Browsing by Author "Di Febo, D."

Now showing 1 - 5 of 5
  • No Thumbnail Available
    ItemMetadata only
    Challenges in developing a multidimensional feature Selective Validation implementation.
    (IEEE, 2010) Archambeault, B.; Duffy, A. P.; Sasse, Hugh G.; Li, X. K.; Scase, M. O.; Shafiullah, M.; Orlandi, A.; Di Febo, D.
  • Thumbnail Image
    ItemOpen Access
    Comparison of Data with Multiple Degrees of Freedom Utilizing the Feature Selective Validation (FSV) Method
    (IEEE, 2016-04-27) Zhang, Gang; Duffy, A. P.; Orlandi, A.; Di Febo, D.; Wang, Lixin; Sasse, Hugh G.
    The feature selective validation method has been shown to provide results that are in broad agreement with the visual assessment of a group of engineers for line, 1-D, data. An implementation using 2-D Fourier transforms and derivatives have been available for some years, but verification of the performance has been difficult to obtain. Further, that approach does not naturally scale well for 3-D and higher degrees of freedom, particularly if there are sizable differences in the number of points in the different directions. This paper describes an approach based on repeated 1-D FSV analyses that overcomes those challenges. The ability of the 2-D case to mirror user perceptions is demonstrated using the LIVE database. Its extension to n-dimensions is also described and includes a suggestion for weighting the algorithm based on the number of data points in a given “direction.”
  • No Thumbnail Available
    ItemOpen Access
    Down-sampled and Under-sampled Data sets in Feature Selective Validation (FSV)
    (IEEE, 2014-06-09) Zhang, Gang; Wang, Lixin; Duffy, A. P.; Sasse, Hugh G.; Di Febo, D.; Orlandi, A.; Aniserowicz, Karol
    Feature Selective Validation (FSV) is a heuristic method for quantifying the (dis)similarity of two data sets. The computational burden of obtaining the FSV values might be unnecessarily high if data sets with large numbers of points are used. While this may not be an important issue per se it is an important issue for future developments in FSV such as real-time processing or where multi-dimensional FSV is needed. Coupled with the issue of data set size, is the issue of data sets having ‘missing’ values. This may come about because of a practical difficulty or because of noise or other confounding factors making some data points unreliable. These issues relate to the question “what is the effect on FSV quantification of reducing or removing data points from a comparison – i.e. down- or under-sampling data?” This paper uses three strategies to achieve this from known data sets. This paper demonstrates, through a representative sample of 16 pairs of data sets, that FSV is robust to changes providing a minimum data set size of approximately 200 points is maintained. It is robust also for up to approximately 10% ‘missing’ data, providing this does not result in a continuous region of missed data.
  • No Thumbnail Available
    ItemMetadata only
    Investigating Confidence Histograms and Classification in FSV: Part I. Fuzzy FSV
    (IEEE, 2013) Di Febo, D.; de Paulis, F.; Orlandi, A.; Zhang, G.; Sasse, Hugh G.; Duffy, A. P.; Wang, L.; Archambeault, B.
    One important aspect of the feature selective validation (FSV) method is that it classifies comparison data into a number of natural-language categories. This allows comparison data generated by FSV to be compared with equivalent “visual” comparisons obtained using the visual rating scale. Previous research has shown a close relationship between visual assessment and FSV generated data using the resulting confidence histograms. In all cases, the category membership functions are “crisp”: that is data on the FSV value axis fall distinctly into one category. An important open question in FSV-based research, and for validation techniques generally, is whether allowed variability in these crisp category membership functions could further improve agreement with the visual assessment. A similar and related question is how robust is FSV to variation in the categorization algorithm. This paper and its associated “part II” present research aimed at developing a better understanding of the categorization of both visual and FSV data using nonsquare or variable boundary category membership functions. This first paper investigates the level of improvement to be expected by applying fuzzy logic to location of the category boundaries. The result is limited improvement to FSV, showing that FSV categorization is actually robust to variations in category boundaries.
  • No Thumbnail Available
    ItemMetadata only
    Investigating Confidence Histograms and Classification in FSV: Part II-Float FSV
    (IEEE, 2013) Di Febo, D.; Archambeault, B.; Zhang, G.; Sasse, Hugh G.; Duffy, A. P.; Wang, L.; de Paulis, F.; Orlandi, A.
    One important aspect of the feature selective validation (FSV) method is that it classifies comparison data into a number of natural-language categories. This allows comparison data generated by FSV to be compared with equivalent “visual” comparisons obtained using the visual rating scale. Previous research has shown a close relationship between visual assessment and FSV generated data using the resulting confidence histograms. In all cases, the category membership functions are “crisp”: that is data on the FSV-value axis falls distinctly into one category. The companion paper to this Investigating Confidence histograms and Classification in FSV: Part I. Fuzzy FSV investigated whether allowing probabilistic membership of categories could improve the comparison between FSV and visual assessment. That paper showed that such an approach produced limited improvement and, as a consequence, showed that FSV confidence histograms are robust to flexibility in category boundaries. This paper investigates the effect of redefining some, but not all, category boundaries based around the mode category. This “float” approach does show some improvement in the comparison between FSV and visual assessment.