Google’s Neural Tangents library gives ‘unprecedented’ insights into AI models’ behavior
Google made available Neural Tangents is an open source software library written in JAX, a system for high-performance machine learning research. Google says it could allow “unprecedented” insight into the models’ behavior and “help open the black box” of machine learning.
Artificial Intelligence research is that increasing the width of models results in more regular behavior and makes them easier to understand. Neural Tangents comes in where data scientists construct and train ensembles of infinite-width networks at once using only five lines of code.
ML models are allowed to become infinitely wide tend to converge to another, simpler class of models called Gaussian processes. As per ML models Infinite-width limit of a finite model requires mathematical expertise and has to be worked out separately for each architecture.
On the basis of query strings they used to predict context, and in tests, the best-performing managed to anticipate product annotations with 97% accuracy for activity categories and 92% for audience categories.