1,721,360 research outputs found
Fusion of Classifiers for Protein Fold Recognition
No full textPredicting the three-dimensional structure of a protein from its amino acid sequence is an important problem in bioinformatics and a challenging task for machine learning algorithms. Given (numerical) features, one of the existing machine learning techniques can be then applied to learn and classify proteins represented by these features. We show that combining Fisher's linear classifier and K-Local Hyperplane Distance Nearest Neighbor we obtain an error rate lower than previously published in the literature
Fusion of classifiers for predicting Protein-Protein interactions
No full textPrediction of protein-protein interaction is a difficult and an important problem in biology. In this paper, we describe a very general method for predicting protein-protein interactions. The interaction mining approach is demonstrated by building a learning system based on experimentally validated protein-protein interactions in the human gastric bacterium Helicobacter pylori. We show that combining linear discriminant classifier and cloud points we obtain an error rate lower than previously published in the literature
Over-complete feature generation and feature selection for biometry
No full textIn this paper a novel method for obtaining an appropriate representation of patterns is presented. The information is extracted using an over-complete global feature combination, and then the most useful features are selected by Sequential Forward Floating Selection (SFFS). This new method has been tested in two problems: trained integration of iris and face biometrics; on-line signature verification system based on global information and a one-class classifier (Parzen Window Classifier). To the best of our knowledge, this is the first work that studies and proposes a set of “artificial” features for combining biometric matchers, created starting from the scores of the matchers. We show that a classifier trained on such set of features gains a noticeable performance improvement with respect to fixed fusion rules and other trained fusion methods. Moreover, we show that an on-line signature matcher based on the “artificial” features gains a noticeable performance improvement with respect to a matcher based on the “original” global features
Machine learning multi-classifiers for peptide classification
No full textIn this paper, we study the performance improvement that it is possible to obtain combining classifiers based on different notions (each trained using a different physicochemical property of amino-acids). This multi-classifier has been tested in three problems: HIV-protease; recognition of T-cell epitopes; predictive vaccinology. We propose a multi-classifier that combines a classifier that approaches the problem as a two-class pattern recognition problem and a method based on a one-class classifier. Several classifiers combined with the “sum rule” enables us to obtain an improvement performance over the best results previously published in the literature
Overview of the combination of biometric matchers
No full textBiometric identity verification refers to technologies used to measure human physical or behavioral characteristics, which offer a radical alternative to passports, ID cards, driving licenses or PIN numbers in authentication. Since biometric systems present several limitations in terms of accuracy, universality, distinctiveness, acceptability, methods for combining biometric matchers have attracted increasing attention of researchers with the aim of improving the ability of systems to handle poor quality and incomplete data, achieving scalability to manage huge databases of users, ensuring interoperability, and protecting user privacy against attacks. The combination of biometric systems, also known as "biometric fusion", can be classified into unimodal biometric if it is based on a single biometric trait and multimodal biometric if it uses several biometric traits for person authentication. The main goal of this study is to analyze different techniques of information fusion applied in the biometric field. This paper overviews several systems and architectures related to the combination of biometric systems, both unimodal and multimodal, classifying them according to a given taxonomy. Moreover, we deal with the problem of biometric system evaluation, discussing both performance indicators and existing benchmarks. As a case study about the combination of biometric matchers, we present an experimental comparison of many different approaches of fusion of matchers at score level, carried out on three very different benchmark databases of scores. Our experiments show that the most valuable performance is obtained by mixed approaches, based on the fusion of scores. The source code of all the method implemented for this research is freely available for future comparisons1. After a detailed analysis of pros and cons of several existing approaches for the combination of biometric matchers and after an experimental evaluation of some of them, we draw our conclusion and suggest some future directions of research, hoping that this work could be a useful start point for newer research
When Fingerprints Are Combined with Iris - A Case Study: FVC2004 and CASIA
No full textThis paper presents novel studies on fusion strategies for personal identification using fingerprint and iris biometrics. The purpose of our paper is to investigate whether the integration of iris and fingerprint biometrics can achieve performance that may not be possible using a single biometric technology. Moreover we are interested in evaluating the correlation among the best state of art algorithms for fingerprint verification presented at FVC2004. We show that the fusion among some competitors of FVC2004 permits a drastically reduction of the performance. Particularly interesting is the result obtained by combining the competitors of FVC2004 and an IRIS matcher in terms of EER (the most used parameter in the evaluation of real identification systems), significantly lower than for other approaches. This indicates that the intrinsic error of the system is very low and tends to 0 for some of the tests carried out. The results of this paper confirm that a multimodal biometric can overcome some of the limitations of a single biometric resulting in a substantial performance improvement
Deep learning in polyp segmentation
No full textreservedA. Note identificative
Laureanda Ezeobi Francisca Chidubem
Titolo della tesi: "Reti neurali profonde per segmentazione di polipi"
Corso di laurea in Ingegneria Informatica
Relatore: Prof. Nanni Loris
B. Elementi di contenuto
La finalità è l'applicazione delle reti neurali. In questo caso, ci si avvale di Mask R-CNN per l'individuazione dei polipi presenti nel tratto gastrointestinale. Prima di mostare, l'esperimento e i suoi risultati, sono stati brevemente spiegati alcuni ma importanti concetti teorici riguardante il machine learning e deep learning.
I risultati di instance segmentation eseguiti sul dataset kvasir-SEG sono in parte buoni. Le reti addestrate con solver SGDM riescono abbastanza bene nell'individuazione dei polipi mentre quelle reti addestrate con solver Adam non riescono tanto. L'esperimento è stato svolto con l'aiuto del software Matla
A genetic approach for building different alphabets for peptide and protein classification
Full text linkAbstract Background In this paper, it is proposed an optimization approach for producing reduced alphabets for peptide classification, using a Genetic Algorithm. The classification task is performed by a multi-classifier system where each classifier (Linear or Radial Basis function Support Vector Machines) is trained using features extracted by different reduced alphabets. Each alphabet is constructed by a Genetic Algorithm whose objective function is the maximization of the area under the ROC-curve obtained in several classification problems. Results The new approach has been tested in three peptide classification problems: HIV-protease, recognition of T-cell epitopes and prediction of peptides that bind human leukocyte antigens. The tests demonstrate that the idea of training a pool classifiers by reduced alphabets, created using a Genetic Algorithm, allows an improvement over other state-of-the-art feature extraction methods. Conclusion The validity of the novel strategy for creating reduced alphabets is demonstrated by the performance improvement obtained by the proposed approach with respect to other reduced alphabets-based methods in the tested problems.</p
A Clustering Method for Automatic Biometric Template Selection
No full textThe problem addressed in this paper is the templateselection and update in biometrics based on clustering. Templateselection is a reliable method to reduce the number of templates used in abiometric system to account for variations observed in a person's biometric data. An efficient method based on clustering with automaticselection of the number of clusters is proposed in this work for finding subgroups of similar templates which are used for prototype selection.
Experimental results confirm the advantage of the new method and the importance of adopting a procedure to perform templateselection
- …
