Year: 2021
Type: preprint
Abstract: In this paper we develop a Multilevel Orthogonal Subspace (MOS) Karhunen--Loève feature theory based on stochastic tensor spaces, for the construction of robust machine learning features. Training data are treated as instances of a random field within a relevant Bochner space. Our key observation is that separate machine learning classes can reside predominantly in mostly distinct subspaces. Using the Karhunen-Loève expansion and a hierarchical expansion of the first (nominal) class, a MOS is constructed to detect anomalous signal components, treating the second class as an outlier of the first. The projection coefficients of the input data into these subspaces are then used to train a Machine Learning (ML) classifier. These coefficients become new features from which much clearer separation surfaces can arise for the underlying classes. We test our approach on blood plasma and cerebrospinal fluid datasets (Alzheimer's Disease Neuroimaging Initiative). The tests show significant increases in accuracy by applying them to Support Vector Machines and (Deep) Neural Networks. Furthermore, from non-invasive blood test, high-accuracy results can be obtained for predicting AD stages such as cognitive normal, mild cognitive impairment and dementia. The non-invasive nature of this test is significant since it can potentially avoid other painful tests.
Source: arXiv (Cornell University)
Institution: Boston University
Language: English
Subfield: Artificial Intelligence
Field: Computer Science
Domain: Physical Sciences
Open Access status: green
Funders: National Science Foundation, U.S. Department of Defense, Alzheimer's Association, Bristol-Myers Squibb, Eli Lilly and Company, Biogen, F. Hoffmann-La Roche, Alzheimer's Disease Neuroimaging Initiative, BioClinica, Eisai, H. Lundbeck A/S, IXICO, National Institutes of Health, Genentech, Division of Mathematical Sciences
Awards: G1012222616, DMS/NIGMS 1: Multilevel stochastic orthogonal subspace transformations for robust machine learning with applications to biomedical data and Alzheimer's disease subtyping, G1298807586, G1554142629, G1650819308, G1880935189, G2189010064, G2340207221, G2423387114, G2471646051, G2522063444, G2835601961, G29951154, G3131374986, G3216574877, G3553156159, G3710456697, G3870774570, G3966393351, G43180419, G4501223628, G4800177743, G5207404921, G5654639750, G5974299961, G6210868510, G6371410416, G6412110892, G6719828067, G6792506723, G6899163935, G6948409098, G7268282379, G7298697282, G764699806, G781725486, G8228018387, G8301434335, ATD: Anomaly detection and functional data analysis with applications to threat detection for multimodal satellite data, G8895907669, G8957659042, G9000039018, G9000122335
OpenAlex ID: W3203376656