Intrinsic Universal Measurements of Non-linear Embeddings
This work addresses a foundational issue in machine learning for researchers seeking to quantify mapping complexity, though it appears incremental as it builds on existing geometric concepts without demonstrating broad empirical impact.
The paper tackles the problem of defining simplicity for non-linear mappings in machine learning by proposing a formal measure of information based on local geometric discrepancies, independent of observations or parameterizations.
A basic problem in machine learning is to find a mapping $f$ from a low dimensional latent space $\mathcal{Y}$ to a high dimensional observation space $\mathcal{X}$. Modern tools such as deep neural networks are capable to represent general non-linear mappings. A learner can easily find a mapping which perfectly fits all the observations. However, such a mapping is often not considered as good, because it is not simple enough and can overfit. How to define simplicity? We try to make a formal definition on the amount of information imposed by a non-linear mapping $f$. Intuitively, we measure the local discrepancy between the pullback geometry and the intrinsic geometry of the latent space. Our definition is based on information geometry and is independent of the empirical observations, nor specific parameterizations. We prove its basic properties and discuss relationships with related machine learning methods.