogenicity prediction show...

In this paper we present a new shape analysis algorithm. The key distinguishing aspect of our algorithm is that it is completely compositional, bottom-up and non-iterative. We present our algorithm as an inference system for computing Hoare triples summarizing heap manipulating programs. Our inference rules are compositional: Hoare triples for a compound statement are computed from the Hoare tr...

This paper describes a series of experiments aimed at producing a bot tom-up parser that will produce partial parses suitable for use in robust interpretation and still be reasonably efficient. In the course of these experiments, we improved parse times by a factor of 18 over our first a t tempt, ending with a system that was twice as fast as our previous parser, which relied on strong top-down...

Traditional data mining methods based on rough set theory focus on extracting models which are good at classifying unseen objects. If one wants to uncover new knowledge from the data, the model must have a high descriptive quality— it must describe the data set in a clear and concise manner, without sacrificing classification performance. Rough modeling, introduced by Kowalczyk (1998), is an ap...

We propose a credit stress testing method that puts together: (1) a bottom-up corporate default prediction model capable of translating shocks to its input variables into the default likelihood of a target portfolio, and (2) a set of stress-testing regressions which relate the presumed adverse macroeconomic scenarios to shocks to the inputs of the default prediction model. This method can produ...

In this paper we introduce several new improvements to the bottom-up model generation (BUMG) paradigm. Our techniques are based on non-trivial transformations of first-order problems into a certain implicational form, namely rangerestricted clauses. These refine existing transformations to range-restricted form by extending the domain of interpretation with new Skolem terms in a more careful an...