Position: A Theory of Deep Learning Must Include Compositional Sparsity
By: David A. Danhofer , Davide D'Ascenzo , Rafael Dubach and more
Potential Business Impact:
Lets computers learn complex things by breaking them down.
Overparametrized Deep Neural Networks (DNNs) have demonstrated remarkable success in a wide variety of domains too high-dimensional for classical shallow networks subject to the curse of dimensionality. However, open questions about fundamental principles, that govern the learning dynamics of DNNs, remain. In this position paper we argue that it is the ability of DNNs to exploit the compositionally sparse structure of the target function driving their success. As such, DNNs can leverage the property that most practically relevant functions can be composed from a small set of constituent functions, each of which relies only on a low-dimensional subset of all inputs. We show that this property is shared by all efficiently Turing-computable functions and is therefore highly likely present in all current learning problems. While some promising theoretical insights on questions concerned with approximation and generalization exist in the setting of compositionally sparse functions, several important questions on the learnability and optimization of DNNs remain. Completing the picture of the role of compositional sparsity in deep learning is essential to a comprehensive theory of artificial, and even general, intelligence.
Similar Papers
A Theoretical Analysis of Compositional Generalization in Neural Networks: A Necessary and Sufficient Condition
Machine Learning (CS)
Teaches computers to understand new word combinations.
The Physics of Data and Tasks: Theories of Locality and Compositionality in Deep Learning
Machine Learning (CS)
Finds hidden patterns in data for smarter learning.
Sparse Computations in Deep Learning Inference
Computational Engineering, Finance, and Science
Makes AI faster and use less power.