Local Interpretable Model-Agnostic Explanations (LIME) and GDPR.
Articles raising awareness of the implications GDPR will have on data science and machine learning are published from multiple sources. How can data scientists and machine learning engineers trust that their models are correct and not biased or discriminatory against minorities in the society?
“ GDPR defines and strengthens data protection for consumers and harmonizes data security rules within the EU.”
GDPR is complicated and might intersect with existing regulations leaving its actual impact on organizations uncertain. But as professionals who are working with and processing data to be used in decision making, GDPR raises awareness of the ethical and moral aspects of data driven decision making.
“If the data scientist’s goal is to create automated processes that affect people’s lives, then he or she should regularly consider ethics in a way that academics in computer science and statistics, generally speaking, do not.”
Some people have expressed the need for a Data Scientist Pledge to ensure that every data scientist will perform their tasks in a moral and ethical manner. A KDnuggets poll on this topic revealed that non-data scientists seem positive in regards to establishing a Data Scientist Pledge, while data scientists themselves are more skeptical (Should there be a Data Scientist Pledge?).
The implications on individuals and groups of people can be devastating if bias contaminates data driven decision. Pro Publica wrote an article in 2016 highlighting the negative effects of machine bias in software used to predict future crimes (Machine Bias). Cathy O’Neil writes in her book “Weapons of math destruction” about how big data increases inequality and threatens democracy (Weapons of Math Destruction) and illustrates the amount of power given to algorithms and machines.
“People have too much trust in numbers to be intrinsically objective”
Algorithms are not inherently fair and objective, they extrapolate on evidence that is fed to them.
Local Interpretable Model-Agnostic Explanations (LIME)
Local Interpretable Model-Agnostic Explanations (LIME) is a novel explanation technique that explains the predictions of any classifier in an interpretable and faithful manner, by learning an interpretable model locally around the prediction (Why should I trust you? and Lime: Explaining the predictions of any machine learning classifier). With the enforcement of GDPR fast approaching, could techniques such as LIME help to better understand machine learning models and prevent biased decision-making?
It remains to be seen how people will interpret and evaluate the transparency offered by projects like LIME in their decision-making process and how big of an impact GDPR will have on data science and machine learning.
Are you ready for May 25, 2018?
