Damián López

Inference of reversible tree languages

In this paper, we study the notion of k-reversibility and k-testability when regular tree languages are involved. We present an inference algorithm for learning a k-testable tree language that runs in polynomial time with respect to the size of the sample used. We also study the tree language classes in relation to other well known ones, and some properties of these languages are proven.

Error-correcting tree language inference

A new tree language inference algorithm is proposed in this work. This algorithm extends a string language inference algorithm which is based on error correction (ECGI). The algorithm proposed here uses the substructures which have already been taken into account in a tree automaton, modifying the automaton in order to force it to accept the new structures presented in the identification process. The proposed algorithm allows the use of more powerful representation primitives in pattern recognition tasks than the string primitives. It also takes advantage of the thoroughly tested ECGI features used in speech and planar shape recognition tasks.

IgTM: An algorithm to predict transmembrane domains and topology in proteins

BackgroundDue to their role of receptors or transporters, membrane proteins play a key role in many important biological functions. In our work we used Grammatical Inference (GI) to localize transmembrane segments. Our GI process is based specifically on the inference of Even Linear Languages.ResultsWe obtained values close to 80% in both specificity and sensitivity. Six datasets have been used for the experiments, considering different encodings for the input sequences. An encoding that includes the topology changes in the sequence (from inside and outside the membrane to it and vice versa) allowed us to obtain the best results. This software is publicly available at: http://www.dsic.upv.es/users/tlcc/bio/bio.htmlConclusionWe compared our results with other well-known methods, that obtain a slightly better precision. However, this work shows that it is possible to apply Grammatical Inference techniques in an effective way to bioinformatics problems.

Syntactic Pattern Recognition by Error Correcting Analysis on Tree Automata

Although the multidimensional primitives are more powerful than string primitives and there also exist some works concerning distance measure between multidimensional objects, there are no many applications of this kind of languages to syntactic pattern recognition tasks. In this work, multidimensional primitives are used for object modelling in a handwritten digit recognition task under a syntactic approach. Two well-known tree language inference algorithms are considered to build the models, using as error model an algorithm obtaining the editing distance between a tree automaton and a tree; the editing distance algorithm gives the measure needed to complete the classification. The experiments carried out show the good performance of the approach.

Protein motif prediction by grammatical inference

The rapid growth of protein sequence databases is exceeding the capacity of biochemically and structurally characterizing new proteins. Therefore, it is very important the development of tools to locate, within protein sequences, those subsequences with an associated function or specific feature. In our work, we propose a method to predict one of those functional motifs (coiled coil), related with protein interaction. Our approach uses even linear languages inference to obtain a transductor which will be used to label unknown sequences. The experiments carried out show that our method outperforms the results of previous approaches.

ERROR CORRECTING ANALYSIS FOR TREE LANGUAGES

To undertake a syntactic approach to a pattern recognition problem, it is necessary to have good grammatical models as well as good parsing algorithms that allow distorted samples to be classified. There are several methods that obtain, by taking two trees as input, the editing distance between them. In the following work, a polynomial time algorithm which processes the distance between a tree and a tree automaton is presented. This measure can be used in pattern recognition problems as an error model inside a syntactic classifier.

Learning decision trees and tree automata for a syntactic pattern recognition task

Decision trees have been widely used for different tasks in artificial intelligence and data mining. Tree automata have been used in pattern recognition tasks to represent some features of objects to be classified. Here we propose a method that combines both approaches to solve a classical problem in pattern recognition such as Optical Character Recognition. We propose a method which is organized in two stages: (1) we use a grammatical inference technique to represent some structural features of the characters and, (2) we obtain edit distances between characters in order to design a decision tree. The combination of both methods benefits from their individual characteristics and is formulated as a coherent unifying strategy.

Learning automata teams

We prove in this work that, under certain conditions, an algorithm that arbitrarily merges states in the prefix tree acceptor of the sample in a consistent way, converges to the minimum DFA for the target language in the limit. This fact is used to learn automata teams, which use the different automata output by this algorithm to classify the test. Experimental results show that the use of automata teams improve the best known results for this type of algorithms. We also prove that the well known Blue-Fringe EDSM algorithm, which represents the state of art in merging states algorithms, suffices a polynomial characteristic set to converge.

Handwritten Digit Recognition through Inferring Graph Grammars. A First Approach

Graph grammars have been widely used in areas such as program semantics, concurrence and parallelism, or graphic generation. This work is concerned with pattern recognition tasks and how to use a class of graph grammars (web grammars) to modelize the objects in both the learning and recognition phases. First, we give some definitions and basic notation referring to web grammars. We then propose an adaptation of a previous method for learning web grammars and we report some preliminary results on a handwritten digit recognition task together with future work to be carried out.

Editing distances between membrane structures

In this work we propose an efficient solution to calculate the minimum editing distance between membrane structures of arbitrary P systems. We use a new model of tree automata based on multisets of states and symbols linked to the finite control. This new model accepts a set of trees with symmetries between their internal nodes (mirrored trees). Once we have calculated the editing distance between an arbitrary tree and an arbitrary multiset tree automaton, we can translate the classical operations of insertion, deletion and substitution into rule applications of membrane dissolving and membrane creation.