{"id":21108,"date":"2019-02-13T11:10:18","date_gmt":"2019-02-13T10:10:18","guid":{"rendered":"https:\/\/www.inovex.de\/blog\/?p=15078"},"modified":"2025-03-19T07:30:05","modified_gmt":"2025-03-19T06:30:05","slug":"machine-learning-interpretability","status":"publish","type":"post","link":"https:\/\/www.inovex.de\/de\/blog\/machine-learning-interpretability\/","title":{"rendered":"Machine Learning Interpretability: Do You Know What Your Model Is Doing?"},"content":{"rendered":"

Machine learning has a great potential to improve data products and business processes. It is used to propose products and news articles<\/a> that we might be interested in as well as to steer autonomous vehicles and to challenge human experts in non-trivial games<\/a>. Although machine learning models perform extraordinary well in solving those tasks, we need to be aware of the latent risks that arise through inadvertently encoding bias<\/a>, responsible for discriminating individuals<\/a> and strengthening preconceptions<\/a>, or mistakenly taking random correlation for causation<\/a>. In her book „Weapons of Math Destruction“<\/a>, Cathy O’Neil even went so far as to say that improvident use of algorithms can perpetuate inequality and threaten democracy. Filter bubbles<\/a>, racist chat bots<\/a>, and foolable face detection<\/a> are prominent examples of malicious outcomes of learning algorithms. With great power comes great responsibility\u2014wise words that every practitioner should keep in mind.<\/p>\n

With the adoption of GDPR, there are now EU-wide regulations concerning automated individual decision-making and profiling (Art. 22, also termed „right to explanation“), engaging companies to give individuals information about processing, to introduce ways for them to request intervention and to even carry out regular checks to make sure that the systems are working as intended<\/a>. Recent research in computational ethics propose to raise awareness to optimization criteria like fairness<\/a>, safety<\/a>\u00a0(2<\/a>) and transparency<\/a> when developing machine learning models. However, unlike usual performance metrics, these constraints are much harder if not impossible to quantify. Nevertheless, designing decision systems that are understandable not only for their creators, but also for their customers and users, is key to achieving trust and acceptance in the mainstream.<\/p>\n

\n

<\/p>\n<\/div>