In the first part of this blog, Basic Concepts and Techniques of AI Model Transparency, we reviewed a few common techniques for AI model transparency such as linear coefficients, local linear approximation, and permutation importance. In particular, the permutation importance is applicable to any black-box models, any accuracy/error functions, and more robust against high-dimensional data (because it handles each feature one by one rather than all features at the same time). One of the drawbacks of the permutation importance is its high computation cost. We have to repeat the evaluation process by (the number of features) * (the number of random shuffling to repeat) * (the number of models). To reduce the computation time, a common approach is to apply downsampling that works well when the positive and negative classes are balanced. However, such naive downsampling makes permutation importance extremely unreliable. Permutation Importance Under Class Imbalance Let us first see…
Transparency of AI/ML models is a topic as old as AI/ML itself. However, transparency has increasingly become more important due to proliferation of enterprise AI applications, critical breakthroughs in black-box ML modeling (e.g. Deep Learning), and greater concerns with increased personal data being used in AI models. The word “transparency” is often used in different contexts, but generally refers to issues like: Interpretability and Explainability Reproducibility and Traceability Ethic, Trust and Fairness This blog focuses on the most “basic” level of transparency, how to explain the impact of input variables (a.k.a. features) in the final prediction. There are many techniques to evaluate the impact of input features. Below are some common techniques and their advantages and disadvantages. Linear coefficients for Linear Models The simplest (but important) technique is linear coefficients (weights) of features. Fig.1 illustrates the idea of linear coefficients based on a simple two-dimensional example (x1 and x2 are…
According to a recent Adweek survey, two-thirds of business executives say COVID-19 hasn’t slowed AI projects. Some 40% said that the pandemic even accelerated their efforts. While the economic activity and business sentiment has deteriorated over the past couple of months, the scope of AI has expanded, the biggest impetus being decreasing costs, improving performance, and increasing efficiencies. So which industry verticals are embracing AI and what are the top applications? In the previous post of this two-part blog series, we discussed how AI is transforming industries, enhancing performance across a wide range of applications in Banking, Fintech, Healthcare, and Industry 4.0. In this final part, we look at the remaining industry verticals along with top enterprise applications:Insurance: Mckinsey’s latest research report on the Insurance industry noted that in the wake of the global pandemic, the insurers should invest in digital and analytics capabilities to make them more customer-centric, simple,…
We often hear about AI in consumer applications such as Alexa voice service (Natural Language Processing), Netflix recommendation engine (Machine Learning), and Facebook Facial Recognition (Deep Learning). However, we don’t hear enough about enterprise AI applications. Mckinsey Global Institute had predicted in 2018 that AI will transform the enterprise world. Today AI is generating tremendous cost savings and improving business operations across several industries. AI’s significance and impact will get even more dramatic in the future. In this two-part blog series, we look at the top industries where enterprise AI applications are being deployed and how AI is adding value. Let’s start with the top four industries and enterprise applications where AI has moved from PoC to production at scale: Banking: Banks are constantly facing pressure from competitors, growing governance, and regulatory requirements. The banking industry must manage financial and operational risk, prevent fraud, reduce customer defaults while keeping costs…
