Probabilistic generative models and learning -application to realistic data and brain measurements-

点击翻译按钮获取中文摘要

Problems of estimating sparse source vector fields from the measured data of them were addressed by assuming a generative model. These fields can be regarded as block-sparse vectors, that have larger values at smaller points and whose elements have groupwise (blockwise) structure. (1) A source estimation method robust against forward calculation errors was proposed, and (2) A reconstruction principle of point source in linear inverse problems was proved.