Biomedical subjects

Michele Piana

Publications and source records attributed to Michele Piana.

3 recordsLinked to original sources

PubMed2006

Modulation of brain and behavioural responses to cognitive visual stimuli with varying signal-to-noise ratios.

OBJECTIVE: To study behavioral and brain responses to variations in signal-to-noise ratio (SNR) of cognitive visual stimuli. METHODS: We presented meaningful words visually, embedded in varying amounts of dynamic noise, and utilized magnetoencephalography (MEG) to measure responses to the words. A multidipole model of the evoked fields was constructed to quantify the strengths and latencies of the neuronal sources at each noise level. The recognition rates of the words were measured in separate behavioral sessions. RESULTS: MEG revealed sequential activation of occipital and occipito-temporal areas (latencies 130-250 and 170-350 ms, respectively) followed by activity in superior temporal cortex (230-640 ms). The strengths and latencies of all identified sources followed functions similar to the SNR of the stimulus. The peak amplitudes and shortest latencies of all sources coincided with the maximum SNR of the stimulus. The occipito-temporal and temporal sources as well as the word recognition rate accurately followed the SNR of the stimulus whereas the early occipital source exhibited a more peaked dependence on the SNR. CONCLUSIONS: Evoked responses expectedly peaked at the maximum SNR of the stimulus. Interestingly, early visual responses showed sharper peaks than longer-latency sources as a function of the noise level. This can be understood as the higher-level processes analyzing the stimuli more holistically and thus being less sensitive to the salience of simple visual features. The similar noise-dependence of the longer-latency sources and the recognition rate provides new evidence for the relevance of these activations in the recognition of written words. SIGNIFICANCE: This study contributes to the understanding of brain activity evoked by degraded stimuli with cognitive content.

Adult↗

PubMed2004

Are loss functions all the same?

In this letter, we investigate the impact of choosing different loss functions from the viewpoint of statistical learning theory. We introduce a convexity assumption, which is met by all loss functions commonly used in the literature, and study how the bound on the estimation error changes with the loss. We also derive a general result on the minimizer of the expected risk for a convex loss function in the case of classification. The main outcome of our analysis is that for classification, the hinge loss appears to be the loss of choice. Other things being equal, the hinge loss leads to a convergence rate practically indistinguishable from the logistic loss rate and much better than the square loss rate. Furthermore, if the hypothesis space is sufficiently rich, the bounds obtained for the hinge loss are not loosened by the thresholding stage.

Learning↗

PubMed2003

New sol-gel synthetic route to phospho-olivines as environmentally friendly cathodes for Li-ion cells.

A new sol-gel synthetic route was developed to prepare single-phase phospho-olivines LiMPO4 (M = Fe or Mn), potential cathode materials for the next generation of Li-ion secondary batteries. Triethyl phosphite was used as phosphate organic precursor, with absolute ethanol as solvent. The sol-gel synthesis ensures homogeneity of the precursors at the nanometric scale and improved reactivity, allowing to obtain very small agglomerates and crystal grain size. Several carbon sources were used in order to improve the electrochemical performances of the samples. Galvanostatic cycling tests of the cathodic materials were made on coin cells with Li metal as anode in order to study the influence of the synthesis and the structural-morphological characteristics on the electrochemical performance of the phosphate/C composite.

Conservation of Energy Resources↗