-->
Bioinformatic research: We have developed interest in bioinformatics, trying to apply our methodologies to different tasks, such as the prediction of the disulfide-bonding state of cysteines in proteins and the bacterial protein family identification.
abstract01.pdf
Hybrid HMM/ANN: In spite of the advances accomplished throughout the last decades by a number of research teams, the processing of sequential data is still a challenging and difficult task in several domains. We focused our attention in this field on hybrid methods, which combine HMMs and ANNs within a single architecture.
abstract02.pdf
Neural Networks for Graph Processing: In several applications the information is naturally represented by graphs.This project presents a new neural model, called Relational Neural Network (RNN), that extends recursive neural networks and can be applied on most of the practically useful kinds of graphs.
abstract03.pdf
Local Minima: Abstract not inserted yet!
abstract03.pdf
Neural heuristics We believe that neural information (combined with classical symbolic technics) can outperform symbolic algorithmic heuristics mantaining solution optimality very close to 100%.
abstract03.pdf
Web crow We designed and implemented a software system, called WebCrow, that represents the first solver for Italian crosswords and the first system that tackles a language game using the Web as knowledge base.
abstract03.pdf
Document image analysis: The main topics of our research in document image analysis are related to printed documents. In the field of printed documents, the most difficult task is to deal with documents with complex layout, where the meaning of a field can be deduced only after analyzing its graphical and logical context. Examples of these documents are forms and invoices.
abstract03.pdf
Graph Matching: In this field we propose a general framework for graph matching which is suitable for different problems of pattern recognition. The pattern representation we assume is at the same time highly structured, like for classic syntactic and structural approaches, and of sub--symbolic nature with real--valued features, like for connectionist and statistic approaches.
abstract03.pdf
Neurosoft: Neurosoft-Sebia is an expert system that is designed to classify human serum electrophoresic curves. Sebia produces a device (Capillarys) that exploits capillary electrophoresis to obtain a curve that is related to the distribution of the serum proteins. These curves are visually analyzed by human experts in order to determine anomalous situations that suggest to perform further detailed diagnostic checks. Neurosoft-Sebia has been developed with the aim of assisting the human experts during the analysis. In particular, the system classifies the profiles into two categories: regular or anomalous.
abstract03.pdf
 
 
  credits .: Artificial Intelligence Research Group of Siena :.