Robust Control Methods for On-Line Statistical Learning
<p/> <p>The issue of controlling that data processing in an experiment results not affected by the presence of outliers is relevant for statistical control and learning studies. Learning schemes should thus be tested for their capacity of handling outliers in the observed training set so...
| Main Author: | |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
SpringerOpen
2001-01-01
|
| Series: | EURASIP Journal on Advances in Signal Processing |
| Subjects: | |
| Online Access: | http://dx.doi.org/10.1155/S1110865701000178 |
| _version_ | 1811267067449442304 |
|---|---|
| author | Capobianco Enrico |
| author_facet | Capobianco Enrico |
| author_sort | Capobianco Enrico |
| collection | DOAJ |
| description | <p/> <p>The issue of controlling that data processing in an experiment results not affected by the presence of outliers is relevant for statistical control and learning studies. Learning schemes should thus be tested for their capacity of handling outliers in the observed training set so to achieve reliable estimates with respect to the crucial bias and variance aspects. We describe possible ways of endowing neural networks with statistically robust properties by defining feasible error criteria. It is convenient to cast neural nets in state space representations and apply both Kalman filter and stochastic approximation procedures in order to suggest statistically robustified solutions for on-line learning.</p> |
| first_indexed | 2024-04-12T20:55:46Z |
| format | Article |
| id | doaj.art-5cc08a19eb774cae8700902a55fb5491 |
| institution | Directory Open Access Journal |
| issn | 1687-6172 1687-6180 |
| language | English |
| last_indexed | 2024-04-12T20:55:46Z |
| publishDate | 2001-01-01 |
| publisher | SpringerOpen |
| record_format | Article |
| series | EURASIP Journal on Advances in Signal Processing |
| spelling | doaj.art-5cc08a19eb774cae8700902a55fb54912022-12-22T03:17:00ZengSpringerOpenEURASIP Journal on Advances in Signal Processing1687-61721687-61802001-01-0120012287964Robust Control Methods for On-Line Statistical LearningCapobianco Enrico<p/> <p>The issue of controlling that data processing in an experiment results not affected by the presence of outliers is relevant for statistical control and learning studies. Learning schemes should thus be tested for their capacity of handling outliers in the observed training set so to achieve reliable estimates with respect to the crucial bias and variance aspects. We describe possible ways of endowing neural networks with statistically robust properties by defining feasible error criteria. It is convenient to cast neural nets in state space representations and apply both Kalman filter and stochastic approximation procedures in order to suggest statistically robustified solutions for on-line learning.</p>http://dx.doi.org/10.1155/S1110865701000178artificial learningstatistical control algorithmsrobustness and efficiency of estimatorsmaximum likelihood inference |
| spellingShingle | Capobianco Enrico Robust Control Methods for On-Line Statistical Learning EURASIP Journal on Advances in Signal Processing artificial learning statistical control algorithms robustness and efficiency of estimators maximum likelihood inference |
| title | Robust Control Methods for On-Line Statistical Learning |
| title_full | Robust Control Methods for On-Line Statistical Learning |
| title_fullStr | Robust Control Methods for On-Line Statistical Learning |
| title_full_unstemmed | Robust Control Methods for On-Line Statistical Learning |
| title_short | Robust Control Methods for On-Line Statistical Learning |
| title_sort | robust control methods for on line statistical learning |
| topic | artificial learning statistical control algorithms robustness and efficiency of estimators maximum likelihood inference |
| url | http://dx.doi.org/10.1155/S1110865701000178 |
| work_keys_str_mv | AT capobiancoenrico robustcontrolmethodsforonlinestatisticallearning |