In several biomedical and bioinformatics applications, one is faced with regression problems that can be stated as Statistical Learning problems. One example is given by the brain activity interpretation through the analysis of functional Magnetic Resonance Imaging (fMRI) data. In recent years it has been shown that the general Statistical Learning problem can be restated as a linear inverse problem (see, e.g., Cucker, Smale, Bull. AMS 2001; De Vito et al., JMLR, 2005; Evgeniou et al., Adv. Comp. Math. 2000). Hence, new algorithms have been proposed to solve this inverse problem in the context of reproducing kernel Hilbert spaces (De Vito et al., Tech. Rep. 2006; Yao et al., Constr. Approx., 2007). These new proposals involve a numerical approach which differs from the ones concerning the classical machine learning techniques and which seems mainly suitable for the just described classes of problems; thus, it is worth to explore how effectively these new algorithms perform when compared with well-known, standard machine learning approaches. The paper will then deal with an experimentation of the new methods on a real-world experiment, as well as with a performance comparison with the "support vector machines" (SVMs) technique.
On recent Machine Learning algorithms for brain activity interpretation
ZANGHIRATI, Gaetano
2007
Abstract
In several biomedical and bioinformatics applications, one is faced with regression problems that can be stated as Statistical Learning problems. One example is given by the brain activity interpretation through the analysis of functional Magnetic Resonance Imaging (fMRI) data. In recent years it has been shown that the general Statistical Learning problem can be restated as a linear inverse problem (see, e.g., Cucker, Smale, Bull. AMS 2001; De Vito et al., JMLR, 2005; Evgeniou et al., Adv. Comp. Math. 2000). Hence, new algorithms have been proposed to solve this inverse problem in the context of reproducing kernel Hilbert spaces (De Vito et al., Tech. Rep. 2006; Yao et al., Constr. Approx., 2007). These new proposals involve a numerical approach which differs from the ones concerning the classical machine learning techniques and which seems mainly suitable for the just described classes of problems; thus, it is worth to explore how effectively these new algorithms perform when compared with well-known, standard machine learning approaches. The paper will then deal with an experimentation of the new methods on a real-world experiment, as well as with a performance comparison with the "support vector machines" (SVMs) technique.I documenti in SFERA sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.
