How Does a Computer “See” Gender? Demonstrating Uncertainty in Deep Learning Classification Models Though Data Visualization and Interactive Play
How Does a Computer “See” Gender? Demonstrating Uncertainty in Deep Learning Classification Models Though Data Visualization and Interactive Play
复制标题
计算机如何通过数据可视化和交互游戏来“看到”深度学习分类模型的不确定性？
DOI：
--
复制
发表时间：
2020
期刊：
影响因子：
--
通讯作者：
Peter Bell
中科院分区：
文献类型：
--
作者：
Christopher Baronavski;Peter Bell
皮尤研究中心最近的一项调查发现，大多数美国人认为面部识别技术可以准确地识别个人，并正确确定性别和种族。此外，大多数美国人表示，他们相信执法机构会负责任地使用面部识别技术[3]。然而，在实际应用中，机器学习应用程序容易受到偏差[1][6]的影响，而面部识别和机器视觉技术的实现可能会产生不可预测的、意外的或不准确的结果[9][4]。
A recent Pew Research Center survey found that a majority of Americans believe facial recognition technology can accurately identify individuals, as well as correctly determine gender and race. Additionally, a majority of Americans say that they trust law enforcement agencies to use facial recognition technology responsibly[3]. However, in practical use, machine learning applications are susceptible to biases[1] [6], and implementations of facial recognition and machine vision technologies may yield unpredictable, unexpected, or inaccurate results[9] [4].