will-sotai

That's a great question!

I like to think of these models as the best of both worlds between handwritten rules and black-box models like DNNs. You get the control and user defined logic of handwritten rules through shape constraints, and the predictive power and flexibility of a black-box model through standard ML training techniques.

I think your description best fits the Calibrated Linear model, which is essentially a regression model with user defined logic/rules/constraints. But Calibrated Lattice and Calibrated Lattice Ensemble models are what we call "universal approximators" (just like DNNs), so you can approximate far more complex functions than simple regression.

I am a former research engineer at Google AI who left a little over a year ago to found SOTAI to make machine learning transparent, interpretable, and accessible for everyone, particularly data scientists. I've found that data scientists are often stuck choosing between handwritten rules / heuristics, which make sense but lack predictive power, and black-box models such as DNNs, which are flexible but often don't make sense.

At Google AI, I researched interpretable machine learning techniques with the goal of providing an alternative option that offers the predictive power of black-box models without the black box. At SOTAI, we've built our platform, Artifice, to help professionals across industries harness the power of AI without sacrificing transparency.

Key features:

- No-code platform: easily prepare data, configure and train models, and analyze results.

- Shape constraints: Embed domain knowledge directly into your models

- Analysis tooling: gain valuable insights with purpose-built tooling.

If you're interested, sign up for our free Personal plan to explore the platform! Our recent blog post showcases real-world examples across various industries, including e-commerce, retail, real estate, finance, healthcare, and more. Use it as a reference for how to best use Artifice!

We're excited to learn from the community so we can continue to improve, so please share your thoughts!

TL;DR:

I left Google AI to found SOTAI and build Artifice, an interpretable machine learning platform to make machine learning make sense, particularly for data scientists. Please let us know what you think!

P.S. please excuse my forgetting to add a description...