Новый подход к интерпретации весов сверточных сетей при работе с ЭЭГ
the capacity to understand the roles of specific multispectral waveforms identified by CNNs could be very helpful. In this study, we present a novel model visualization-based approach that adapts the traditional CNN architecture to increase interpretability and combines that inherent interpretability with a systematic evaluation of the model via a series of novel explainability methods ... We evaluate our approach within the context of automated sleep stage classification and find that, for the most part, our explainability results are highly consistent with clinical guidelines.
Ellis CA, Miller RL and Calhoun VD (2022) A systematic approach for explaining time and frequency features extracted by convolutional neural networks from raw electroencephalography data. Front. Neuroinform. 16:872035. https://doi.org/10.3389/fninf.2022.872035
the capacity to understand the roles of specific multispectral waveforms identified by CNNs could be very helpful. In this study, we present a novel model visualization-based approach that adapts the traditional CNN architecture to increase interpretability and combines that inherent interpretability with a systematic evaluation of the model via a series of novel explainability methods ... We evaluate our approach within the context of automated sleep stage classification and find that, for the most part, our explainability results are highly consistent with clinical guidelines.
Ellis CA, Miller RL and Calhoun VD (2022) A systematic approach for explaining time and frequency features extracted by convolutional neural networks from raw electroencephalography data. Front. Neuroinform. 16:872035. https://doi.org/10.3389/fninf.2022.872035