Giovanni Soda

Local feedback multilayered networks

In this paper, we investigate the capabilities of local feedback multilayered networks, a particular class of recurrent networks, in which feedback connections are only allowed from neurons to themselves. In this class, learning can be accomplished by an algorithm that is local in both space and time. We describe the limits and properties of these networks and give some insights on their use for solving practical problems.

Artificial neural networks for document analysis and recognition

Artificial neural networks have been extensively applied to document analysis and recognition. Most efforts have been devoted to the recognition of isolated handwritten and printed characters with widely recognized successful results. However, many other document processing tasks, like preprocessing, layout analysis, character segmentation, word recognition, and signature verification, have been effectively faced with very promising results. This paper surveys the most significant problems in the area of offline document image processing, where connectionist-based approaches have been applied. Similarities and differences between approaches belonging to different categories are discussed. A particular emphasis is given on the crucial role of prior knowledge for the conception of both appropriate architectures and learning algorithms. Finally, the paper provides a critical analysts on the reviewed approaches and depicts the most promising research guidelines in the field. In particular, a second generation of connectionist-based models are foreseen which are based on appropriate graphical representations of the learning environment.

Unified integration of explicit knowledge and learning by example in recurrent networks

Proposes a novel unified approach for integrating explicit knowledge and learning by example in recurrent networks. The explicit knowledge is represented by automaton rules, which are directly injected into the connections of a network. This can be accomplished by using a technique based on linear programming, instead of learning from random initial weights. Learning is conceived as a refinement process and is mainly responsible for uncertain information management. We present preliminary results for problems of automatic speech recognition. >

INFORMys: a flexible invoice-like form-reader system

We describe a flexible form-reader system capable of extracting textual information from accounting documents, like invoices and bills of service companies. In this kind of document, the extraction of some information fields cannot take place without having detected the corresponding instruction fields, which are only constrained to range in given domains. We propose modeling the documents layout by means of attributed relational graphs, which turn out to be very effective for form registration, as well as for performing a focused search for instruction fields. This search is carried out by means of a hybrid model, where proper algorithms, based on morphological operations and connected components, are integrated with connectionist models. Experimental results are given in order to assess the actual performance of the system.

Representation of finite state automata in recurrent radial basis function networks

In this paper, we propose some techniques for injecting finite state automata into Recurrent Radial Basis Function networks (R2BF). When providing proper hints and constraining the weight space properly, we show that these networks behave as automata. A technique is suggested for forcing the learning process to develop automata representations that is based on adding a proper penalty function to the ordinary cost. Successful experimental results are shown for inductive inference of regular grammars.

Logo Recognition by Recursive Neural Networks

In this paper we propose recognizing logo images by using an adaptive model referred to as recursive artificial neural network. At first, logo images are converted into a structured representation based on contour trees. Recursive neural networks are then learnt using the contourtrees as inputs to the neural nets. On the other hand, the contour-tree is constructed by associating a node with each exterior or interior contour extracted from the logo instance. Nodes in the tree are labeled by a feature vector, which describes the contour by means of its perimeter, surrounded area, and a synthetic representation of its curvature plot. The contour-tree representation contains the topological structured information of logo and continuous values pertaining to each contour node. Hence symbolic and sub-symbolic information coexist in the contour-tree representation of logo image. Experimental results are reported on 40 real logos distorted with artificial noise and performance of recursive neural network is compared with another two types of neural approaches.

Towards Incremental Parsing of Natural Language Using Recursive Neural Networks

In this paper we develop novel algorithmic ideas for building a natural language parser grounded upon the hypothesis of incrementality. Although widely accepted and experimentally supported under a cognitive perspective as a model of the human parser, the incrementality assumption has never been exploited for building automatic parsers of unconstrained real texts. The essentials of the hypothesis are that words are processed in a left-to-right fashion, and the syntactic structure is kept totally connected at each step.Our proposal relies on a machine learning technique for predicting the correctness of partial syntactic structures that are built during the parsing process. A recursive neural network architecture is employed for computing predictions after a training phase on examples drawn from a corpus of parsed sentences, the Penn Treebank. Our results indicate the viability of the approach and lay out the premises for a novel generation of algorithms for natural language processing which more closely model human parsing. These algorithms may prove very useful in the development of efficient parsers.

Font adaptive word indexing of modern printed documents

We propose an approach for the word-level indexing of modern printed documents which are difficult to recognize using current OCR engines. By means of word-level indexing, it is possible to retrieve the position of words in a document, enabling queries involving proximity of terms. Web search engines implement this kind of indexing, allowing users to retrieve Web pages on the basis of their textual content. Nowadays, digital libraries hold collections of digitized documents that can be retrieved either by browsing the document images or relying on appropriate metadata assembled by domain experts. Word indexing tools would therefore increase the access to these collections. The proposed system is designed to index homogeneous document collections by automatically adapting to different languages and font styles without relying on OCR engines for character recognition. The approach is based on three main ideas: the use of self organizing maps (SOM) to perform unsupervised character clustering, the definition of one suitable vector-based word representation whose size depends on the word aspect-ratio, and the run-time alignment of the query word with indexed words to deal with broken and touching characters. The most appropriate applications are for processing modern printed documents (17th to 19th centuries) where current OCR engines are less accurate. Our experimental analysis addresses six data sets containing documents ranging from books of the 17th century to contemporary journals

Hidden Markov Models for Text Categorization in Multi-Page Documents

In the traditional setting, text categorization is formulated as a concept learning problem where each instance is a single isolated document. However, this perspective is not appropriate in the case of many digital libraries that offer as contents scanned and optically read books or magazines. In this paper, we propose a more general formulation of text categorization, allowing documents to be organized as sequences of pages. We introduce a novel hybrid system specifically designed for multi-page text documents. The architecture relies on hidden Markov models whose emissions are bag-of-words resulting from a multinomial word event model, as in the generative portion of the Naive Bayes classifier. The rationale behind our proposal is that taking into account contextual information provided by the whole page sequence can help disambiguation and improves single page classification accuracy. Our results on two datasets of scanned journals from the Making of America collection confirm the importance of using whole page sequences. The empirical evaluation indicates that the error rate (as obtained by running the Naive Bayes classifier on isolated pages) can be significantly reduced if contextual information is incorporated.

Automatic document classification and indexing in high-volume applications

Abstract. In this paper a system for analysis and automatic indexing of imaged documents for high-volume applications is described. This system, named STRETCH (STorage and RETrieval by Content of imaged documents), is based on an Archiving and Retrieval Engine, which overcomes the bottleneck of document profiling bypassing some limitations of existing pre-defined indexing schemes. The engine exploits a structured document representation and can activate appropriate methods to characterise and automatically index heterogeneous documents with variable layout. The originality of STRETCH lies principally in the possibility for unskilled users to define the indexes relevant to the document domains of their interest by simply presenting visual examples and applying reliable automatic information extraction methods (document classification, flexible reading strategies) to index the documents automatically, thus creating archives as desired. STRETCH offers ease of use and application programming and the ability to dynamically adapt to new types of documents. The system has been tested in two applications in particular, one concerning passive invoices and the other bank documents. In these applications, several classes of documents are involved. The indexing strategy first automatically classifies the document, thus avoiding pre-sorting, then locates and reads the information pertaining to the specific document class. Experimental results are encouraging overall; in particular, document classification results fulfill the requirements of high-volume application. Integration into production lines is under execution.