Illuminating the Black Box: Interpreting Deep Neural Network Models for Psychiatric Research
Psychiatric research is often confronted with complex abstractions and dynamics that are not readily accessible or well-defined to our perception and measurements, making data-driven methods an appealing approach. Deep neural networks (DNNs) are capable of automatically learning abstractions in the...
| Main Author: | |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
Frontiers Media S.A.
2020-10-01
|
| Series: | Frontiers in Psychiatry |
| Subjects: | |
| Online Access: | https://www.frontiersin.org/articles/10.3389/fpsyt.2020.551299/full |
| _version_ | 1818890381912178688 |
|---|---|
| author | Yi-han Sheu Yi-han Sheu Yi-han Sheu |
| author_facet | Yi-han Sheu Yi-han Sheu Yi-han Sheu |
| author_sort | Yi-han Sheu |
| collection | DOAJ |
| description | Psychiatric research is often confronted with complex abstractions and dynamics that are not readily accessible or well-defined to our perception and measurements, making data-driven methods an appealing approach. Deep neural networks (DNNs) are capable of automatically learning abstractions in the data that can be entirely novel and have demonstrated superior performance over classical machine learning models across a range of tasks and, therefore, serve as a promising tool for making new discoveries in psychiatry. A key concern for the wider application of DNNs is their reputation as a “black box” approach—i.e., they are said to lack transparency or interpretability of how input data are transformed to model outputs. In fact, several existing and emerging tools are providing improvements in interpretability. However, most reviews of interpretability for DNNs focus on theoretical and/or engineering perspectives. This article reviews approaches to DNN interpretability issues that may be relevant to their application in psychiatric research and practice. It describes a framework for understanding these methods, reviews the conceptual basis of specific methods and their potential limitations, and discusses prospects for their implementation and future directions. |
| first_indexed | 2024-12-19T17:24:01Z |
| format | Article |
| id | doaj.art-ef45c16018bd4ff7bd4c2e2187671531 |
| institution | Directory Open Access Journal |
| issn | 1664-0640 |
| language | English |
| last_indexed | 2024-12-19T17:24:01Z |
| publishDate | 2020-10-01 |
| publisher | Frontiers Media S.A. |
| record_format | Article |
| series | Frontiers in Psychiatry |
| spelling | doaj.art-ef45c16018bd4ff7bd4c2e21876715312022-12-21T20:12:36ZengFrontiers Media S.A.Frontiers in Psychiatry1664-06402020-10-011110.3389/fpsyt.2020.551299551299Illuminating the Black Box: Interpreting Deep Neural Network Models for Psychiatric ResearchYi-han Sheu0Yi-han Sheu1Yi-han Sheu2Psychiatric Neurodevelopmental and Genetics Unit, Department of Psychiatry, Massachusetts General Hospital, Boston, MA, United StatesDepartment of Psychiatry, Harvard Medical School, Boston, MA, United StatesThe Stanley Center, Broad Institute of Harvard and Massachusetts Institute of Technology (MIT), Cambridge, MA, United StatesPsychiatric research is often confronted with complex abstractions and dynamics that are not readily accessible or well-defined to our perception and measurements, making data-driven methods an appealing approach. Deep neural networks (DNNs) are capable of automatically learning abstractions in the data that can be entirely novel and have demonstrated superior performance over classical machine learning models across a range of tasks and, therefore, serve as a promising tool for making new discoveries in psychiatry. A key concern for the wider application of DNNs is their reputation as a “black box” approach—i.e., they are said to lack transparency or interpretability of how input data are transformed to model outputs. In fact, several existing and emerging tools are providing improvements in interpretability. However, most reviews of interpretability for DNNs focus on theoretical and/or engineering perspectives. This article reviews approaches to DNN interpretability issues that may be relevant to their application in psychiatric research and practice. It describes a framework for understanding these methods, reviews the conceptual basis of specific methods and their potential limitations, and discusses prospects for their implementation and future directions.https://www.frontiersin.org/articles/10.3389/fpsyt.2020.551299/fullmodel interpretabilityexplainable AIdeep learningdeep neural networksmachine learningpsychiatry |
| spellingShingle | Yi-han Sheu Yi-han Sheu Yi-han Sheu Illuminating the Black Box: Interpreting Deep Neural Network Models for Psychiatric Research Frontiers in Psychiatry model interpretability explainable AI deep learning deep neural networks machine learning psychiatry |
| title | Illuminating the Black Box: Interpreting Deep Neural Network Models for Psychiatric Research |
| title_full | Illuminating the Black Box: Interpreting Deep Neural Network Models for Psychiatric Research |
| title_fullStr | Illuminating the Black Box: Interpreting Deep Neural Network Models for Psychiatric Research |
| title_full_unstemmed | Illuminating the Black Box: Interpreting Deep Neural Network Models for Psychiatric Research |
| title_short | Illuminating the Black Box: Interpreting Deep Neural Network Models for Psychiatric Research |
| title_sort | illuminating the black box interpreting deep neural network models for psychiatric research |
| topic | model interpretability explainable AI deep learning deep neural networks machine learning psychiatry |
| url | https://www.frontiersin.org/articles/10.3389/fpsyt.2020.551299/full |
| work_keys_str_mv | AT yihansheu illuminatingtheblackboxinterpretingdeepneuralnetworkmodelsforpsychiatricresearch AT yihansheu illuminatingtheblackboxinterpretingdeepneuralnetworkmodelsforpsychiatricresearch AT yihansheu illuminatingtheblackboxinterpretingdeepneuralnetworkmodelsforpsychiatricresearch |