Search papers, labs, and topics across Lattice.
This paper provides approximation rates and excess risk bounds for learning sparse compositional functions, represented as DAGs, using Frobenius norm-constrained deep neural networks in overparameterized regimes. The key insight is that by controlling the Frobenius norm, deep networks can effectively exploit the compositional structure of target functions, avoiding the curse of dimensionality. The derived rates apply to a broad class of functions, including multi-index models and binary tree structures, demonstrating the power of hierarchical representations.
Norm-constrained deep networks can dodge the curse of dimensionality when learning compositional functions, unlocking efficient learning even in overparameterized regimes.
The ability of deep neural networks to learn hierarchical features is widely regarded as a key mechanism underlying their success in high-dimensional learning. Existing theory partially supports this view by establishing approximation rates based on parameter counts and sample complexity guarantees for compositional models without incurring the curse of dimensionality (CoD). To study overparameterized regimes, where the number of parameters exceeds the sample size, we develop a framework that measures complexity via the parameter norm. Within this approach, we establish approximation rates and excess risk bounds for learning sparse compositional functions whose compositional structure is represented by directed acyclic graphs (DAGs), using Frobenius norm-constrained deep neural networks. Our results have broad applicability since every function that is efficiently Turing computable admits sparse compositional representations. In particular, we cover a range of representative models, including multi-index models, binary tree structures, and general compositional architectures. The rates we derive show that deep networks can exploit the compositional structure of the target functions, effectively avoiding the CoD through hierarchical representations.
