161.
Improving Theories for Inductive Logic Programming Systems with Ground Reduced Programs
- Akihiro Yamamoto,Fachgebiet Intellektik,Fachbereich Informatik,Technische Hochschule Darmstadt
In this paper we improve the theory of inverse entailment given by Muggleton. The
theory of inverse entailment is a foundation of the Progol system, one of the most
famous ILP systems. We first point out that the theory is incomplete in general.
Secondly we prove that the theory is complete if the background knowledge given
to the system is a ground reduced program, every training example is a ground
unit clause, and the hypothesis space is the set of all definite clauses. The proof is
obtained by showing that every ground reduced logic program is logically equivalent
to the conjunction of all atoms in its least...
162.
Using Multiple Clause Constructors in Inductive Logic Programming for Semantic Parsing
- Raymond J. Mooney
In this paper, we explored a learning approach which combines different learning methods in inductive logic programming (ILP) to allow a learner to produce more expressive hypotheses than that of each individual learner. Such a learning approach may be useful when the performance of the task depends on solving a large amount of classification problems and each has its own characteristics which may or may not fit a particular learning method. The task of semantic parser acquisition in two different domains was attempted and preliminary results demonstrated that such an approach is promising.
163.
Pruning Nodes in the Alpha-Beta Method Using Inductive Logic Programming
- Nobuhiro Inuzuka,Hayato Fujimoto,Tomofumi Nakano,Hidenori Itoh
This paper reports a preliminary research of
application of inductive logic programming
for pruning nodes in the alpha-beta gametree
search method, which finds an appropriate
move by looking game-trees in a limited
depth ahead. The alpha-beta method
reduces the number of nodes by keeping and
updating lower and upper bounds of static
evaluation. Pruning effect depends on an accidental
order of nodes visited, because we
can expect large pruning after we have large
update. This paper proposes a method to
learn rules to sort nodes to yield effective
pruning, by using inductive logic programming
framework. The method induces a logic
program of a binary relation among nodes,
and sorts nodes based on the relation. We
inspected the...
164.
An Inductive Logic Programming Framework to Learn a Concept from Ambiguous Examples
- Dominique Bouthinon,Henry Soldano
. We address a learning problem with the following peculiarity : we search for
characteristic features common to a learning set of objects related to a target concept. In
particular we approach the cases where descriptions of objects are ambiguous : they represent
several incompatible realities. Ambiguity arises because each description only contains
indirect information from which assumptions can be derived about the object. We suppose here
that a set of constraints allows the identification of "coherent" sub-descriptions inside each
object.
We formally study this problem, using an Inductive Logic Programming framework close to
characteristic induction from interpretations. In particular, we exhibit conditions which allow a
pruned search of...
165.
An Inductive Logic Programming Framework to Learn a Concept from Ambiguous Examples
- Dominique Bouthinon,Henry Soldano
. We address a learning problem with the following peculiarity : we search for
characteristic features common to a learning set of objects related to a target concept. In
particular we approach the cases where descriptions of objects are ambiguous : they represent
several incompatible realities. Ambiguity arises because each description only contains
indirect information from which assumptions can be derived about the object. We suppose here
that a set of constraints allows the identification of "coherent" sub-descriptions inside each
object.
We formally study this problem, using an Inductive Logic Programming framework close to
characteristic induction from interpretations. In particular, we exhibit conditions which allow a
pruned search of...
166.
An Inductive Logic Programming Framework to Learn a Concept from Ambiguous Examples
- Dominique Bouthinon,Henry Soldano
. We address a learning problem with the following peculiarity : we search for
characteristic features common to a learning set of objects related to a target concept. In
particular we approach the cases where descriptions of objects are ambiguous : they represent
several incompatible realities. Ambiguity arises because each description only contains
indirect information from which assumptions can be derived about the object. We suppose here
that a set of constraints allows the identification of "coherent" sub-descriptions inside each
object.
We formally study this problem, using an Inductive Logic Programming framework close to
characteristic induction from interpretations. In particular, we exhibit conditions which allow a
pruned search of...
167.
The Application of Inductive Logic Programming to Finite Element Mesh Design
- Bojan Dolsak And,Stephen Muggleton
Finite element methods are used extensively by engineers and modelling scientists to analyse stresses in physical structures. These structures are represented quantitatively as finite collections of elements. The deformation of each element is computed using linear algebraic equations. In order to design a numerical model of a physical structure it is necessary to decide the appropriate resolution for modelling each component part. Considerable expertise is required in choosing these resolution values. Too fine a mesh leads to unnecessary computational overheads when executing the model. Too coarse a mesh produces intolerable approximation errors. In this paper we demonstrate that rules for...
168.
Inductive Logic Programming: issues, results and the challenge of Learning Language in Logic
- Stephen Muggleton
Inductive Logic Programming (ILP) is the area of AI which deals with
the induction of hypothesised predicate definitions from examples and
background knowledge. Logic programs are used as a single representation
for examples, background knowledge and hypotheses. ILP is differentiated
from most other forms of Machine Learning (ML) both by its use of an
expressive representation language and its ability to make use of logically
encoded background knowledge. This has allowed successful applications
of ILP in areas such as molecular biology and natural language which
both have rich sources of background knowledge and both benefit from
the use of an expressive concept representation languages. For instance,
the ILP system Progol...
169.
A Comparative Study of Inductive Logic Programming Methods for Software Fault Prediction
- William W. Cohen,Prem Devanbu
We evaluate inductive logic programming
(ILP) methods for predicting fault density in
C++ classes. In this problem, each training
example is a C++ class definition, represented
as a calling tree, and labeled as "positive
" iff faults (i.e., errors) were discovered
in its implementation. We compare two ILP
systems, FOIL and FLIPPER, and explore
the reasons for their differing performance,
using both natural and artificial data. We
then propose two extensions to FLIPPER:
a user-directed bias towards easy-to-evaluate
clauses, and an extension that allows FLIPPER
to learn "counting clauses". Counting
clauses augment logic programs with a
variation of the "number restrictions" used
in description logics, and significantly improve
performance on this problem when
prior knowledge is used.
170.
An Inductive Logic Programming Approach to the Classification of Phases in Business Cycles
- Katharina Morik
A workbench for knowledge acquisition and data analysis is presented and its use for the classification of business cycles is investigated. Inductive Logic Programming (ILP) allows to model relations between intervals, e.g. time or value intervals. Moreover, the user of the workbench is supported in inspecting the learned rules, not only with respect to their coverage, accuracy, and redundancy, but also regarding consistency (i.e., logical contradictions). The application of ILP requires pre-processing in order to establish time and value intervals. To this end, top-down induction of decision trees is used. This paper describes the workbench MOBAL, its learning algorithm RDT,...
171.
Relational Data Mining with Inductive Logic Programming for Link Discovery
- Raymond J. Mooney,Prem Melville,Jude Shavlik,Ines De Castro Dutra,Vtor Santos Costa
Link discovery (LD) is an important task in data mining
for counter-terrorism and is the focus of DARPA's Evidence
Extraction and Link Discovery (EELD) research program.
Link discovery concerns the identification of complex relational
patterns that indicate potentially threatening activities
in large amounts of relational data. Most data-mining
methods assume data is in the form of a feature-vector (a
single relational table) and cannot handle multi-relational
data. Inductive logic programming is a form of relational
data mining that discovers rules in first-order logic from
multi-relational data. This paper discusses the application
of ILP to learning patterns for link discovery.
172.
Relational Data Mining with Inductive Logic Programming for Link Discovery
- Raymond J. Mooney,Prem Melville,Jude Shavlik,Ines De Castro Dutra,Vtor Santos Costa
Link discovery (LD) is an important task in data mining
for counter-terrorism and is the focus of DARPA's Evidence
Extraction and Link Discovery (EELD) research program.
Link discovery concerns the identification of complex relational
patterns that indicate potentially threatening activities
in large amounts of relational data. Most data-mining
methods assume data is in the form of a feature-vector (a
single relational table) and cannot handle multi-relational
data. Inductive logic programming is a form of relational
data mining that discovers rules in first-order logic from
multi-relational data. This paper discusses the application
of ILP to learning patterns for link discovery.
173.
Discovering Rules to Design Newspapers: an Inductive Constraint Logic Programming Approach
- Marc Bernard,Francois Jacquenet
Inductive Logic Programming (ILP) combines both Machine Learning and Logic
Programming techniques. ILP uses first-order predicate logic restricted to Horn clauses
as an underlying language. Thus, programs induced by an ILP system inherit the
classical limitations of PROLOG programs. Constraint Logic Programming avoids some
of the limitations of Logic Programming and so ILP aims to induce programs that
employ this paradigm. Current ILP systems which induce constrained logic programs
extend systems based on the normal semantics of ILP. In this paper we introduce ICLog,
a new system which induces constrained logic programs and relies on an extension
of a non-monotonic semantics-based system. We then present an application of...
174.
Implementation issues in Inductive Logic
- Robert Kolter
We propose several algorithms for ecient Testing of logical Implication
in the case of ground objects. Because the problem of Testing a
set of propositional formulas for (un)satis
ability is NP-complete there's
strong evidence that there exist examples for which every algorithm which
solves the problem of testing for (un)satis
ability has a runtime that is
exponential in the length of the input. So will have our algorithms. We
will therefore point out classes of logic programs for which our algorithms
have a lower runtime.
175.
An Initial Experiment into Stereochemistry-based Drug Design using Inductive Logic Programming
- Stephen Muggleton,Ashwin Srinivasan
Previous applications of Inductive Logic Programming to
drug design have not addressed stereochemistry, or the three-dimensional
aspects of molecules. While some success is possible without consideration
of stereochemistry, researchers within the pharmaceutical industry
consider stereochemistry to be central to most drug design problems.
This paper reports on an experimental application of the ILP system
P-Progol to stereochemistry-based drug design. The experiment tests
whether P-Progol can identify the structure responsible for the activity
of ACE (angiotensin-converting enzyme) inhibitors from 28 positive
examples, that is, from 28 molecules that display the activity of ACE
inhibition. ACE inhibitors are a widely-used form of medication for the
treatment of hypertension. It should be stressed that...
176.
Automatically Exploring Hypotheses about Fault Prediction: a Comparative Study of Inductive Logic Programming Methods
- William W. Cohen,Premkumar T. Devanbu
We evaluate a class of learning algorithms known as inductive logic programming (ILP) methods on the task of predicting fault occurrence in C++ classes. Using these methods, a large space of possible hypotheses is searched in an automated fashion; further, the hypotheses are based directly on an abstract logical representation of the software, rather than on manually proposed numerical metrics that predict fault density.
177.
Inductive Logic Programming and Case-Based Reasoning for Nuclear Power Plants Monitoring
- Claire Nicolini,Francois Jacquenet,Marc Bernard
In this paper, we present the SECAPI system, a
decision-aid tool in case of malfunction of nuclear
power plants from the French nuclear power board
(CEA). In case of malfunction of the plant, we do
not have any formal specification of the running process.
So it is not possible to implement some classical
techniques such as Expert Systems. Machine learning
techniques integrated in SECAPI, combine concepts
of Case-Based Reasoning (CBR) and Inductive Logic
Programming (ILP). The use of such a tool, allows to
store past experiments which will be used later to explain
new malfunctions and to induce rules that model
the most frequent malfunctions.
Keywords : Inductive Logic Programming, CaseBased
Reasoning, Machine Learning,...
178.
Advantages of Decision Lists and Implicit Negatives in Inductive Logic Programming
- Mary Elaine Califf,Raymond J. Mooney
This paper demonstrates the capabilities of Foidl, an inductive logic programming
(ILP) system whose distinguishing characteristics are the ability to produce first-order
decision lists, the use of an output completeness assumption to provide implicit negative
examples, and the use of intensional background knowledge. The development of Foidl
was originally motivated by the problem of learning to generate the past tense of English
verbs; however, this paper demonstrates its superior performance on two different
sets of benchmark ILP problems. Tests on the finite element mesh design problem
show that Foidl's decision lists enable it to produce better results than all other ILP
systems whose results on this problem have...
179.
A Comparison of Two Methods Employing Inductive Logic Programming for Corpus-based Parser Construction
- John M. Zelle,Raymond J. Mooney
This paper presents results from recent experiments
with Chill, a corpus-based parser acquisition
system. Chill treats grammar acquisition
as the learning of search-control rules within a
logic program. Unlike many current corpus-based
approaches that use propositional or probabilistic
learning algorithms, Chill uses techniques from
inductive logic programming (ILP) to learn relational
representations. The reported experiments
compare Chill's performance to that of a more
naive application of ILP to parser acquisition. The
results show that ILP techniques, as employed in
Chill, are a viable alternative to propositional
methods and that the control-rule framework is
fundamental to Chill's success.
180.
An Experimental Comparison of Genetic Programming and Inductive Logic Programming on Learning Recursive List Functions
- Mary Elaine Califf,Raymond J. Mooney
This paper experimentally compares three approaches to program
induction: inductive logic programming (ILP), genetic programming (GP),
and genetic logic programming (GLP) (a variant of GP for inducing Prolog
programs). Each of these methods was used to induce four simple,
recursive, list-manipulation functions. The results indicate that ILP is
the most likely to induce a correct program from small sets of random
examples, while GP is generally less accurate. GLP performs the worst,
and is rarely able to induce a correct program. Interpretations of these
results in terms of differences in search methods and inductive biases
are presented.
Keywords: Genetic Programming, Inductive Logic Programming, Empirical
Comparison
This paper will also be submitted to...
