Search

Deep neural networks (DNNs) are an indispensable machine learning tool despite the difficulty of diagnosing what aspects of a model’s input drive its decisions. In countless real-world domains, from legislation and law enforcement to healthcare, such diagnosis is essential to ensure that DNN decisions are driven by aspects appropriate in the context of its use. The development of methods and studies enabling the explanation of a DNN’s decisions has thus blossomed into an active and broad area of research. The field’s complexity is exacerbated by competing definitions of what it means “to explain” the actions of a DNN and to evaluate an approach’s “ability to explain”. This article offers a field guide to explore the space of explainable deep learning for those in the AI/ML field who are uninitiated. The field guide: i) Introduces three simple dimensions defining the space of foundational methods that contribute to explainable deep learning, ii) discusses the evaluations for model explanations, iii) places explainability in the context of other related deep learning research areas, and iv) discusses user-oriented explanation design and future directions. We hope the guide is seen as a starting point for those embarking on this research field. [ABSTRACT FROM AUTHOR]

Multivariate pattern analysis is a technique that allows the decoding of conceptual information such as the semantic category of a perceived object from neuroimaging data. Impressive single-trial classification results have been reported in studies that used fMRI. Here, we investigate the possibility to identify conceptual representations from event-related EEG based on the presentation of an object in different modalities: its spoken name, its visual representation and its written name. We used Bayesian logistic regression with a multivariate Laplace prior for classification. Marked differences in classification performance were observed for the tested modalities. Highest accuracies (89% correctly classified trials) were attained when classifying object drawings. In auditory and orthographical modalities, results were lower though still significant for some subjects. The employed classification method allowed for a precise temporal localization of the features that contributed to the performance of the classifier for three modalities. These findings could help to further understand the mechanisms underlying conceptual representations. The study also provides a first step towards the use of concept decoding in the context of real-time brain-computer interface applications. [ABSTRACT FROM AUTHOR]